The SIMIND

Monte Carlo
Program
Version 6.2
08/09/2020

Michael Ljungberg, PhD

Professor
Medical Radiation Physics
Department of Clinical Sciences, Lund
Lund University
SE-221 85 Lund, Sweden
[email protected]
 The SIMIND Monte Carlo Program Page 2 of 64

The Monte Carlo simulation code, simind, describes a standard clinical SPECT camera, and it
can easily be modified for nearly any type of calculation or measurement encountered in SPECT
imaging. simind has been developed by Professor Michael Ljungberg (Medical Radiation Physics,
Department of Clinical Sciences, Lund, Lund University, Sweden).

The entire code is written in FORTRAN-90, including versions that are fully operational on Linux
systems (x86), Mac systems running MacOS, and Windows systems (x86). The majority of the
main code structure is similar for all operating systems; however, in the case of a unique operating
system, additional information on the code pertaining to the specific system is provided.

The advantage of a Monte Carlo program is that it facilitates a detailed study of the radiation
transport which allows for inverstigating the reasons for the degradation of images captured by
nuclear medicine scintillation camera systems. For example, the images above show a series of
simulations. The first image corresponds to an imaging situation without patient motion and with
perfect camera resolution. The second image includes patient movements (respiration and
heartbeat). The third image is an image with normal SPECT system resolution for patient
movements. The next two images show degradation due to photon attenuation and photon
attenuation plus scatter contribution, respectively. In the last image, a realistic noise level has
been added.

The main references for the program are (1-4).

Version 6.2
 The SIMIND Monte Carlo Program Page 3 of 64

Introduction ........................................................................................................................ 5
Main page in Change .......................................................................................................... 7
Index Page 1 ........................................................................................................................ 9
Index Page 2 ...................................................................................................................... 12
Index Page 3 ...................................................................................................................... 15
Index Page 4 ...................................................................................................................... 17
Index Page 5 ...................................................................................................................... 20
Index Page 6 ...................................................................................................................... 21
Index Page 6 ...................................................................................................................... 25
Simulation Flags .............................................................................................................. 26
Scoring Routines .............................................................................................................. 27
The ”scattwin” scoring routine ......................................................................................... 28
The ”listmode” scoring routine ......................................................................................... 29
The ”forced collimation” scoring routine .......................................................................... 30
The ”penetrate” scoring routine ........................................................................................ 30
Source Routines ................................................................................................................ 31
The ”image-based sources” source routine ....................................................................... 31
The ”myocardiac” source routine ...................................................................................... 33
The ”multiple spheres” source routine .............................................................................. 33
Running simind ................................................................................................................ 34
Basic runtime-switches for simind ................................................................................... 36
Hidden flags ...................................................................................................................... 38
Some information about scripts ....................................................................................... 39
The “bim” program ........................................................................................................... 41
The “bis” program ............................................................................................................. 42
The “smc2castor” program................................................................................................ 42
Description of some files of importance ............................................................................ 43
The simind.ini file ......................................................................................................... 43
The error-file .................................................................................................................. 43
The Interfile-key file ..................................................................................................... 43
The collimator database ................................................................................................ 43
The result-file ................................................................................................................ 43
File-extensions and file-formats ................................................................................... 45
Information of available voxel-based phantoms .............................................................. 46
The Zubal phantom ....................................................................................................... 46
The NCAT/XCAT phantom ........................................................................................... 46
Tutorials ............................................................................................................................. 47
1. Point source simulation in a cylinder .......................................................................... 47
2. Jaszczak SPECT simulation if six spheres in a cylinder ............................................ 48
3. Scattered radiation ....................................................................................................... 49
4. Simulation of spatial resolution................................................................................... 49
5. Bone-scan simulation using an MDP-like radiopharmaceutical (SPECT) ................ 50
6. Bone-scan simulation using an MDP-like radiopharmaceutical (Whole body) ......... 51
7. Simulation of photon penetration in an LEGP collimator.......................................... 51
8. Some examples of castor reconstructions ..................................................................... 52
FAQ ...................................................................................................................................... 53
Installation ........................................................................................................................ 53
Phantoms .......................................................................................................................... 53
Output ............................................................................................................................... 54
Runtime ............................................................................................................................. 54

Version 6.2
 The SIMIND Monte Carlo Program Page 4 of 64

Simulations ....................................................................................................................... 54
Installation procedure .................................................................................................... 55
Linux/MacOS ................................................................................................................... 55
Windows ............................................................................................................................ 55
The castor programs ......................................................................................................... 55
Version History ................................................................................................................. 57
References .......................................................................................................................... 63

Version 6.2
 The SIMIND Monte Carlo Program Page 5 of 64

INTRODUCTION
The simind system has two main programs, namely change and simind. The change program
provides a way of defining the system to be simulated and writing data to external data files.

The actual Monte Carlo simulation is performed by the simind program, which reads input files
created by change and outputs results to the screen or to different data files. Thus, several input
files can be prepared and loaded into a command file for submission to a batch queue, which is a
convenient way of working because Monte Carlo simulations by default are time-consuming
processes.

This manual provides a detailed explanation of the structure and purpose of simind and all
related subroutines that are important for the user to understand. In addition, it provides
technical instructions for programming the code for specific applications.

The primitive outlines of the topics addressed herein include an in-depth description of simind's
subroutine files, the manipulation of gamma camera parameters, the implementation of
phantoms, the general operation of simind, and instructions for acquiring planar and SPECT
images.

The change program enables the user to easily define the desired imaging system. The change
includes a series of menus that prompt the user to input parameters specific to the description of
the system. These parameters are then written to a data file used in simind.
A simind file name assumes a base name and a three-character extension separated by a period.
The output file may have the same base file name as the input file but will usually differ in terms
of the file extension. The output file can easily be redirected to other file names.
A default file, namely simind.smc, is always created when invoking the change program. This
file may be used when performing tests that do not need to be stored in separate files. Each time
that changes are made in change, the modifications are written to simind.smc. Even when
exporting data to smc files with unique names, the content in simind.smc is updated. Thus, this
file should be regarded as a temporary file.

A file name can be entered after the program name change at the operating system command
prompt. If this file already exists, it will be used when reading data into change. All
modifications will be saved to this file and to simind.smc at exit. If only the command change
is entered, data can be exported/imported by Options 5 and 6 in the main menu. The coordinate
system is defined in the figure below.

Version 6.2
 The SIMIND Monte Carlo Program Page 6 of 64

When simulating voxel-based phantoms, the first density/activity image is located towards +X
and the last one is located towards –X. When simulating SPECT, the camera rotates in the ZY
plane either clockwise or counter-clockwise (index 30).

The flow-chart below describing the link between the change program and simind and the files
that are either used or produced in a simulation project.

Version 6.2
 The SIMIND Monte Carlo Program Page 7 of 64

Main page in Change

Main-1: Not in use at the moment.

Main-2: This option allows the user to read the menus and change individual values. An index
value is prompted, and after selecting the appropriate one, the value of that particular index is
given. The menu is immediately updated. A carriage return, given as a value, returns a null value.

Main-3: Menu showing the possible simulation flags that can be used to control the program and
that allow the user to invoke or evoke certain types of simulation features. These flags are Boolean
and are set/reset by typing the corresponding flag number.

Main-4: Allows the user to export current change data to a named SMC file. To transfer changes
to multiple files, please refer to Option 8 within this section. A carriage return given at the file
name prompter returns to the main menu without action.

Main-5: Allows the user to import data from a *.smc file previously stored. A carriage return
given at the file name prompter returns to the main menu without action.

Main-6: Allows the user to delete all information (parameter values) from change (current smc
file) and thus define the simulated system from scratch. All detector parameters are set to zero
and the simulation flags are set to FALSE.

Main-7: Allows the user to insert a line of comments about the individual simulation that will
then appear in the result file (*.res) written by simind.

Main-8: Allows the user to transfer current changes from an SMC file to other previously created
SMC files. By simply invoking the wildcard command, the user can control data transfer from one
SMC file to any other SMC file as required. The program prompts for [Y,N,A,Q]. "Y" means that
changes will be transferred to the current file, "N" means no transfer, "A" means transfer without

Version 6.2
 The SIMIND Monte Carlo Program Page 8 of 64

further confirmation (ALL), and "Q" means simply quit the option and return to the main menu.
The counter that keeps track of the changes is reset

• by entering the program change,

• by clearing all data by Value 8, or
• after a transfer by Value 10.

Main-9: Tables of energy-dependent cross-sectional values for the photoelectric, Compton,

coherent, and pair production interactions unique to the phantom. This is used for both an
analytical phantom and a voxel-based photon. The typical phantom cross section is water, given
by the file h2o.cr3.

Main-10: Tables of energy-dependent cross-sectional values for the photoelectric, Compton,

coherent, and pair production interactions unique to the phantom. The typical phantom cross
section is water, given by the file h2o.cr3. This cross-section table is used only when simulating
voxel-based phantoms. The cross section is used for those voxels that have a density value higher
than a defined threshold. This threshold is defined in the simind.ini file located in the smc_dir
folder. A typical cross-section table is bone.cr3, but it could also be the same as the selection in
Main-9.

Main-11: Tables of energy-dependent cross-sectional values for the photoelectric, Compton,

coherent, and pair production interactions unique to the cover material incorporated into simind.
In the cover, interactions occur but no scoring of deposit energy is made. A typical cross-section
material for this compartment is al.cr3.

Main-12: Tables of energy-dependent cross-sectional values for the photoelectric, Compton,

coherent, and pair production interactions unique to the detector crystal material incorporated
into simind. The crystal of the detector is typically sodium iodide, and it is denoted by the file
name, nai.cr3.

Main-13: Allows the user to input a name for a file that stores density maps for simulations of a
non-homogeneous phantom. The file extension default is *.smi.

Note: If you are using images created from MCAT or NCAT/XCAT software, you should only give
the base name of the file. simind adds the characters 'atn_av'. It is assumed that the
MCAT/NCAT/XCAT files have the extension 'bin'.

Main-14: Allows the user to input a name for a file that stores source maps for simulation of a
nonhomogeneous source. Similarly, the file extension default is *.smi. The file name is given to
simind by the data file.

Note: If you are using images created from MCAT or NCAT/XCAT software, you should only give
the base name of the file. simind adds the characters 'act_av'. It is assumed that the
MCAT/NCAT/XCAT files have the extension 'bin'.

Main-15: Allows the user to input a file name for the backscattering media. This simulates the
backscattering of photons from light guides and photo-multiplier tubes (PMTs). However, the
simulation is based on a solid material whose size is equal to that of the crystal but with a user-
selected thickness. Index-11 in MENU 2 gives the thickness. The number of scatter orders is set
in the simind.ini file.

Version 6.2
 The SIMIND Monte Carlo Program Page 9 of 64

Index Page 1

Index-1: The energy in keV of the photon is given here. simind simulates only the transport of
photons and disregards the transport of secondary electrons. The energy imparted by the electrons
is assumed to be locally absorbed at the interaction site. Even if the cross-section tables include
photon energies of up to several MeV, the user should be aware of the possibility of inaccuracy of
the results when simulating small crystal sizes and high photon energies.

If the value is negative, a call to the isotope routine is made with the purpose of defining the
initial photon energy from this routine instead of using the value defined by Index-1. However,
the absolute value of Index-1 is used to define the location and width of the energy window and
other parameters [keV].

Index-2: The x-dimensions of the standard source shapes are defined by this parameter. Please
note that the numerical values entered for the corresponding dimensions are half of the total
dimensions. Standard source geometries can be point, ellipsoidal, rectangular, or cylindrical
(elliptical) in both the horizontal and the vertical directions. The choice for the actual shape of the
sources is given in Index-15 described below [cm].

If voxel-based source maps are simulated, then this value defines the half-length of all slices
defined by the phantom. For example, assuming 128 slices and a value of 40 for Index-2, the slice
thickness will be 2*40/128 = 0.63 cm.

Index-3: The y-dimensions of the standard source shapes are defined. Please note that the
numerical values entered for the corresponding dimensions are half of the total dimensions.
Standard source geometries can be point, ellipsoidal, rectangular, or cylindrical (elliptical) in both
the horizontal and the vertical directions. The choice for the actual shape of the sources is given
in Index-15 described below [cm].

For voxel phantoms, this value is not used. Index-31 is used to define the voxel sides of the
phantom (not the slice thickness).

Version 6.2
 The SIMIND Monte Carlo Program Page 10 of 64

Index-4: The z-dimensions of the standard source shapes are defined. Please note that the
numerical values entered for the corresponding dimensions are half of the total dimensions.
Standard source geometries can be point, ellipsoidal, rectangular, or cylindrical (elliptical) in both
the horizontal and the vertical directions. The choice for the actual shape of the sources is given
in Index-15 described below [cm].

For voxel phantoms, this value is not used. Index-31 is used to define the voxel sides of the
phantom (not the slice thickness).

Index-5: The x-dimension of the standard phantoms is defined. The dimensions of the phantoms
are defined similarly to the source distribution described above in that all dimension values are
just half of their total values [cm].

If voxel-based phantoms are simulated, the values define the half-length of all slices defined by
the phantom. For example, assuming 128 slices and a value of 40 for Index-2, the slice thickness
will be 2*40/128 = 0.63 cm.

Index-6: The y-dimension of the standard phantoms is defined. The dimensions of the phantoms
are defined similarly to the source distribution described above in that all dimension values are
just half of their total values. For voxel phantoms, this value is not used. Index-31 is used to define
the voxel sides of the phantom (not the slice thickness) [cm].

Index-7: The z-dimension of the standard phantoms is defined. The dimensions of the phantoms
are defined similarly to the source distribution described above in that all dimension values are
just half of their total values. For voxel phantoms, this value is not used. Index-31 is used to define
the voxel sides of the phantom (not the slice thickness) [cm].

Index-8: The length of the cylindrical crystal in the x-direction. If Index-10 is set to zero, then
the detector is assumed to be of a cylindrical shape with a radius equal to the value defined Index-
8 [cm].

Index-9: The thickness of the scintillation crystal is given here [cm].

Index-10: The width of the detector. If this index is equal to zero, then the detector is assumed
to be of a cylindrical shape with a radius equal to the value defined by Index-8 [cm].

Index-11: This value determines the thickness of the volume behind the scintillation crystal that
simulates the light guides and PM tubes [cm]. The number of scatter-orders in the backscattering
media is defined in the simind.ini file.

Index-12: The distance from the origin of the coordinate system (see Figure 1) to the lowest part
of the detector. Note that, depending on the defined simulation flags, this can be the collimator
lower surface, the protection layer, or the scintillation crystal itself [cm].

If this is set to a negative number when using analytical phantoms, the distance will be equal to
Index-5 plus the absolute value of this negative number. This is convenient when simulating
various phantoms using the same phantom surface to camera distance.

Index-13: A protective cover that has the same shape as the scintillation crystal can be simulated.
However, no imparted energy is scored within the cover. The material can also be different and is
defined by the file name of the cross-section table in the main menu. Index-13 then defines the
thickness of this protective cover [cm].

Note: In this version of simind, the protective cover works only for photons that are impinging
on the lower side of the detector.

Version 6.2
 The SIMIND Monte Carlo Program Page 11 of 64

Index-14: Integer flag indicating the type of phantom that is to be simulated. Available standard
phantoms are

2 = Rectangular phantom,
3 = Vertical cylindrical phantom, and
4 = Horizontal cylindrical phantom.

Note: The cylindrical phantom can be elliptical if either Index-5 and -6 or Index-6 and -7
(depending on the direction of the phantom) are chosen not to be equal.

If the flag value is negative, the phantom is simulated using binary density maps. The file name
storing the particular maps are defined in Option 14 of the main menu and should have the
following extensions:

-1 = An density distribution map (16-bit integers,1000*density, extension: *.dmi)

-2 = The Zubal Voxel-Man density distribution (8-bit, extension: *.dat)
-3 = The Zubal brain density distribution (8-bit, extension: *.dat)
-4 = The Zubal whole-body Voxel-Man density distribution (8-bit, extension: *.dat)
-5 = Code-based Zubal phantom (8-bit, extension: *.dat)
-6 = Code-based NCAT/XCAT phantom (32-bit float, extension: *.bin)
-7 = Generic NCAT/XCAT phantom (32-bit float, extension: *.bin)
-8 = Generic MCAT phantom (32-bit float, extension: *.bin)

The non-homogeneous phantom is further defined by Index-31 to -38, which are described below.

Note: Because the coordinate system for MCAT/NCAT axially is rotated by 180 degrees compared
to simind, these images are flipped vertically when reading into simind.

Note: NCAT and MCAT add the label "atn_av" to the base name. This is also done by simind, so
you should only give the base name for the phantom in Index-14 in the main menu.

Note: When simulating time frames generated by NCAT or MCAT, refer to the information
concerning the switches /SF and /DF under the section "Running simind".

Index-15: Integer flag indicating the type of source distribution that is to be simulated. The
available standard source shapes are as follows:

1 = Ellipsoidal source
2 = Rectangular source
3 = Vertical cylindrical source
4 = Horizontal cylindrical source
5 = Point source (x,y,z = 0,0,0)
6 = Simple cardiac phantom (refer to the information about this below)
7 = Multiple spheres phantom (old jaszak program)

If a horizontal cylinder is selected, then Index-2 is the axial source length and Index-3 and -4
define the other axes, respectively. If a vertical cylinder is chosen, then Index-4 defines the axial
source length and Index-2 and -3 define the remaining axes.

Note: The cylindrical source distribution can be elliptical if either Index-2 and -3 or Index-3 and -
4 (depending on the direction of the source) are chosen not to be identical.

If the flag value is negative, a call is made to a user-written source routine. This source routine
should then provide the coordinates of a sampled decay position through the parameter list. This
procedure is described further in the manual. If the default source map routine is used, it supports
the following phantoms.

Version 6.2
 The SIMIND Monte Carlo Program Page 12 of 64

-1 = An source distribution map (16-bit integer, extension: *.smi)

-2 = The Zubal Voxel-Man source distribution (8-bit, extension: *.dat)
-3 = The Zubal brain source distribution (8-bit, extension: *.dat)
-4 = The Zubal whole-body Voxel-Man source distribution (8-bit, extension: *.dat)
-5 = Code-based Zubal phantom (8-bit, extension: *.bin)
-6 = Code-based NCAT phantom (32-bit float, extension: *.bin)
-7 = Generic NCAT phantom (32-bit float, extension: *.bin)
-8 = Generic MCAT phantom (32-bit float, extension: *.bin)

Note: The difference between -7 and -8 is that the NCAT/XCAT images are rotated by 180 degrees
relative to the MCAT images.

Note: NCAT and MCAT add the label "act_av" to the base name. This is also done by simind, so
you should only give the base name for the phantom in Index-14 in the main menu.

Note: When simulating time-frames generated by NCAT or MCAT, refer to the information
concerning the switches /SF and /DF under the section "Running simind"

Note: When using voxel-based images as a source, the value of Index-26 is not used. Instead, the
sum of all pixel values in the source maps is used as photons/projections. This value can be
increased by an integer factor n using the switch /NN:n

Note: For Index-2 to Index-8, the voxel size for the source maps follows the voxel size for the
density map and the slice thickness, given in Index-31 and calculated by the number of density
images and physical axial phantom length, respectively. However, you can set switches /PX: and
/TH to override these defaults. Refer to the documentation for the source map provided later in
the manual.

Index Page 2

Version 6.2
 The SIMIND Monte Carlo Program Page 13 of 64

Index-16: A translation of the x-coordinate of the sampled decay location that has been generated
by the standard source geometries or the source routines supplied can be made here. Normally, a
decay point is sampled symmetrically around the origin of the coordinate system. Nevertheless,
it may be required to move the source, e.g. a point source location, either closer to the surface or
to a larger depth.

Index-17: A translation of the y-coordinate of the sampled decay location that has been generated
by the standard source geometries or the source routines supplied can be made here. Normally, a
decay point is sampled symmetrically around the origin of the coordinate system. Nevertheless,
it may be required to move the source, e.g. a point source location, either closer to the surface or
to a larger depth.

Index-18: A translation of the z-coordinate of the sampled decay location that has been generated
by the standard source geometries or the source routines supplied can be made here. Normally, a
decay point is sampled symmetrically around the origin of the coordinate system. Nevertheless,
it may be required to move the source, e.g. a point source location, either closer to the surface or
to a larger depth [cm].

Note: A positive value of z moves the source towards the camera.

Index-19: This integer flag controls the solid angle in which the photon is to be emitted. The
calculation of the solid angle depends on the actual value of Index-19.

1) If the value is 2, then the solid angle is calculated using the following rules in descending
priority:

a) If a collimator is used, the solid angle will be calculated from the hole sizes and the effective
collimator thickness.
b) If a phantom is simulated, then the solid angle is calculated from the phantom dimension,
given by Index-5, -6, and -7, and the detector dimensions such that a photon will always have
a chance to strike any part of the detector from any point within the phantom.
c) If no phantom is simulated, the solid angle is calculated from the source dimension, given by
Index-2, -3, and -4, and the detector dimensions such that a photon will always have a chance
to strike any part of the detector from any point within the source distribution.

2) If the value is 3, the solid angle is calculated using the phantom dimension and camera
dimension regardless of whether a collimator has been selected. This is convenient in simulations
where penetration is important (Index-53 = 1) and when it is necessary to simulate photons
impinging on the whole camera surface.

3) If the value is less than or equal to zero, then the polar angle that defines the solid angle will
be calculated from the absolute value given in this index [degrees].

Note: The calculations 1 and 2 above use the values defined in change. Therefore, the proper
value for the solid angle may not be calculated if an external source routine is used. Hence, the
user should give an absolute value of the solid angle to be sure that the whole detector is covered.
This restriction is only valid when simulating an ordinary detector without collimators.

Note: If a value of zero is given to simulate a narrow-beam situation, then some basic parameters
in the result file will not be calculated. The value cannot be equal to or greater than -90. In such
cases, the maximum angle is set to 89.99 degrees.

Index-20: This index refers to the upper energy window threshold. Instead of using the absolute
values of the upper threshold in keV, a relative window definition can be given by using a negative
sign. For example, a value of -20 given to either Index-20 or -21 will be regarded as a 20% (+-10%)
energy window centred on the primary photon energy, given by Index-1 [either keV or %].

Version 6.2
 The SIMIND Monte Carlo Program Page 14 of 64

Index-21: This index refers to the lower energy window threshold. Instead of using the absolute
values of the lower threshold in keV, a relative window definition can be given by using a negative
sign. For example, a value of -20 given to either Index-20 or -21 will be regarded as a 20% (+-10%)
energy window centred on the primary photon energy, given by Index-1 [either keV or %].

Index-22: The energy resolution of the detector is modelled from an on-line convolution of the
imparted energy from each photon history using an energy-dependent Gaussian function that
varies with 1/Sort(E). The reference point for the energy resolution is given as the percentage full-
width at half-maximum (FWHM) at 140 keV [%]. The simulation Flag-12 can turn off the energy
resolution.

Index-23: This index refers to the intrinsic spatial resolution of the crystal. The simulation of
this parameter is based on the same concept as that of the energy resolution described above. The
spatial resolution is provided as the FWHM at 140 keV and will blur the x- and y-coordinates by
a Gaussian function [cm].

Index-24: The number of emitted photons per decay is used to calculate basic detector
parameters, such as the sensitivity (cps/MBq) of the detector system. This value can also be
defined from the spectrum file. The priority is as follows:

1) If there is a line in the spectrum file starting with the word GAMMA = value, then the emitted
photons per MBq will be set to 'value'.
2) If a spectrum file without an explicit GAMMA is defined in the file, then the GAMMA is
calculated from the sum of abundances for all photon energies.
3) If no spectrum file is used, the GAMMA obtained from Index-24 in change is used.

Option 1 above will be useful if only a part of an energy spectrum is to be simulated while retaining
the proper value of GAMMA.

Index-25: The activity of the source distribution is used to obtain correct values of the sensitivity
and different calculated count rates, as well as correct values for the pixels in the projection data.
It does not decide the number of photons that are simulated [MBq].

Note: The corresponding acquisition time per projection is assumed to be 1 s. If the numerical
values in the projection data should reflect a different acquisition time, then use Index-25 such
that the value reflects the product of activity times the acquisition time.

Index-26: This index represents the number of photon histories that are simulated per projection
angle. If a SPECT study is to be simulated, the total number of compiled photon histories will be
this value times the number of projection angles. The value is given in thousands of histories.
Note that doubling this value does not increase the count level in the final projection nor the
energy pulse-height distribution because all data are normalized to the defined activity and
acquisition time of 1 s. However, it improves the statistical uncertainty.

Note: This value does not affect the number of histories when simulating voxel-based phantoms.
In these cases, the sum of all voxel values will determine the number of histories per projection.
This value can be increased by the multiplier, defined by the switch /NN.

Index-27: simind has a limited number of energy channels for calculating the energy pulse-
height distribution. The default value is 512 channels, but it can be set to a different value by
Index-80. Index-27 then represents the number of keV/channel. If the defined upper energy
window (given by Index-20) is greater than the maximum energy defined by the number of bins
times the keV/channel in the spectra, then simind will increment the keV/channels to a proper
value that covers the upper energy threshold. Nevertheless, this check will not decrement the
keV/channel value [keV/channel].

Version 6.2
 The SIMIND Monte Carlo Program Page 15 of 64

Index-28: The pixel size is used to define the area on the crystal that is represented by the
projection image. The matrix is always centred on the cylindrical axis of the crystal and no
translation can be made. The actual dimension of the projection matrix is given by Index-76 and
-77. However, the created image can be zoomed in or out arbitrarily by giving a smaller/larger
value of the pixel size. An error message will appear if an image with zero pixel size is simulated
[cm].

Index-29: This value defines the number of SPECT projections to be simulated. Each projection
is stored in a SPECT file or in the current directory.

Index-30: This option defines the SPECT rotation. The value is the rotation angle and the sign
defines the rotation direction: negative implies counter-clockwise and positive implies clockwise.
To be comparable with older versions of simind, the following translations remain [degrees].

0 = 360 degrees counter-clockwise

1 = 180 degrees counter-clockwise
2 = 360 degrees clockwise
3 = 180 degrees clockwise

Index Page 3

The non-homogeneous phantom simulation is based on using a set of integer matrices that
describes the density distribution in a section of the phantom. The file format is described in the
section "FILE EXTENSION and FILE FORMATS". Each cell in the density maps must store the
density times a factor of 1000. The half-length of the phantom is defined by Index-5. The choices
available in this menu for setting up the non-homogeneous phantom and the different SPECT
parameters include the following:

Index-31: This value represents the pixel size of the density maps being used to simulate the
non-standard phantoms. This value has nothing to do with the pixel size used to form the images
and projection matrices. Instead, the value is used to scale the distance from a relative cell
position in the 'matrix' space to a physical distance in the coordinate system [cm].

Version 6.2
 The SIMIND Monte Carlo Program Page 16 of 64

Index-32: Integer value that refers to the direction of the phantom that is defined by the density
maps.

0 = density map where (i,j) corresponds to direction y,z (Transversal slices)

1 = density map where (i,j) corresponds to direction x,y (Sagittal slices)
2 = density map where (i,j) corresponds to direction x,z (Coronal slices)

Index-33: This index defines the image number in the density map file that will be read as the
first image. If the value is less than zero, then the block number (number of blocks of 512 bytes)
in the binary file that defines the first density map is defined. Block zero indicates the beginning
of the file. This can be used to skip a file header.

Index-34: Value equal to the number of density maps that will be read into simind. The memory
for these maps is allocated at runtime.

Index-35: Value for the density limit that defines the border. If the density image is slightly
noisy, a threshold can be used to properly define the border of the phantom. The ray-tracing
technique assumes that the border of the phantom has been reached if a density value that is
lower than the given threshold is found [gcm-3].

Index-36: This index refers to the column location (I dir) of the pixel cell in the density map that
corresponds to the origin of the coordinate system. Normally, the cell with index 0,0 should be at
the centre. However, if a translation has been performed in the maps, then the shift of the centre
pixel cell from the origin can be accounted for [cm].

Index-37: This index refers to the row location (J dir) of the pixel cell in the density map that
corresponds to the origin of the coordinate system. Normally, the cell with index 0,0 should be at
the centre. However, if a translation has been performed in the maps, then the shift of the centre
pixel cell from the origin can be accounted for [cm].

Index-38: This value is the step size that is used to ray-trace the photon path through the density
maps. This value should be less than the pixel size of the density map (Index-31). However, the
actual computational time may depend on this value, so do not select a low value that is too small
[cm].

Index-39: This index refers to the location (K dir) of the slice in the density map that corresponds
to the origin of the coordinate system. This allows the user to shift the centre of the phantom in
the x-direction relative to the origin of the system [cm].

Index-40: The density threshold that defines the change from the cross-section file for soft
tissue (Index-9 in the main menu) to the cross-section file for bone (Index-10 in the main menu).
If the value is set to zero, then the default value, defined in the system file simind.ini (located
in the smc_dir directory), is used.

Index-41: This index refers to the starting angle of the detector in a SPECT simulation. A positive
value indicates a clockwise rotation, while a negative value indicates a counter-clockwise rotation.
A value of zero means that the detector is centred on the z-axis of the coordinate system [degrees].

Index-42: This value controls the non-circular orbit of the SPECT camera. This orbit can operate
in different ways. A positive value determines a parameter called ”orbital rotation fraction” and
is used to simulate a non-circular SPECT study based on a major and minor axis. This fraction
defines the maximum distance from the origin to the lowest part of the detector in the y-direction
relative to the distance given in Index-12. For example, a fraction of 1.5 yields a maximum radius
of rotation along the y-axis of 15 cm if Index-12 is set to 10 cm. This value can be less than unity
but not less than or equal to zero.

Version 6.2
 The SIMIND Monte Carlo Program Page 17 of 64

If the value is negative, the distance from the origin to the detector surface is calculated for each
angle using the density map. This mimics a SPECT system with photosensitive detection of the
outline. The absolute value of Index-42 will define the distance (air gap) between the surface of
the phantom and the detector. The distances are stored for each projection angle in a file whose
base name is the same as the input name with the extension *.cor. Note that it is important for
the density value that defines the border (Index-35) to be higher than 0.0; otherwise, the distances
calculated will correspond to the density matrix borders and not to the object inside the density
matrix. The *.cor file is a simple ASCII text file, and it can then be used in a reconstruction
program that can compensate for the distance-dependent spatial resolution.

Index-43: This index allows a shift of the camera head relative to the phantom. This can be useful
if a large phantom is simulated with a source distribution outside the field-of-view of the camera
[cm].

Index-44: This index specifies a shift of the camera in the y-direction relative to the phantom. If
it is zero, the camera centre and the phantom centre are aligned to each other [cm].

Index-45: A special code-based phantom format is supported when simulating a voxel-based

phantom. Instead of defining the attenuation and activity directly in the images, these values are
defined in a table. This facilitates a setup of multiple source distributions for a given phantom
because only one binary voxel phantom is needed. These phantoms should be either in the 8-bit
mode (Index-14, -15 = -5) or as float (32-bit values) arrays (Index-14, -15 = -6). Each voxel should
have a value from 0 to 255, where the number identifies particular structures/organs. The
definition of the organ, its code, and corresponding densities and relative activity concentrations
are given in a *.zub file (refer to vox_man.zub in the smc_dir directory as an example). In a *.zub
text file, there are different sections that are valid for different phantoms. You may also develop
your own phantom and add a code section to the text file as long as it follows the format.

The supported phantoms for this code-based feature are the Zubal torso phantom vox_man1.dat
(Index-14, -15 = -1), the Zubal brain phantom vox_brn.dat, and the whole-body version of the
Zubal torso phantom vox_man3.dat. Further, a segmented RSD Torso rsd.dat is available.
Because these phantoms have the same code for different regions, you must select the proper code
page in the *.zub file.

This is a number indicated in the text file. If you create your own code-based phantom in a byte
format, then you can add your own section in the *.zub file if you follow the format of the earlier
phantom and select a unique number. The main advantage of this feature is that you can easily
change the densities and activity per voxel by editing a simple text file instead of creating large
image files.

Index Page 4

Version 6.2
 The SIMIND Monte Carlo Program Page 18 of 64

Included in change is a large database that covers most of the commercial collimators available
for both SPECT and planar imaging. The collimator parameters are stored in a file, called
collim.col, in the special SMC directory. The contents of the file are listed in Appendix A.
A code name must be given at the prompt to include a specific collimator. For example, if a General
Electric Low-Energy General-Purpose collimator is to be simulated, the following command can
be entered:
Index_Number…> *ge-legp
The values for that particular collimator will then be updated. When simulating a Standard
collimator that is not included in the database, simply input the inner hole diameter (Index 46 -
Hole Size X), the septa thickness ( Index 48- Distance between holes in x-direction), the collimator
thickness (Index 52: Collimator Thicknes), and the type of collimator ( Index 54 Hole Shape: 2=
Cir, 3= Hex, 4=Rect). If this procedure is followed, then the command given at the prompt is:

Index_Number…> *standard

At this point, the new collimator parameters will be calculated.

Note: This calculation assumes a 'normal' collimator, i.e. four holes in each corner and one in the
middle if a hexagonal or elliptical/circular hole shape has been chosen.

Index-46: This parameter is the length of the collimator hole in the x-direction [cm]. For
hexagonal holes, this value defines the inner square-to-square hole diameter commonly referred
to in many collimator data-sheets as an inscribed circle or a flat-to-flat distance [cm].

Index-47: This parameter is the length of the collimator hole in the y-direction [cm].

Note: For a hex-hole collimator, this value is normally not given by the user. Instead, it is
calculated using the command *standard (refer to the above text).

Index-48: This parameter is the distance between two collimator holes in the x-direction. In the
data-sheet, this distance is often stated as the septal thickness [cm].

Version 6.2
 The SIMIND Monte Carlo Program Page 19 of 64

Index-49: This parameter is the distance between two collimator holes in the y-direction [cm].
This value is normally not given by the user. Instead, it is calculated using the command
*standard (refer to the above text).

Index-50: This parameter is the distance of the shift in the x-direction from the centre to adjacent
holes. This parameter holds only for circular or hexagonal collimator holes [cm]. For a hex-hole
collimator, this value is normally not given by the user. Instead, it is calculated using the
command *standard (refer to the above text).

Index-51: This parameter is the distance of the shift in the y-direction from the centre to adjacent
holes. This parameter holds only for circular or hexagonal collimator holes [cm]. This value is
normally not given by the user. Instead, it is calculated using the command *standard (refer to
the above text).

Index-52: This parameter is the physical thickness of the collimator [cm].

Index-53: The type of collimator routine used.

0 = the earlier default collimator with analytical weighting is used (developed by Eric C. Frey et
al.).
1 = the parallel hole collimator with hex or rectangular holes that allow for penetration and
scattering is used (used before in the SIMINDC and SCATTWINC programs).

Index-54: This parameter denotes the geometric shape of the collimator holes. The possible
options for this flag at present are 3 (hexagonal holes) and 4 (rectangular holes). This option is
only used in collimator 1 (Index-53).

Index-55: This parameter defines the type of collimator used when Index-53 is set to 0. The
possible options are

0 = Parallel-hole collimator (Index-51 = 0,1)

1 = Pinhole collimator (Index-51 = 1)
2 = Converging collimator (Index-51 = 0)
3 = Fan-beam collimator (Index-51 = 0)
4 = Diverging collimator (Index-51 = 0)
5 = Slant-hole collimator (Index-51 = 1)

Index-56: Collimator-dependent variable 1.

a) Parallel-hole collimators: This will define a possible air gap between the collimator and the
cover of the detector.
b) Pinhole collimator (Index-53 = 1): This value defines the air gap between the pinhole and the
cover of the detector.

Index-57: Collimator-dependent variable 2.

a) When simulating fan-beam, cone-beam, or diverging collimators with collimator Index-53 =

0, this value defines the focal distance.

Index-58: Collimator-dependent variable 3. It will be used in future releases depending on the

choice of the collimators.

Index-59: If this is 1, then the parallel-hole collimator that handles septa penetration (Index-53
= 1) will be arbitrarily moved (one hole diameter + one septa) to avoid visible hole patterns in the
image.

Version 6.2
 The SIMIND Monte Carlo Program Page 20 of 64

Index-60: Collimator-dependent variable 4. It will be used in future releases depending on the

choice of the collimators.

Index Page 5

Index-61: The energy for the transmission photons. If a value of zero is defined, then the energy
will be equal to the energy of the emission photons (Index-1) [keV].

Index-62: The polar angle in which the transmission photons are emitted. If the value is 0, then
the polar angle as defined by the acceptance angle of the collimator holes is used [degrees].

Index-63: The azimuthal angle in which the transmission photons are emitted. If the value is 0,
then the sampled range of the azimuthal angle will be randomly distributed between 0 and 360
[degrees].

Index-64: The source length of the transmission source along the x-axis [cm]

Index-65: The source width of the transmission source along the y-axis. If the value is 0, then the
source width is equal to the number of columns in the density map times the pixel size for the
density map [cm].

Index-66: The shift of the transmission source in the x-direction [cm].

Index-67: The shift of the transmission source in the y-direction [cm].

Index-68: The shift of the transmission source in the z-direction [cm].

Index-69: Unused.

Index-70: During the simulation, a corresponding blank projection is calculated. This projection
is not generally stored in a file but is necessary to calculate the attenuation projections. This

Version 6.2
 The SIMIND Monte Carlo Program Page 21 of 64

option defines the way in which the transmission projections are stored. The result is stored in
the SPECT *.a00 file

2 = Raw transmission projection

1 = The ratio of the transmission projection to the correlated blank projection
0 = The log of the ratio of the transmission projection to the correlated blank projection

Index Page 6

Index-76: The size of the image matrix in terms of the number of columns. The first column is 1.

Index-77: The size of the image matrix in terms of the number of rows. The first row is 1.

Index-78: Matrix size of the density map in the I direction (column).

Index-79: Matrix size of the source map in the I direction (column).

Note: The Zubal phantom (Index-14 and -15 = -2, -3, -4) has specific hard-coded values equal to
128×128 (Index-2), 256×256 (Index-3), and 192×96 (Index-4). This is because these are known
phantoms located in the smc_dir folder.

Index-80: Number of channels in the energy pulse-height distributions. The maximum energy
will then be this value times the keV/channel defined by Index-27.

Index-81: Matrix size of the density map in the J direction (row). If this value is zero, then it will
be set equal to the value of Index-78.

Index-82: Matrix size of the source map in the J direction (row). If this value is zero, then it will
be set equal to the value of Index-79.

Index-83: Cut-off energy used to terminate the photon histories below the lower energy threshold
when simulating scatter in the phantom. This value can be used in two ways.

Version 6.2
 The SIMIND Monte Carlo Program Page 22 of 64

1) If the value is positive, then the cut-off value is the actual value in keV.
2) If the value is negative, then the cut-off value is defined as the lower energy window threshold
minus the value of Index-83.

When using the scat twin scoring routine (Index-84 = 1) together with a negative number of the
cut-off energy, the lowest value of the energy window settings defined in the *.win file will be used
as the cut-off value.

Note: The program will be terminated if the cut-off value is higher than the lowest threshold of
the energy window.

Note: Because of the energy resolution, an event with a lower energy than the lower energy
window value may still enter the energy window because of the Gaussian blurring. Therefore, you
need to be careful about the section of the value so that you do not miss these events.

Note: The maximum scatter order is set to 10 automatically when using a cut-off value.

Index-84: Standard scoring routine compiled and linked to the program that can be used. Refer
to the separate links for additional information about each of the routines.

0 = A dummy scoring routine

1 = The ”scattwin” scoring routine
2 = The ”list mode” scoring routine
3 = The ”forced collimation” scoring routine
4 = The "penetrate" scoring routine

Note: Refer to the "Scoring Routines" chapter for additional information.

Index-85: This option allows for certain results to be written to a text file in a comma-separated
value (CSV) format. The advantage is that it is then easy to import results into, e.g. an Excel
sheet. There are different ways of writing data, and the output is described below. Please review
the page on "The RES file" for further information on each of the parameters stated below. In
general, the file name will be output.csv. However, if a third file is given at the command line,
then this file name will be used. For example, by giving the command "simind input output/85:1
csvname", a file named csvname.csv will be created.

• value = 1: 'ResFile','ComReg','FulReg', 'PilReg','Foto1','Compt1','Effici(W)','SD','Effici(D)',

'Cps/MBq','Cpm/mCi','Peak/C1','Peak/C2','Peak/Tot', 'FWHM','FWTM'

• value = 2: 'ResFile','ScattPrim', 'ScattTot','ScatterOrder from 1 to ISCUT','FWHM','FWTM'

• value = 3: 'File','HV0','HoleDiam','CollZ','CollDep1','CollDep2','CollDep3','CollDep4',
'CpsMBq', 'Geom (D)','Geom (W)','Pen (D)','Pen (W)','Scatt (D)','Scatt (W)'

• value = 4: Scatter/Total (W), ScatterWeight(W), PrimaryWeight(W) for each projection

angle. In this option a new file is always created!

• value = 5: The activities and concentrations for the voxel-based phantom are written.

Note: The results of the FWHM and FWTM are only relevant if a point source located at the centre
(0,0,0) has been simulated.

Note: If you use this option on the command line with a negative value, i.e. /85:-1, then a blank
line is added to the CSV file. This is useful to separate certain data from others.

Version 6.2
 The SIMIND Monte Carlo Program Page 23 of 64

Note: If no CSV file is present, then a new file is created. Otherwise, a line is added to the present
content. If you use this option in a script, be sure to add a command that deletes the CSV file
before starting the simulation series.

Note: Send Michael an email if you have a specific combination of results that you would like to
be sent to a CSV file that is not covered by the above options.

Index-86: This option allows for the simulation of dynamic scintigraphy. There are several
requirements that need to be fulfilled.

1. This option only works with an image-based phantom that is based on codes. For additional
details about these phantoms, please refer to the information related to Index-14 and Index-
15.
2. For each of the sources having an activity distribution that will change dynamically, a time-
activity curve (TAC) needs to be available in a data file and in a specific file format, which is
described below.
3. The actual time sequence is provided by the switch /TS. The value for this switch is a character
string with the format xA1_B1_A2_B2_...An_Bnx, where A is the time per frame for sequence
“n” and B is the number of images for that sequence.

The file format for the TAC is as follows:

16-bit integer = number of points in the curve

32-bit integer = not currently in use, so it can be anything
32-bit float array = an array of ‘float’ values for the time points and time steps indicated with a
negative number.

The first value in the array of time points must be negative and the magnitude of the value reflects
the time step for the TAC values that follow. If a negative number is found as the next number in
the time point sequence, then the time step is changed to this value. Thus, it is possible to have a
single TAC file with different time bins. The actual value is interpolated to the actual time in the
/TS switch (which is the mid-point in the time interval).

When an image for a new time point (defined by the /TS switch) is simulated, simind creates a
new activity map from the information in the code file. To connect to the TAC file in order to
obtain the proper activity distribution, the format for the code file (*.zub file) needs to be according
to that of the following example (using the xcat phantom).

hrt_myoLV 1 1040 1,hrt_myolv.tac

hrt_myoRV 2 1035 1,hrt_myorv.tac
hrt_myoLA 3 1053 1,hrt_myola.tac
hrt_myoRA 4 1048 1,hrt_myora.tac
body 9 1000 1,body.tac
muscle 10 1050 1,muscle.tac
brain 11 1040 1,brain.tac
sinus 12 1000 1,sinus.tac
liver 13 1060 1,liver.tac
gall_bladder 14 1026 1,gall_bladder.tac
l_lung 15 260 1,l_lung.tac
r_lung 16 260 1,r_lung.tac
esophagus 17 1000 1,esophagus.tac
laryngopharynx_activity 18 1000 1,laryngopharynx_activity.tac

The difference here, as compared to the template file vox_man.zub located in the smc_dir folder, is
the kinetic TAC file. The number that precedes the file name must be there, and it can be used as
a scaling factor. The extension needs to be ’tac’. For those organs that do not have a TAC associated

Version 6.2
 The SIMIND Monte Carlo Program Page 24 of 64

to it, a fixed activity value based on the number in the fourth column will be used (standard code
usage).

Version 6.2
 The SIMIND Monte Carlo Program Page 25 of 64

Index Page 6

It is possible to perform a Monte Carlo simulation of solid-state cadmium zinc telluride (CZT) in
this version of simind. The detection of an event in these types of detectors is determined by
released charges (electron/hole pairs) instead of scintillation light. Each image pixel has its own
corresponding detector. There are several properties that differ from those of the common Anger-
type scintillation camera detection of radiation. These are described in a paper that also includes
several validation studies of the energy spectrum, system sensitivity, and spatial resolution. For
additional information, please refer to reference (5).

Index-91: The voltage between the anode and the cathode (V). A typical value is 600 V.

Index-92: This value is the mobility life for the electrons. The parameter is expressed in units of
10-3cm2/V. A typical value would be 5.

Index-93: This is the mobility life for the holes, and it is expressed in units of 10-3cm2/V. The
mobility life is generally lower than that for electrons, and a typical value would be 0.4.

Index-94: This is the gap between the active pixels, expressed as a fraction of the detector size.

Index-95: This is the physical detector size. The detector is assumed to be of square shape [cm].

Index-96: Tau is an exponential decay term that facilitates the simplification of Ramo’s
formulation of induced charge (6-8).

Index-97: The Hecht formula simulating signal loss due to recombination. If the value is 1, then
the Hecht formula for both electrons and holes is used. If the value is zero, then

Index-98: Some different energy resolution can be modelled. If the value is 0, then the energy
resolution model described by Chen (9) is used. If the value is 1, then the Chen model is used, but
it can be modified by the energy resolution parameters provided in the simind.ini file. If the
value is 2, then a Gaussian model is used. For additional information, please refer to reference

Version 6.2
 The SIMIND Monte Carlo Program Page 26 of 64

(5).

Index-99: The cloud of charge also has lateral mobility when drifting towards the anode. This
drift is modelled as a Gaussian function (10). The value entered should be the diffusion constant.

Simulation Flags

Option 3 in the main menu is a menu consisting of logical flags that control various operational
aspects of simind. These flags are Boolean and are set/reset by giving the appropriate flag
number. The flags are defined as follows:
Flag-1: Determines whether on-line printout and results will be written to the terminal during
the actual simulation. This flag should be turned off when submitting a command procedure to a
batch queue.

Flag-2: Determines whether a binary image file (*.bim) that will store a calculated planar image
should be created. This flag does not affect the actual calculation of the matrix. The file format
for the image is 32-bit float.

Flag-3: Determines whether a binary energy pulse-height distribution file (*.bis) should be
created. As for Flag-2, this does not affect the calculation of the different detector parameters that
are derived from the energy spectra. The file format for the spectrum file is 32-bit float.

Flag-4: Includes simulation of a scintillation camera collimator. The values in Index-45 to -53 are
not in use although they are defined.

Flag-5: Determines if a SPECT simulation will be performed. If this flag is set, then the total
number of photon histories in the SPECT simulation will be the number of photon histories per
projection multiplied by the number of projection angles. Invoking Flag-5 will set Flag-2 to FALSE
because a special SPECT file with the extension *.a00 is produced instead. The file format for the
projection images is 32-bit float.

Version 6.2
 The SIMIND Monte Carlo Program Page 27 of 64

Flag-6: Includes the simulation of characteristic k-a X-ray photons emitted from a photoelectric
absorption site in the crystal, the collimator, and the protective cover. However, whether
interactions in these different compartments are to be actually simulated depends on what is set
in the simind.ini file located in the smc_dir folder.

Flag-7: Includes a volume behind the crystal to simulate scattering in light guides and PMT. The
thickness is given by Index-11. The scatter order is 3 by default, and it can be changed by editing
the simind.ini file located in the smc_dir folder.

Flag-8: Controls the sequence of the random number generator. If it is set to FALSE, then the
same initial seed will be used in every simulation. Then, no statistical differences will be observed
between different simulations of identical imaging systems. This function can be useful when
tracing errors. However, for a normal procedure, the flag should be set to TRUE.

Flag-9: Simulation of transmission SPECT imaging. Transmission simulations require an image-

based phantom to work.

Flag-10: This option includes the simulation of the protective cover. The cover is defined by its
cross-section file (in the main menu of change) and the thickness of the layer (Index-13).

Flag-11: This option takes into account the simulation of the phantom. The dimensions of the
phantom are given by Index-5 to -7 or, if Index-14 is negative, by Index-31 to -38. Be sure to have
a valid file name for the non-homogeneous phantom file when setting Index-14 to a negative
number.

Flag-12: This flag incorporates the simulation of both the limited energy resolution and the
intrinsic spatial resolution in the simulation. The actual values for the system are defined by
Index-22 and -23. If this flag is reset, then the energy pulse-height spectra will reflect the
imparted energy in the scintillation crystal and not counts from simulated scintillation light that
may be affected by statistical fluctuations.

Flag-13: This option forces the photons to interact in the crystal by invoking a ’forced direction’
variance reduction technique. It is useful when simulating high-energy photons that are
impinging on thin crystals or when simulating a low-density detector. The calculated detector
parameter is not affected by this option because the weight of each photon is adjusted for the
forced interaction.

Flag-14: If this flag is TRUE, then the files storing the image and the SPECT projection data will
have a header of the Interfile version 3.3 format. This format has been developed to establish a
mode of communication between different nuclear medicine units. The specific simind Interfile
parameters have a prefix of ;# and are thus treated as comments by an external system.

Flag-15: If this flag is TRUE, then a file of the phantom that is re-sampled to the pixel size and
slice thickness for the corresponding SPECT image is stored. The base name of the file is the same
as that of the density map file (Index-14 in the main menu). The contents of the file are in units
of density times 1000. The extension is *.ict for the actual images and *.hct for the interfile header.
Note that these files are only created if voxel-based phantoms are used and Flag-11, which allows
for interaction in the phantom, is set to TRUE.

SCORING ROUTINES
The following scoring routines are now included in simind. For further details, refer to the
specific information for the particular routine by clicking the appropriate link to the left.

Version 6.2
 The SIMIND Monte Carlo Program Page 28 of 64

• 0 = dummy routine.
• 1 = A routine that provides multiple energy window simulations (Scattwin).
• 2 = A list mode scoring routine (Listmode).
• 3 = Forced collimation routine.
• 4 = Separation of events into components (Penetrate).

The ”scattwin” scoring routine

This routine has been developed as an aid to perform multiple window simulations easily without
the need for programming. Because f90 and later compilers allow for dynamical allocatable
arrays, multiple windows can easily be defined at runtime. At runtime, simind looks for a file
whose name is the same as that of the input file but with the extension *.win. If this cannot be
found, then the program looks for the file scattwin.win. The structure of the file is relatively
simple, as shown in the following example:
126.0,154.0,0
92.0,124.0,0
124.0,126.0,0
154.0,156.0,0
124.0,126.0,-4
Each row represents a window where the first number is the lower threshold in keV and the
second number is the upper threshold. The third number is a flag indicating two different
situations.
• If the number is negative, '-n', then this window will add data from the window defined in row
'n' in the window file. This automatically sums two windows. In the example above, data in
window 4 is added to the last window.
• If the number is greater than 0, then the scatter images saved for this energy window only
contain events from photons scattered with order 'n' . You can then, e.g. investigate the scatter
image in an energy window for each order using the following *.win file:
126.0,154.0,1
126.0,154.0,2
126.0,154.0,3
126.0,154.0,4
126.0,154.0,5
The files created by scattwin have the format
out_air_w1.a00
out_tot_w1.a00
out_sca_w1.a00
where air, tot, and sca stand for 'air','total', and 'scatter', respectively, and ’w1’ represents the first
energy window. Primary images can be obtained by subtracting the scatter images from the total
images. The number in the extension stands for the window number, i.e. the row number in the
*.win file.
For easy use of the simind-generated scatter images and for the castor reconstruction program, it
is possible to write two keywords in the *.win file depending on whether you select the DW method
or the TEW method.
The DEW method: The scatter is estimated on the basis of the previous energy window. For
example, assume this energy window file
126.0,154.0,0
92.0,124.0.74,dew
This now works in such that the scatter file for the second energy window 92-124 is scaled to
match the energy window in the previous setting, i.e. the scaling factor will be [154-126]/[124-92].
In addition, the data will be scaled optionally by the constant 0.5. Thus, the total scaling will be
[154-126]/[124-92]*0.74. The scaled version will be stored in the file *sca_w2.a00. It is possible to

Version 6.2
 The SIMIND Monte Carlo Program Page 29 of 64

add a new set of window settings and obtain two different scaled scatter files in the same
simulation.
The TEW method: This works as follows. Assume an energy window file as in this example:
126.0,154.0,0
154.0,158.0,0
122.0,126.0,1.0,tew
Now, the scaling assumes that the primary photopeak window is the second previous window
setting, i.e. in this case, the first line. The scatter data in the third window will then be scaled by
the factor [Data2*W2 + Data3*W3]*0.5 * W1, where W1 = [154-126], W2 = [158-154], and W3 = [126-
122]. Here, there is no optional scaling, i.e. the factor is 1.0 but there could have been a value.
The triple energy window data will be stored in *_sca_w3.a00.
The BIS file saved by this routine contains spectra for the total photons (spectra 1), primary
unscatter spectrum (spectra 2) and scattered photons (spectra 3) and the for each of the scatter
orders selected (1,2,3,4 …...) The spectra can be extracted using the bis program, e.g.
bis test spectra1/spe:1 # Extracted spectra = the total events
bis test spectra2/spe:2 # Extracted spectra = the primary events
bis test spectra3/spe:3 # Extracted spectra = the scatter events
bis test spectra4/spe:4 # Extracted spectra = the first scatter order

The results for each of the energy windows is printed in the middle section of the result file.

Switches

/FW This switch can be used to define the window file name. For example, if a window file
has the name "windows.win", then this file is used if the command is followed by the
switch "/FW:windows". Note that no extension ".win" is needed here.

The ”listmode” scoring routine

This routine saves information about the photon history in a binary list mode file. The file will
have the extension *.lmf. To save disk space, each record of a history in the binary file will have
nine 16-bit integer values, one 32-bit float value, and one 8-bit value in the order indicated in the
table. To preserve the decimal precision, the floating values have been multiplied with a scaling
factor before being truncated to integers. Therefore, you should divide the values by the
corresponding scaling factors before using them.
OU = 0 (default)
1 = x0 * 100 integer, 16-bit X0,Y0,Z0 is the sampled emission point
2 = y0 * 100 integer, 16-bit
3 = z0 * 100 integer, 16-bit

4 = xphant * 100 integer, 16-bit This is the last interaction point in the
5 = yphant * 100 integer, 16-bit phantom before leaving the phantom.
6 = zphant * 100 integer, 16-bit

7 = xcrystal * 100 integer, 16-bit The X,Y,Z position of the centroid of

8 = ycrystal * 100 integer, 16-bit interactions in the crystal. This location can
9 = zcrystal * 100 integer, 16-bit be convolved with the spatial intrinsic
resolution if this flag has been selected.

10 = energy in integer, 16-bit Units are keV. A negative sign indicates that
crystal * 10 a new projection has started. Please be aware
of this in your reformatting program. This
has been made to save file space.

Version 6.2
 The SIMIND Monte Carlo Program Page 30 of 64

11 = photon weight float,64-bit If (2) = 0, this is a primary photon that has

12 = scatter order integer, 8-bit not been scattered in the phantom.

Switches

/OU If this switch is given, where n is a given number, then different list mode formats can
be stored. For the present version, only one format is defined, which is given below.

The ”forced collimation” scoring routine

This is an option of including a forced collimation in the simulation to speed up the simulation.
Owing to technical reasons, this option cannot be made as an ordinary collimator routine. Instead,
it must be a scoring routine. This means that if the forced collimator option is selected, in the
present version, it cannot be combined with other scoring routines, such as scattwin.
The principle of the method is that multiple images are created on the basis of the distance
between the lower collimator surface and the last interaction point. When the simulation of a
certain projection is completed, each of these multiple images is convolved with a Gaussian
function with sigma calculated from the distance between the plane for those interaction points
and the camera surface. Subsequently, the images are added to a final image.
The following assumptions have been made:
• The ordinary collimator is turned off and the direction of the photon after the last
interaction point is directed perpendicular to the crystal surface. The attenuation is then
calculated on the basis of the distance from the last interaction point to the surface of the
phantom. This means that the difference in attenuation as a function of the azimuthal
angle close to the surfaces is not taken into consideration.
• The shape of the Gaussian will be symmetric.
• No backscatter from the volume behind the crystal is included.
• No events from penetration or scatter in the collimator are included.
To run the program, select Index-84 = 3. The default distance between the planes is 1.0 cm and
64 planes are generated. This can be changed by the switch /DP (distance between planes), so if
/DP:0.5 is given, then 64/0.5 = 128 planes are used. Finally, the images will be saved to *.bim or
*.a00 files.
Note: Because this technically is a scoring routine, it is not possible to combine this feature with
the ’Scattwin’, ’Penetrate’, or ‘List Model’ routines in this version.

The ”penetrate” scoring routine

This scoring routine separates the components of an image or spectrum. Further, 18 different
images and energy spectra are stored. The file names for the images are stored with the extension
*.b01 to *.b18. Different energy spectra are stored in consecutive order in the *.bis file. The scoring
routine is run by selecting Index-84=4. The different image files and spectra are separated
according to the following table and where events are coming from

*.b01= All type of interactions.

*.b18= photons without scattering and attenuation in the phantom.
*.b19 = photons without scattering and attenuation in the phantom and geometrically
collimated for primary photon energy (indicated by a negative energy in the *.isd
file).

The following files are with no backscatter from the compartment behind the crystal and show
events coming from.

Version 6.2
 The SIMIND Monte Carlo Program Page 31 of 64

*.b02= Geometrical collimated primary attenuated photons from the phantom.

*.b03= Penetration of a septa from primary attenuated photons from the phantom.
*.b04= Scatter from the collimator from primary attenuated photons from the
*.b05= phantom.
*.b06= X-rays from the collimator from primary attenuated photons from the
*.b07= phantom.
*.b08= Geometrical collimated from scattered photons from the phantom.
*.b09 = Penetration of a septa from scattered photons from the phantom.
Scatter from the collimator from scattered photons from the phantom
X-rays from the collimator from scattered photons from the phantom.

The following are with backscatter from the compartment behind the crystal.

*.b10 = Geometrical collimated from primary attenuated photons from the phantom.
*.b11 = Penetration of a septa from primary attenuated photons from the phantom.
*.b12 = Scatter from the collimator from primary attenuated photons from the
*.b13 = phantom.
*.b14 = X-rays from the collimator from primary attenuated photons from the
*.b15 = phantom.
*.b16 = Geometrical collimated from scattered photons from the phantom.
*.b17 = Penetration of a septa from scattered photons from the phantom.
Scatter from the collimator from scattered photons from the phantom.
X-rays from the collimator from scattered photons from the phantom.
The content in the energy spectrum file follows the order of appearance described above, but
different spectra are stored in a single file. To access the *.bis file, you need to open the file and
read in to a 32-bit float vector dimension with two dimensions according to the spectra
(NrChannels,18). The value of NrChannels is defined by Index-80. You can then select the proper
energy spectra by using the same numbers as above. For example, the distribution of events from
photons scattered in the collimator and in the phantom but without backscattering will be plotted
as spectra(*,8).

SOURCE ROUTINES
The following source routines are now included in simind in addition to the geometrical shapes
defined by Index-1 to -5, as described above.

• Index-15 < 0 -> image-based source routine linked by default. The magnitude of the value
determines the type of standard image file. For additional information, refer to the next
sub-chapter.
• Index-15 = 6 -> myocardial source routine.
• Index-15 = 7 -> multiple source routine.

The ”image-based sources” source routine

This simulates a source distribution in three dimensions by using the information in a file that
contains integer matrices when generating the location for the decays in the phantom. With this
procedure, very geometrically complex source distributions can be simulated in a relatively easy
manner. The dimension of each map is defined by Index-79 and -82 in the change program. The
number of images is unlimited. The sources are normally centred on the origin of the coordinate
system but can be shifted by Index-16 to -18 in the change program. The routine simulates
decays randomly within the voxel volume corresponding to the location of the source map voxel.
The number of decays from that location is determined from the contents of the source map voxel.
A pixel size (/PX described below) is needed for conversion to physical coordinates because the
location to a matrix cell is referred by relative indices (i,j,k). Note that this pixel size has no
connection to the pixel size that defines the projections being simulated (Index-28). Depending on

Version 6.2
 The SIMIND Monte Carlo Program Page 32 of 64

the desired direction of the source, the half-axial source length (that determines the slice
thickness) is defined by Index-2-4. The routine reads the number of images stored in the input
file. The name of the file is defined by Value 14 in the main menu of change. The default file
extension is *.smi, but it may be different depending on the selected type of phantom.
A start block can be defined to avoid file headers, and a limited number of images can be defined.
The routine scans through the source maps and sums up all voxels values. The sum will then be
equal to the number of simulated photon histories. Thus, when simulating voxel-based source
distribution, Index-26 in change is not used. To use the ”Image-Based Routine”, Index-15 in
change needs to be negative.
A tumour file can be read into the source map routine and the count is incremented according to
the definition in the tumour file. One tumour is defined by one row in the file as floating-point
values separated by a space.
• The first three values (1-3) define the centre of the tumour in pixel units.
• The next values (4-6) refer to the centre location of the tumour also in pixel units.
• Value 7 is the voxel value and is a relative value associated with the main source
distribution.
• Value 8 is the density of the tumour. If the value is 0, then the density will be given by
the original values in the voxels that the sphere occupies.
• Value 9 and 10 determine the distribution of activity in the tumour. If both values are 0,
then it will be a uniform distributed activity. Otherwise, the shape of the activity
distribution will change according to a Butterworth filter, where 9 is the cut-off and 10 is
the order (steepness of the curve).
Note: Index-26 is not used when simulating source maps. Instead, the number of photon histories
is determined by calculating the sum of the voxel values in the source map. This ensures that the
proper distribution is sampled. If the number of histories is too low to obtain good statistics in the
final projections, then the switch /NN (described below) can be used to easily increase the number
of photon histories by multiplication with a factor.

Switches

/PX Pixel size of the source maps. This switch is needed for proper function [cm].

/DI The general direction of the source map defined by an integer (1-3) in the same way as
the non-homogeneous phantom (Index-32 in change). No switch means a direction when
indices i,j correspond to y,z.

/TH The value sets the slice thickness for the images [cm].

/SB Start block (512 bytes) when reading source maps from the file. No value means starting
from beginning of the file.

/1S This switch defines the position in the file of the first image to be used. If omitted, the
first image is the first image in the file.

/NN Multiplier that scales the number of counts in the source maps with an integer. This is
employed to allow longer simulations with better statistics using the same source map
file. Note that the magnitude of the final event in the images does not depend on this
value. The idea is that the count levels in the images should reflect the activity level in
MBq as defined by Index-25 and measured with an acquisition time of 1 s regardless of
the number of photons simulated. Thus, a larger value of NN improves the statistics in
the image but the total sum of events in the image should remain the same.

/IF An input tumour file is read in. The file name should be the same as the main input file
name but with the extension *.inp. The format is the same as that for the multiple source
routine with the addition of an optional density value (times 1000 according to the
same logic as that of the densities in the *.zub files) as the number 7 in the sequence and

Version 6.2
The SIMIND Monte Carlo Program Page 33 of 64

two values that define the distribution of the activity in the tumour. If the number 7 is
absent or has a value equal to 0.0, then the tumour density will be the same as that of
the organ it occupies. Note that if a tumour infiltrates two organs, the density of the
tumour will have two values if it is not explicitly defined.
There are two types of tumours. Tumour type 1 has the centre in the pixel defined by
value 4,5,6 in the *.inp file. If value 1,2,3 is equal to zero, then the tumour will be a
single pixel. Tumour type 2 has the centre at the surface of the pixel that is between i,j,x
and i-1,j-1,k-1. The minimum size for this tumour is 8 pixels.
Note the following options:
1. /IF = The tumour type as defined in the simind.ini is used for all tumours defined
in the *.inp file.
2. /IF:2 = Tumour type 2 is used for all tumours in the *.inp file.
3. /IF:-2 = Tumour type 2 is used for all tumours in the *.inp file but the background
activity is set to zero.
4. /IF:2.3 = Tumour type 2 is used with the Zubal background but only the tumour
corresponding to row 2 in the inp file is used.

The ”myocardiac” source routine

This is a source routine that simulates a heart consisting of a combination of a half sphere shell
and a cylindrical shell, both with the same thickness. The source distribution can very accurately
be matched to the physical cardiac insert manufactured by DataSPECTRUM Inc. A defect can be
inserted in the half cylinder at an arbitrary location. The boundaries of the heart are defined by
Index-2,3,4 in the change program.
Switches

/A1 Shift of the heart in the xy-direction (default value 122 degrees)
/A2 Shift of the heart in the yz-direction (default value 52 degrees)
/A3 Shift of the heart in the zx-direction (default value 0 degrees)

/L1 Location of defect 0 = ant, 180 = inferior, -999 = No defect

/L2 Angular size of the defect (± degrees) (default value 30 degrees)
/L3 Start from Base (default value 0.6 cm)
/L4 Extent of defect i axis direction (default value 2.0 cm)
/L5 Transgression in % (default value 1.0 = full transgression)
/L6 Activity ratio in defect (default value 1.0 = same as surrounding activity)
/M1 Thickness of the myocardial wall (default value 1.0 cm)
/M2 Thickness of the plastic wall (default value 0.2 cm)
/M3 Total length of the chamber (default value 8.0 cm)
/M4 Total diameter of the chamber (default value 6.1 cm)

The ”multiple spheres” source routine

This source routine simulates a Jaszczak cylindrical uniform source phantom (Data Spectrum,
Inc., Chapel Hill, North Carolina). The sources must be defined in an input text file with the
extension *.inp. The routine first searches for a file named input.inp, where input is the
base name of the input smc file created by change. If this does not exist, then the routine
attempts to open a default input file jaszak.inp. If this is not found, then simind is
terminated.

Version 6.2
 The SIMIND Monte Carlo Program Page 34 of 64

The input file: This file is an ASCII file where each row defines a sphere. Float 32-bit values
separated by a comma are used. The values at the first three positions define the size of the source.
The values at positions 4-6 are the location in the coordinate system for the centre of the source.
Note that this is not the centre of gravity but the starting point for each of the axes that define
the ellipsoid. The value at position 7 in the row is a relative value of the desired activity
concentration (MBq/cc). The program normalizes the decays in each source so that the number of
photon histories per projection provided by Index-26 in change can be simulated. However, if
these values are negative, then they are treated as absolute activity concentrations in units of
MBq/cc. The total activity (Index-25) is then recalculated to match the sum of the activities for all
spheres and, if defined, also together with the background activity.
The value at position 8 determines whether the source shape will be
• a sphere (0.0)
• a cylindrical horizontal rod (1.0)
• a rectangular horizontal rod (2.0)
• a hexagonal horizontal rod (3.0)
• a cylindrical vertical rod (4.0)
• a cone (5.0)
As an example, consider the following two lines in the input file.
1.0,1.0,1.0,0.0,0.0,0.0,1.5,0.0
2.0,2.0,2.0,3.0,0.0,0.0,3.0,1.0
The first source is spherical with a radius of 1.0 cm and is centred at the origin. The relative
activity concentration is 1.5. The second source is cylindrical with a radius of 2 cm and the source
is shifted 3 cm towards the positive x-axis. The relative activity concentration in this source is
twice as much as that in the first source.
Switches

/BG This switch defines the background activity concentration in the same units as the
concentration defined for sources in the input file.

/HO This switch generates hot spheres, i.e. no background activity and only activity in the
sources defined in the input file.

/CO This switch results in a cold sphere simulation where a uniform distribution of decays is
simulated by information from Index -2 to -4 and -15 in change but with the exclusion of
the decays within the spheres, as defined by the input file. The activity concentration has
no meaning here.

/IF A file name for the input file with the different source definitions can be given. The extension
*.inp is assumed.

/DI Change direction of the whole source distribution. Values can be 1,2 or 3

If the

RUNNING SIMIND

A simind command at the operating system prompt consists of the following parts:
• The program name, simind.
• An input file (optional).
• An output file (optional).
• Control switches (optional).

Version 6.2
 The SIMIND Monte Carlo Program Page 35 of 64

The example below considers a Windows operating system with possible differences from other
operating systems in the prompt and the upper case of the characters

simind input output/sc:3

The input file in the command line above, input.smc, defines a file created by change that
contains the data for the particular simulation. No extensions are necessary because change
always creates files with the *.smc extension. The output file(s) created by simind stores basic
results from the simulation, a binary image, and an energy pulse-height spectrum. If not explicitly
given, the output base file name, output, will be the same as the input file name, but different
files are separated by different extensions. If no input file name is given, the default file,
simind.smc, will be used and the base name of the output file will also be simind.

Example of the program running on a PC

The user can control the input parameters created originally by change at the command level.
This means that a single data file (e.g. input.smc) can be used for multiple simulations. Each
value given by an index or by a flag in change can be overwritten by a matching control switch.
By giving a two-character alphanumeric corresponding value, this new value takes precedence
over the value defined by change. Additional characters may be given for clarity but only the first
two are used. For example, overwritten the value in the input file input.smc to change the
number of histories (Index-26 in change) to 10,000 is accomplished by the following command:
simind input output/26:10
The character ":" or "=" can be used to separate data from the switch name. Note that Index-1 to -
9 must be assigned 01-09 in order to fulfil the minimum two-character definition of a switch. The
two switches "TR" and "FA" (TRUE and FALSE) are used to control the simulation flags defined

Version 6.2
 The SIMIND Monte Carlo Program Page 36 of 64

in change. The following command can be entered to reset Flag-1 (indicating no output to the
screen) and set simulation Flag-5 (invoke a SPECT simulation):
simind input output/fa:1/tr:5

The control switches may be placed arbitrarily together with the program name, the input file, or
the output file. Input files, output files, and program names are separated by spaces. Trailing
spaces are removed.
Limitations for Linux/UNIX/MacOS
There will be a conflict between the slashes used by simind and the slashes used to define a
certain directory in Linux/UNIX. simind cannot distinguish between a slash that points to a
switch and a slash that is part of a UNIX file directory. To overcome this problem, the directory
name can be specified using two backslashes '\\'. simind will, after being decoded for control
switches, change these backslashes to ordinary slashes '/' so that the result files can be stored in
other directories.
It is also not possible to add control switches directly to the program name in Linux/UNIX. The
operating system assumes that this will be the program name. Thus, control switches on a
Linux/UNIX machine should be either added after the input or output file or separated by spaces.
All file names are converted to lower case to simplify the coding and comparability for different
operating systems.

Basic runtime-switches for simind

/CC Collimator code in a format according to the description in the change program. Note that
you should not include the '*', i.e. the character to be used as a prefix in the change
program. For example, simind input/cc:ge-legp gives this collimator in the simulation
regarding what is given in the smc file [character].

/DF This switch is used when simulating a heartbeat and/or respiratory movements using
MCAT/NCAT/XCAT phantom. If the base name is test, then by adding the switch /DF:3,
simind will try and open a file named test_atn_3.bin. If this switch is not given, then it
will open the file test_atn_avg.bin, which is the standard name for the average density
image.

/ES An energy offset relative to Index-1 and expressed in keV for the centre energy of the photo
peak window. A negative value means a lower value relative to Index-1. If this shift is not
given, then the centre value of the energy window is defined by Index-1.

/FZ The name for the Zubal file can be given as the parameter. Give the name without the
extension 'zub' [character].

/FI The name for the Input file can be given as the parameter. Give the name without the
extension 'isd' [character].

/FD The base name for the density map (*.dmi) or the NCAT attenuation map (*.bin)
[character].

/FS The base name for the source map (*.smi) or the NCAT activity map (*.bin) [character].

/I2 Image files are stored as 16-bit integer matrices. By default, a scaling factor is calculated
from the first projection so that the maximum in that projection will be 1000. If a data value
is gives after /I2, then that value will be used as a scaling factor. For example, /I2:150 scales
the data to a maximum value of 150. If the value is negative, Poisson noise is added to the
projection data.

/IN With this switch, you can change any value in the simind.ini file at the execution level.
The switch contains two parameters. The first is the ini number, which is given in the

Version 6.2
 The SIMIND Monte Carlo Program Page 37 of 64

simind.ini file. The second value is the actual value that is to be changed. For example,
/IN:x4,8x sets the maximum scatter order in the backscatter layer to 8. Here, ‘x’ is used as
brackets because real brackets do not work in the UNIX environment.

/LO The number of photon histories before printout of the on-going status line. Requires a data
value. The default value is 1000.

/LF A sampling technique based on linear sampling of the polar angle for the photon direction
is used. Note that this is only useful when simulating the collimator routine that allows for
penetration (Index-53 = 1).

/OR Parameter that changes the orientation of the density map. Value 2 transposes the maps,
value 3 swaps the maps such that Index-(1,1) is at the lower left, and value 4 both
transposes and swaps.

/PR Start simulation at projection number given by the switch. Default is projection 1. If the
value is negative, then this indicates the stop projection number. The order of positive and
negative numbers is of no importance.

/PU In some cases, it is more convenient to express the centre of the source in terms of pixel
units instead of cm and let the computer calculate the physical centre position. By using
the /PU (= pixel units) switch, one can give the shift of the source in pixel units according
to
simind input/16:38/17:128/18:64/PU
From the phantom length and the pixel size for the density maps, the pixel size is calculated
as the true source shift in cm. If the command looks like
simind input/16:38/17:128/18:64/PU:0.25
then the position is scaled down by a factor of four. This is used when the pixel unit is valid,
e.g. a 512 matrix, but the actual density map used in the simulation is only a 128 matrix.

/QF Quit the simulation if an earlier result file exists. This flag is useful if a batch simulation
has been going on and the computer has been restarted following a power failure. This may
then prevent recalculation of already existing data files. Note that for this to work as
intended, there should not be any *.res before the start of the simulation.

/RR The random number generator is based on seeds that are stored in *.num files in the
smc_dir directory. When simind starts, it reads the seed, and when the simulation ends,
the last seed is stored in the files for the next run. However, when running on clusters or
machines with multiple cores, multiple runs may read the same seed and thus duplicate
the results. If different integer values are set for this /RR switch, random numbers are
created at the very beginning of the run on the basis of this value. This ensures that when
the actual simulation begins, the random numbers will be different. The most convenient
way to use this switch is of course when defining the run in a script.

/SC The maximum number of scatter orders allowed in the phantom. Requires a data value. A
large number requires a longer computational time because more scattered photons will be
followed in the phantom. In some cases (e.g. when simulating images obtained in a narrow
energy window), a lower value may be sufficient because the energy window rejects the
lower energy photon with a low probability. Therefore, the user must be careful so that the
proper number of scatter orders is simulated for the particular simulation. It may be
necessary to perform some 'trial-and-error' simulations here to fully understand how /SC
works. The default value is 3 and this value can be changed in the simind.ini file. If the
scatter order is a negative value, only the scatter distribution is simulated and stored in
the images and spectrum.

/SF This switch is used when simulating a heartbeat and/or respiratory movements using
MCAT/NCAT/XCAT phantoms. Because each segment from MCAT/NCAT/XCAT has its
own image file with a particular number, this switch makes a patch to the file name of

Version 6.2
 The SIMIND Monte Carlo Program Page 38 of 64

the density file. For example, suppose that you have generated 8 segments of a beating
heart using the base name test. The output from MCAT/NCAT/XCAT software gives you
files named test_act_1.bin, test_act_2.bin,....test_act_8.bin. If you add
the switch /SF:3, then you will use the source map test_act_3.bin even if the input
file says test. If this switch is not given, then it will open the file test_act_avg.bin,
which is the standard name for the average source image from MCAT/NCAT/XCAT.

/TS A time shift can be added to the date and time description in the interfile header. The
initial time and date is set in the simind.ini file and the value of the /TS switch is the
extension of this time expressed in hours. If the value is negative, then the activity value
(Index-25) will be reduced by a factor corresponding to the physical decay of the time
shift. This is done only if the half-life value (expressed in units of hours) is provided in
the *.isd file used. This option does not work for single photon energies. All the *.isd files
included in the smc_dir have information about the half-life.

/UA Set the density equal to the data buffer or if not present equal to 1.0 (= water). This
makes the phantom equal to uniform attenuation regardless of the initial voxel value in
the phantom.

/WB Whole-body simulation of anterior and posterior views. Note that you might need to
adjust the matrix size and detector length (Index-8). When this switch is given, the
following parameters are set.
Flag 15 - Interfile = .TRUE.
Flag 5 - SPECT simulation = .TRUE.
Flag 11 - Phantom flag = .TRUE.
Number of projections = 2
Rotation type = 0 (360 degrees)
Matrix size = n*512 x n*192 where n is the value of the WB Switch
Pixel size = initial pixel size/n (defined by Index-28 in change)
In the simind.ini file, there is a flag T/F that control the format of the aligned attenuation
map defined by Flag-15. If the flag in simind.ini is set to T, then one planar scout file with
the dimension [Index-76 x index-77] is created as an aligned attenuation map. The
numerical values are the projected my*d. If the flag is set to F, then the aligned file is
defined as transversal slices with the number of slices equal to Index-77 and the matrix
size (square) defined from Index-76. The value in the simind.ini file can be changed using
a negative sign in the /WB switch. For example, if the value in simind.ini is set to T,
then /wb:-2 will change it to F (or vice versa).

/Xn 'n' specifies the cross sections (1-6) that can be changed. /X1 to /X5 refer to the values 9
(soft), 10 (bone), 11(cover), 12(crystal), and 15 (backscatter), respectively, in the main
change menu. /X6 is the material for the collimator routine (if Index-53 = 1). The default
value for /X6 is defined in simind.ini [character].

Hidden flags
This table summarizes the 'hidden flags' in the change data. The negative sign of an index value
is used here as a flag and, in some cases, can also be used as the actual value. A negative value of
the following indices indicates:

1 Call to the isotop routine for the emission photons. The default routine linked is a
spectra routine that reads information about energy and abundance from a text file. The
default name for the file is spectra.isd. However, if there is a file that has the same base
name as the input SMC file but with extension *.isd, then this file will be used. The user
can also specify the name using the switch /IF:

Version 6.2
 The SIMIND Monte Carlo Program Page 39 of 64

14 Simulation of a non-homogeneous voxel-based phantom. There are three defined

phantoms available (-2 to -4). Depending on the value, different types of voxel-based
phantoms are used. simind is written such that the XCAT phantoms are easy to import
but there is, owing to license reasons, no XCAT phantom in the simind distribution.

15 Call to the source routine. Depending on the value, different types of voxel-based
phantoms are used.

19 Photons are emitted within a solid angle defined by the absolute value of the index. A
value of 0 indicates a narrow-beam simulation. Note that when this option is selected,
some of the results in the *.res files are not defined (e.g.the cps/MBq that goes to zero).

20,21 Treat the absolute value as a percentage FWHM of the energy resolution relative to the
photon energy in Index-1 and calculate the thresholds of the energy window from this
value.

25 If the activity is set to zero, then no reduction in the value owing to decay is calculated
when using this in combination with the TS switch. Otherwise, the value is changed on
the basis of the half-life provided in the isotop file (*.isd) if this value is present in the
file.

33 A negative value indicates the start block instead of the start image when reading from
the density map.

Non-uniform rotation. The value determines the ’security distance’, i.e. the minimal
41
distance between the surface and the collimator.

61 Call to the isotop routine for the transmission photons.

If a negative sign is given, then a space is written to the CSV file after the data have been
written. This is useful to separate the data row into groups. Use this in a switch value in
85
a script file rather than defining it in the SMC file. Otherwise, a space is written for each
simulation.

Some information about scripts

Monte Carlo simulations are by default very time consuming and therefore it is very convenient to
learn some basic things about how to use scripts in Windows and UNIX. A script can be very
advanced and nearly like a small program, but in its simplest form, it is a file that contains a series
of commands that could be given at the command line prompt. The reason for scripting in Monte
Carlo simulations is to set up a series of simulations that will start without the need for keyboard
commands. Usually, in an MC projection, one investigates some behaviour by varying one
parameter. The reason simind has the structure for execution is that this is very convenient for
scripts. For example, if one wants to investigate the spatial resolution in an image (FWHM) as a
function of five different distances between the source and the camera, one could in principle create
five change files (*.smc) where Index-12 is set to the five different distances. However, for a large
project, this becomes difficult to maintain. Therefore, the ability to input changes in the *.smc file
at the execution stage becomes important. Suppose that we want to study the distances 5, 10, 15,
20, and 30 cm. Then, we can write a script file that contains the following lines.

simind camera result1/12:5

simind camera result2/12:10
simind camera result3/12:15
simind camera result4/12:20
simind camera result5/12:30

In Windows, the script file must always have the extension .bat but in UNIX there is no such
requirement. I prefer to use .csh because I use the tcsh shell.

Version 6.2
 The SIMIND Monte Carlo Program Page 40 of 64

Thus, if the file run.bat (or run.csh) contain the above lines, you can use the following
commands to start the simulation:

c:\simind> run.bat (Windows)

simind# csh run.csh (Linux/MacOS)

Note that the name of the result file must be unique (result1, result2, etc.); otherwise, the files
from a previous simulation will be overwritten. The switch /12 refers to Index-12 in change, so
regardless of what it is set to in the camera.smc file, the value will be changed to the one given by
the /12 switch. All indices in change have a corresponding switch.

An advantage of scripting is that it is possible to pass information from the command line into the
script without editing the script itself. One can define command line parameter as numbers where
the number matches the position of a sequence of characters in the command. For Windows, the
command variables are then defined by a % character while in UNIX/Linux/MacOS, it is the $ sign.
Suppose that the script file (for Windows) looks like this instead.

simind camera result1%2%3/12:5%1

simind camera result2%2%3/12:10%1
simind camera result3%2%3/12:15%1
simind camera result4%2%3/12:20%1
simind camera result5%2%3/12:30%1

If the command now looks like this:

c:\simind> run.bat /01:200 (Windows)

$HOME/simind# csh run.csh /01:200 (Linux/Mac)

then /01:200 will be inserted at the place where %1 is, i.e. at the end of the command. This means
that I now overwrite Index-1 in the change file and instead simulate a photon energy of 200 keV.
Thus, we do not need to edit the run.bat file nor do we need to modify the camera.smc file. Now,
we have a problem because the result files will be overwritten if we use this for different energies.
Therefore, we modify the command to look like this:

c:\simind> run.bat /01:200 200kev (Windows)

$HOME/simind# csh run.csh /01:200 200kev (Linux/Mac)

Note that ”200kev” appears at the second position after the script file name and therefore the
”200kev” characters will be inserted at the position corresponding to %2. Note that %3 is not used
and therefore nothing is inserted at this position. The result name will have the format
result1_100kev.res. However, parameter %3 can be useful in the following example.

If one creates a new file called run1.bat, one can define commands in it that start the file run.bat
for different scenarios.

call run.bat /01:100/fa:12 _100kev _nores

call run.bat /01:200/fa:12 _200kev _nores
call run.bat /01:300/fa:12 _300kev _nores

call run.bat /01:100/tr:12 _100kev _res

call run.bat /01:200/tr:12 _200kev _res
call run.bat /01:300/tr:12 _300kev _res

This example uses run.bat to simulate five distances and photon energies with and without
energy resolutions. The third parameter is used to get unique file names and the first parameter
is used both to set the photo energy and to turn on or off the energy resolution (Flag-12 in change).
We now have a case where one script calls another script. In Windows, we use call for this

Version 6.2
 The SIMIND Monte Carlo Program Page 41 of 64

purpose. In UNIX, we can instead use csh to call another script. The result name will have the
format result1_100kev_nores.res; the underscore character makes it easier to understand
the name of the result file.

Depending on whether it is Windows or UNIX or sh,csh,tcsh, or bash shells in UNIX, there is

a syntax for performing a really complicated script involving loops, if-then-else statements, etc.,
but it is beyond the scope of this manual.

The “bim” program

The bim program is a simple utility program that performs certain calculations on the created
projection image in the *.bim file. The program is invoked in the same manner as simind, i.e.
giving an input file name and additional switches after the program name.
bim input output/switches
The default input file name is simind.bim and the output is simind.prn if no base name for the
output file is given. BIM outputs a very simple 'image', based on keyboard characters, of the matrix
together with calculated data and a two-row data of a profile through the matrix. The profile can
then be plotted or included in a spreadsheet program for further analysis. Wildcards are allowed
according to the rules described in Option 10 in the change description.

Switches
/HO Horizontal profile (default).
/-F Do not save result in file with extension 'prn'.
/VE Vertical profile.
/-O Do not write results to the screen.
/AD Number of added lines centred around /LI.
/LI Select a line number for the profile.
/SM Smooth matrix with a filter.
/1R Save only y-values in the prn file.
/2R Save only x- and y-values in the prn file.
/FW Calculate Full Width at % Maximum (def = 50).
/II Read I*2 matrix.
/IA Read I*2 matrix and remove sign.
/IU Read a I*2 matrix unsigned (0-65535).
/OI Write the matrix as a I*2 matrix with extension 'int'.
/OU Write the matrix as a I*2 matrix norm to 65534.
/OR Write a R*4 matrix with extension 'mtx'.
/+*- Algebra with matrices.
/YP Addition value to y-values (matrix).
/MA Addition slice from start image.
/UT Upper threshold limit (default = 100%).
/LT Lower threshold limit (default = 0%)
/NP Normalize slice peak to value.
/NA Normalize slice to slice area times value.
/LG Calculate logarithm log(mtx(i,j)+1).
/IM Starting with image number.
/LO Process a number of images.
/PL Simple plot on the screen.
/SB Start at block when reading a file.
/XP Add value to x-values (Slice).
/YM Multiply value to y-values (Matrix).

Version 6.2
 The SIMIND Monte Carlo Program Page 42 of 64

/NT Normalize slice to mtx area times value.

/64 Output only matrix. No profile data.
/CA Calculate region of interest.
/SW Swap bytes in 16-bit integer input array.
/XM Multiply value with x-values (Slice).

The “bis” program

The bis program works on the calculated energy spectra from simind. Its main purpose is to
generate a two-column ASCII string of the spectra to be used for plotting or printing. The default
format is FORMAT(2F12.3) but it can be changed if the source code is available. The program is
invoked in the same manner as simind.
bis input.bis output.prn/switches
The default input file is simind.bis and output is simind.bis if no base name for the output file is
given. Wildcards are allowed according to the rules described in Option 10 in the change
description.
Switches
/BI binary from ASCII PRN -> BIS
/CO Convolve with Gaussian. FWHM = 'value'*0.1
/+*- Algebra with files
/SM Smoothing with filter = 'value'
/SP Convert Spectrum Number
/LT Lower Limit Integration (CHANNEL)
/UT Higher Limit Integration (CHANNEL)
/AS ASCII from BINARY BIS -> PRN (default)
/NP Normalize Peak to 'value'
/YA Addition 'value' to y-values (Slice)
/YM Multiply 'value' to y-values (Slice)
/XA Addition 'value' to x-values (Slice)
/XM Multiply 'value' to x-values (Slice)
/H Help file (this table)
/UT Higher Limit Integration (CHANNEL)
/NR Number of channels in the BIS spectra
/OY Generate y-values only
/MX Maximum number of data points
/HI Generate a histogram in the prn file
/-O NO output to terminal
/CH New keV/channel
/NA Normalize Area to 'value'

The “smc2castor” program

By nature, simind is a virtual SPECT camera. This means that it only creates projection images.
No reconstruction software is included or supported. This has been a source of questions because
one wants to reconstruct data and stand-alone software is not always available. Although it is
possible on some clinical systems to import the Interfile file, it is often difficult and requires a
deeper understanding of the Interfile file format and how to make simind to write such a file
properly. In March 2017, the Castor reconstruction software was released for download in the
public domain. Information about the program can be found by visiting http://castor-project.org/.
However, the file format for this reconstruction software was not based on the Interfile, at least
with regard to the input data. Therefore, I have written a conversion program called smc2castor,

Version 6.2
 The SIMIND Monte Carlo Program Page 43 of 64

which makes it easy to reconstruct simind simulated projections. The general syntax for the
program is

smc2castor smc_spect_projs.h00 result_file [scatter file]/switches

where smc_spect_projs.h00 is the header file for the simind SPECT simulation and result_file is
the result file for the reconstructed data. The smc2castor program creates a castor file and can
spawn a reconstruction by adding the switch /ru. It is assumed that the castor software has been
installed properly according to the instruction. The CASTOR_CONFIG environmental variable is
set to the castor config folder, and the executables are accessible from the path.
The following switches are available:

/AT Include attenuation correction

/HE Help
/IT Number of iterations (default = 2)
/IT Number of iterations
/SI Save all iterations
/MB Activity
1=save in the phantom (default = 1 MBq) all iterations
/PR Projector
2=only last (default)
/RU Run the castor reconstruction
1=Classis Siddons (default)
/SF Scaling factor scatter (default = 1.0)
2=Joseph
/SU Number of subsets (default = 8)
/TI Time per projection (default = 1 s)

Description of some files of importance

The simind.ini file

simind initially reads a file, called simind.ini, that is located in the smc_dir folder. This file
includes default values for the program, but it is also possible to some extent to control the physics
used in simind. Each line has a number that can be used to change the default value to a new
value at runtime using the new switch /IN.

The error-file
This file includes the error text displayed by simind.

The Interfile-key file

This text file includes the tags necessary to write the header files in the Interfile format. It is not
meaningful for a user to edit this file. Note that there are several additions to the specified Interfile
syntax but these are starting with the characters ‘;#’. Thus, when importing these into a
workstation, these tags are ignored because of the ‘;’, i.e. a remark sign in the interfile format.

The collimator database

This file includes data for the most common collimator files. It is based on a 6-character code that
is associated with the dimensions of the collimator. The user can easily modify this file and add
new collimators. If a user lacks some common collimators and wants to include them for other
users, please send the data to Michael so that he can include them in this file for further releases.

The result-file
simind calculates some basic detector parameters by default. These appear on the screen and in
the result file (*.res) together with the input parameters, defined in change. Collimator data are
only written to the result file if a collimator has been included. SPECT input parameters are
written only if the SPECT flag is set. Moreover, the scatter fraction and the percentage scatter,
described below, are only written if a phantom has been simulated.

Version 6.2
 The SIMIND Monte Carlo Program Page 44 of 64

The calculated detector parameters are listed below.

Solid Angle This is the calculated solid angle through which the photons are forced.
The angle is normalized to 4π, that is, a value of unity equals an
isotropic emission.

Compiler This shows the compiler that has been used to compile the module for
which the result file has been obtained. This information may be of
value when comparing results from versions of simind for different
operating systems.

Collimator Efficiency The calculated geometric collimator efficiency of the defined collimator.
This parameter is defined as the ratio of the number of photons passing
through the collimator to the number of emitted photons.

Compton area The integral of simulated events from the first channel to the lower
energy window channel.

Photo area The integral of simulated events between the lower energy window
channel and the upper energy window channel.

Pileup area The integral of simulated events from the upper energy window channel
to the highest channel in the energy spectra.

Scatter/Primary This is the ratio between the events from photons scattered in the
phantom to events from photons that have penetrated the phantom
without interaction.

Scatter/Total This is the ratio between the events from photons scattered in the
phantom to all events from photons.

ScatterOrder 1,2,3.. This is the percentage of the scatter events originated from photons that
have been scattered once in the phantom.

DetectorHits This is the number of photons that have reached the scintillation
crystal.

DetectorHits/CPUsec Just the number of detector hits per simulation time. Can be useful to
compare the performance for different computers.

MaxValue Spectra This shows the maximum value in the calculated energy spectra. Note
that this value may be less than unity because the spectra reflect the
accumulated weights of the interacted photons.

CountRate Spectrum This shows the calculated count rate for the whole energy spectra. This
value is based on the activity, given by Index-25 in change.

CountRate Window This shows the calculated count rate for the defined energy window.

Efficiency Window This is the fraction of the number of photons detected within the energy
window to the number of photons impinging on the crystal.

Efficiency Spectrum This is the fraction of the number of photons detected in the whole
energy spectrum and the number of photons impinging on the crystal.

Sensitivity Cps/MBq The count rate in the energy window divided by the activity defined for
the simulation (Index-25 in change) [cps/MBq].

Sensitivity Cpm/uCi The count rate in the energy window divided by the activity defined for
the simulation (Index-25 in change) [cpm/uCi].

Version 6.2
 The SIMIND Monte Carlo Program Page 45 of 64

Comments An optional comment line defined in change (Index-9 in the main

menu).

Message Sometimes, a routine can give a message to the result file concerning
something that may be serious but not fatal for the simulation. The text
is then displayed after the "Message" text.

Additional information dependent on selection in change

1. If image-based phantoms with values -2 to -6 are used (based on the *.zub files), then the
volumes, activity, number of pixels, and activity concentrations are given in the result file for
each organ/structure.
2. If the multiple-source routine is used in the simulation, then information about the activity
in each of the spheres is given in the res file.

File-extensions and file-formats

The various file types encountered in simind are listed in the following sections. On UNIX, Linux,
or MacOS systems, all file names are in lower case!

Input files used by simind

*.cr3 Represents table listings of cross-sectional data for various materials. These files
are written in ASCII and have an obscure file format. The first four floating values
in the first record are the density for the material, the highest atomic number in
the compound, and the corresponding K-alpha energy and fluorescence yield for
that material. The next value is an integer and is the number of cross-section
values in the four columns below. The four additional floating values are reserved
for future development of pile-up simulations. The different cross sections for photo
absorption, Compton, coherent, and pair production are then tabulated in column
form. Finally, atomic form factors and the incoherent scatter functions are
tabulated in a special compressed format.
*.dmi Represents a binary density map file used to generate a non-homogeneous
phantom. These files should be in a 16-bit integer format. The density of the
phantom should be a factor of 1000, i.e. for water (density = 1.0 g/cm3), a pixel value
of 1000 should be used.
*.smc Represents a listing of the input data created by the program change. This file is
an ASCII file that contains the information defined in the change program. It
has a defined format and should not be changed.
*.win File that includes the window settings used by the scattwin program. Each row is
one window with lower and upper energy thresholds in keV. The file format is in
ASCII.
*.inp Represents a file that stores input data used by certain score, source, and isotop
routines. The format may depend on the routine but it is suggested that the proper
extension be used for clarity.
*.smi Represents a binary source distribution. These files are similar in format to the
files for the density maps (*.dmi) but represent a source distribution instead of an
arbitrary density distribution. These files should be in an integer 16-bit format.

Output files created by simind

*.a00 Represents the results of simulated SPECT projections. This binary file consists of
projection data in 32-bit float format.
*.bim Represents a binary matrix image created by the output of simind. This file
contains 32-bit float binary values.

Version 6.2
 The SIMIND Monte Carlo Program Page 46 of 64

*.bis Represents a binary energy spectra image created by the output of simind. This
file contains 32-bit float binary file.
*.res Represents the results of the basic detector parameters calculated from the
simulation. It also contains input parameters and statistics of the simulation.
.him, *.h00 ASCII file that includes header information for the Interfile v3.3 file format. The
*.hct extension differs depending on the image data. ‘him’ is a planar image with the
extension ‘bim’, ‘h00’ is the header file for SPECT projections with the extension
‘a00’, and ‘hct’ is the header file for the aligned density/attenuation file useful for
post processing stages, such as attenuation correction.
*.cor A file that contains the used centre-to-collimator distance for each projection as
calculated from the voxel-based phantom. This file will not be created when using
analytical phantoms (Index-14 > 0).
*.csv A comma-separated file useful for importing to spreadsheets such as Excel. Created
by Index-84.
*.lmf Listmode files that include details about each event. The format may change
dependent on the choice. Please refer to the description of the listmode scoring
routine.

Information of available voxel-based phantoms

The following phantoms are available as binary images in smc_dir.

vox_man1.dat 128×128×243 Original Zubal torso coded phantom with no

integer, 8-bit arms and legs. Note: This is used when then
change Index-14,15 = -2
vox_man3.dat 192×96×498 Modified Zubal phantom with added legs and
integer, 8-bit arms, which is straightened out and aligned
along the body to be useful for whole-body
imaging. Note: This is used when change Index-
14,15 = -4.
vox_brn.dat 256×256 Original Zubal brain phantom file. Note: This is
integer, 16-bit used when change Index-14,15 = -3
alders.dmi 64×64×60 A CT scan of an Alderson phantom. The density
integer, 16-bit is scaled by a factor of 1000.
rod_cyl.smi 512×512×1 Cylindrical CT-based hot-source map of a
integer, 16-bit Quality Control pie phantom.
rod_ell.smi 512×512×1 Elliptical CT-based hot-source map of a Quality
integer, 16-bit Control pie phantom.
rod_ellh.smi 256×256×1 Elliptical CT-based hot-source map of a Quality
integer, 16-bit Control pie phantom.
rod_ellc.smi 256×256×1 Elliptical CT-based cold-source map of a Quality
integer, 16-bit Control pie phantom.

The Zubal phantom

The phantom used in this site has been developed by George Zubal, Ph.D., Dept. of Diagnostic
Radiology, Yale School of Medicine, 333 Cedar Street, New Haven, CT 06510. The phantoms
included are located in the smc_dir directory and have the names vox_man.dat,vox_brn.dat
and vox_man3.dat. The main references are (11-13).

The NCAT/XCAT phantom

This phantom is provided by Dr Paul Segars from Duke University. The phantom is actually a
software package that creates attenuation and activity maps from a user-specified input file, which
are is used as input in simind. Because this family of phantoms have proven to be very useful, the

Version 6.2
 The SIMIND Monte Carlo Program Page 47 of 64

simind program has routines that makes it easy to input such data. Additional information about
the phantom can be found in references (14-16).

TUTORIALS
Here are some examples to get started. These examples assume that you have a display program
to show images represented by float values (32-bit float). ImageJ is one example of such a program
that handles short float images. If, for some reason, you do not have the tutorials, then you can
download the zip file from the link 'Files' to the left.

1. Point source simulation in a cylinder

This tutorial shows how to simulate a simple point source in a cylindrical phantom. The
assumption for the camera is as follows. The corresponding index numbers in the change
program are also given

Collimator: LEGP
Energy resolution: 10.0% at 140 keV (Index-22)
Phantom radius: 11 cm (Index-6,7)
Phantom length: 20 cm (Index-5)
Height from phantom to camera: 5 cm (Index-12)
Energy window: 20% (Index-20,21)
Pixel size 0.54 cm (64x64) (Index-27)
Crystal thickness: 0.952 (40x54 cm2) (Index-9,8,10)
Photon energy 140 keV (Index-1)

denotes the prompter sign in the command window

Run 1: Perform a simple run by the command

simind point point1
The output from the program is now in the files point1.res, point1.bis, and point1.bim. Print
out point1.res and identify the different results from the description in the on-line manual. To
get a readable version of the spectra, perform the command
bis point1
which now produces a new file point1.prn. This is a two-column text file of the energy spectra
which can be imported to Excel or plotted using, e.g. the freeware gnuplot program. The file
point1.bim is a floating point 64×64 matrix that can be briefly viewed by the command
bim point1
To convert to an integer image for display using, e.g. the ImageJ program, give the command
bim point1/oi/nm:1000
A new point1 file with the extension *.int is now created and the maximum pixel value is
normalized to 1000. You can now display this image in a display program
Run 2: Change the point source location by changing the z-value (Index-18). This can be done
conveniently by using switches, e.g.
simind point point1/sc:10/18:4
simind point point2/sc:10/18:2
simind point point3/sc:10/18:-2
simind point point4/sc:10/18:-4
Note that the shift is relative origin so the actual depth is Index-7 – Index-18, or for the example
above, the depth is 7,9,13,15 cm. Plot the spectra for the different runs and also check from the
result files how the sensitivity (cps/MBq) and the scatter-to-total fraction vary with the depth.
Try also to turn off the phantom by adding the switch /fa:11 (Flag-11 is set to FALSE). What
happens to the spectrum, sensitivity, and scatter-to-total fraction?

Version 6.2
 The SIMIND Monte Carlo Program Page 48 of 64

Run 3: See how the shape of an energy spectrum changes by performing the following
simulation:
simind point point5/sc:10/fa:11/fa:12
simind point point6/sc:10/tr:11/fa:12
simind point point7/sc:10/tr:11/fa:12/tr:10
simind point point8/sc:10/tr:11/fa:12/tr:10/tr:7
simind point point9/sc:10/tr:11/tr:12/tr:10/tr:7

The simulation will be as follow

• No phantom, no cover, no backscatter, and no energy resolution (cover and backscatter
switches not needed because these are set as FALSE in the point.smc inputfile)
• Phantom added
• Phantom and cover added
• Phantom, cover, and backscatter added
• Phantom, cover, and backscatter added with energy resolution
Perform the simulations for a couple of photon energies up to 700 keV and determine the energy
range in which the cover and the backscatter compartments become important. The energy
switch is /01: so add, e.g. /01:300 to the commands above for a simulation of 300 keV photons.

2. Jaszczak SPECT simulation if six spheres in a cylinder

The aim of this tutorial is to show how a SPECT simulation can be made using a source routine
jaszak.fso that simulates spheres of different sizes and locations inside a phantom. The routine is
invoked by setting Index-15 = 7. The source routine reads data from jaszak.inp, in which each row
defines a sphere. Run the command
simind jaszak jaszak1
The output files are, in addition to the res file, a file called jaszak1.a00, which contains 64 float
64×64 projections. Moreover, the file jaszak1.h00 is stored, which is an Interfile header file. This
file may be of interest to use. If integer maps are needed for display purposes, one can use the
following command to obtain such a file.
bim jaszak1.a00/oi/nm:100/loop:64
The shape, location, and activity of the spheres are controlled by the file jaszak.inp. Each row
represents one sphere. The first three define the dimension and shape, the next three are the
position relative to the origin, and the last value is a relative activity concentration in the sphere.
If the input file has a different name, simind looks for this name.inp file and if this is not found,
it looks for the default jaszak.inp file. Try and add or change the sphere definitions so that you
become familiar with how they need to be set up. Then, try the following simulation.
simind jaszak jaszak2/co
to make a cold-spot simulation where the activity in the spheres is disregarded and assumed to
be non-active while the remainder of the phantom is assumed to be uniformly filled with activity.
The third case is
simind jaszak jaszak3/bg:0.5
where the activity in the spheres defined by the values in the inp file are used as well as
background activity equal to the value defined by the /bg switch.
Exercises to investigate the SPECT reconstruction:
1. Simulate different numbers of projection angles (Index-29) and different rotation modes
(Index-30).
2. Try also changing the start angle for a certain simulation (Index-41).
3. Change the matrix size for the projection angles (Index-76,77). You need to remember to
change the pixel size to half the value (Index-28).

Version 6.2
 The SIMIND Monte Carlo Program Page 49 of 64

The scatter fraction in the result file and all other detector parameters are obtained by the sum
of all data for all projections simulated. Index-15 in change needs to be 7 in this version in order
to invoke the jaszak.fsc routine.

3. Scattered radiation

Study the contribution from photons, scattered in the phantom, and that have generated events in
a fixed energy window (15% is defined in the input files) as a function of different photon energies
(Index-1 in change). In the res file, you can find values for 'scatter-to-total', ‘scatter-to-primary’
together with the relative contribution of photons with different scatter orders. At which energy is
the scatter-contribution greatest? Think about why this is the case?

simind legp out1/sc:8/15:4/tr:11/01:75

Study the scatter contribution within the energy window as a function of the position of the point
source. Note that the minor axis of the elliptical phantom is 11 cm. The simind command for this
will be

simind legp out1 /sc:8/15:5/18:5/tr:11/84:1/01:140

Index 18, which define the shift in the z-direction relative to origin, should therefore be varied
within the range of [±10.5 cm] and where a negative value makes the point-source move towards
the camera. The switch /TR:11 assure that photon interactions in the phantom are simulated. /SC:8
defines the maximum order of photon scatterings that is allowed. The switch /01: defines the
photon energy 140 keV and /15:5 define a point source of radius [0,0,0]. Study the variation in the
scatter-tot-total fraction, as can be obtained from the result file, and plot these values as a function
of the source depth.

Plot the energy spectrum for some different depths from the binary file out1.bis. Because we run
the scattwin routine, energy spectrum is stored in a different way as compared to the default
simulation mode. Spectra is here stored in such a way that the first 512 floating point values
represent the total energy spectrum, that is all events. The next 512 events represent the total
events from scattered photons, and the consecutive next 512 values are all events from photons
scattered 1 order, 2 order, 3 orders, etc. You can use the bis program to extract the data into text
files *.prn that is suitable to plot, for example, with gnuplot.

bis out1 total/sp:1/nc:512 # total spectrum

bis out1 total/sp:2/nc:512 # total primary spectrum
bis out1 total/sp:3/nc:512 # total scatter spectrum
bis out1 total/sp:4/nc:512 # spectrum of 1st order scatter events
bis out1 total/sp:5/nc:512 # spectrum of 2nd order scatter events
bis out1 total/sp:6/nc:512 # spectrum of 3rd order scatter events
..
..

Study the energy spectrums appearance for different photon energies (75, 140, 364 and 511 keV
for a point in the middle of the phantom with the command

simind legp out/sc:8/15:5/tr:11/84:1/01:75

Plot both primary spectra and spread in the same chart. Why do the primary events occur in
Compton distribution at high energies? Are not primary photons just that - primary?

4. Simulation of spatial resolution

Simulate the spatial resolution for a bar phantom and 140 keV photons as a function of distance
and collimators. Display the bim files using some display program and evaluate the pie slice,

Version 6.2
 The SIMIND Monte Carlo Program Page 50 of 64

which can be resolved for different distances and collimators. Run the simulations with the
following command.

simind lehr test1/px:0.15/th:0.1/nn:1000/dir:3/12:30

where /12: indicates the distance from the collimator to the source. /PX is a switch that defines the
pixel size of the image of the bar phantom (rod_ellh.smi), which is used for the simulation. Note
that this pixel size it not the same as the pixel size of the image that simind calculates. /TH is the
thickness of the rods, /DIR denotes the direction of the source, and /NN is a multiplier (refer to the
page on "Image-Based Source" for additional details about how to simulate 3D source distribution
using images. Repeat the simulation for LEHR, MEGP, and HEGP. You should get something like
this. The bim images are stored as float 128×128 in the file test1.bim

5. Bone-scan simulation using an MDP-like radiopharmaceutical (SPECT)

This tutorial aims to be familiar with the setup of the voxel-based Zubal phantom and how the
input file to simind for that phantom is defined. The Zubal phantom consists of a set of 8-bit-
coded images where each voxel has a unique value that can be related to the organ or structure
that the voxel belongs to. These coded images can then be used to define either density maps or
activity maps provided that the user has appropriate density and activity values.
simind is written such that it is easy to use this phantom. The program reads in the 8-bit-coded
images and from a special user-written table, creates activity maps. A template of such a file is
located in the simind/smc_dir directory and is called vox_man.zub. In this tutorial, a special
file called mdp_ect.zub is used. The file is of ASCII type and consists of four columns
1. The name of the organ,
2. The unique code of that organ,
3. The density value, expressed as g/cm3*1000, and
4. A value representing the activity concentration (MBq/cc).
The last value is a relative value but is regarded as 'activity per voxel'. It is important that the
file is filled with data so that the positions of the values remain the same.
simind assumes a file with the same name as the input file but with the extension *.zub. If this
file cannot be found, then a standard file 'smap.zub' is used. If there is no such file, the program
is terminated. Index-14 and/or 15 should be set to -2 for the torso phantom simulation and -3 for
the whole-body phantom. The size of the phantom in the y- and z-directions is determined by the

Version 6.2
 The SIMIND Monte Carlo Program Page 51 of 64

phantom pixel size (Index-31) and the size in the x-direction is determined by the source (and
phantom) length, i.e. Index-2 and -5.
The following example performs a simulation of an MDP-99m-Tc SPECT with and without
interactions in the phantom.
simind mdp_ect mdp_ect1/tr:11
simind mdp_ect mdp_ect2/fa:11
Here, the file mdp_ect.zub is used as input to define the activity distribution.
Tasks: (here, the SPECT option can be turned off by the switch /FA:5)
1. Change the phantom pixel size (Index-31) to different values and investigate the output in
the image file.
2. Change Index-2 and -5 and view the output of the images file. Note that if you have selected
the phantom to be true, then the sizes in Index-2 and -5 should be equal. If the activity
distribution is outside the phantom (Index-2 > Index-5), then an error occurs.
3. Change the image pixel size (Index-28) to different values and investigate the output in the
file mdp_ect1.bim
4. What is the difference between 1 and 2? Hint: investigate the scatter to total ratio values.
The step size (Index-34) determines how the path length of the photon in the voxel phantom should
be incremented. The size relates to the voxel size in that a value of the step size that is too large
can result in missing some voxels. A value that is too small may result in incremental changes to
get the photon leaving the voxel. Try to change the step size and see how this changes (a) the
speed of the calculation, (b) the accuracy of the calculated results (e.g. scatter-to-total ratio, and (c)
the quality of the image.

6. Bone-scan simulation using an MDP-like radiopharmaceutical (Whole body)

This tutorial shows the possibility of simulating whole-body studies with the extended Zubal
phantom (called vox_man3.dat), which has been developed by Dr Katarina Sjögreen, Lund
University. The major improvement here is that the arms are in a realistic position for WB
investigations.
In simind, there is a special switch called /WB that makes simulation of WB studies easier. It
makes the simulation a SPECT study with two projections, and the posterior projection is
mirrored. The matrix size is by default 192×512, but if a value /WB:2 is given, then the pixel size
will be half and the matrix size will be 384×1024. The interfile will also have a slightly different
layout and the ability to read this has been tested on an ADAC Pegasys workstation
Simulation
simind mdp_wb mdp_wb1/wb/fa:11 # without phantom interactions
simind mdp_wb mdp_wb2/wb/tr:11 # with phantom interactions
simind mdp_wb mdp_wb3/wb/tr:11/if # add lesions
Note that for the third simulation, simind is looking for a file named mdp_wb.inp. If found, it
reads from the file and replaces those pixel values corresponding to the locations and sizes given
in the input file with new values. The format is similar to the input file format for the "Multiple
Sphere Routine" with the difference that the size and location are given as pixel numbers instead
of spatial coordinates (x,y,z) and that the activity concentration needs to be an integer.
Tasks: change the WB value from 1 to 2 and verify that you can read the new image.

7. Simulation of photon penetration in an LEGP collimator

Study the effect of septum penetration as a function of photon energy for the energies 75, 140, 245,
364, and 511 keV for a low-energy general-purpose collimator. In the result file, the fractions of
"geometric collimation", "penetration", and "scatter in the collimator" are given.

1) Use the following command

simind legp result1/lf/19:-45/15:4/01:245/53:1

Version 6.2
 The SIMIND Monte Carlo Program Page 52 of 64

where switch /01 is the photon energy in keV and / 53:1 represents a collimator that includes the
interaction used. The switch /19:-45 implies that the acceptance angle for photons emitted towards
the camera is 45 degrees and not determined by the collimator hole dimensions. /LF means linear
sampling of the polar angle and makes the simulation somewhat shorter. /15:4 mean a cylindrical
horizontal source. You should get something like this for an LEGP if you look at the bim files.

The bim images are stored as float 128×128 in the file test1.bim.

2) Plot the values 'geometrical', 'penetration', and 'scatter' from the result file as a function of
photon energy.

3) Repeat the simulations for LEHR and HEGP collimators and study the differences.

simind lehr result2/lf/19:-45/15:4/01:245/53:1

simind hegp result3/lf/19:-45/15:4/01:245/53:1

8. Some examples of castor reconstructions

Perform a simulation according to tutorial 5 to get a SPECT projection file mdp_ect.h00 and
mdp_ect.a00. These files are then the Interfile header (*.h00) and the data file (*.a00). Give the
command

smc2castor mdp_ect.h00 ect/run

which will, if all things are set up properly, reconstruct the data to an Interfile ect_it2.hdr and the
data file. ect_it2.img. The data format for the images will be 32-bit floating as will the data format
from simind. The number of iterations and subsets can be changed by the switches /IT and /SU.

If you want to incorporate attenuation correction of simind generated data, you need to set Flag-
15 in change to TRUE. By doing this, simind will produce a map of density values in a format that
is registered to the pixel size and matrix size of the SPECT projections. These maps *.hct and *.ict
can therefore be used as an attenuation map in the castor reconstruction. However, castor assumes
that the data format is in units of attenuation coefficients (cm-1). To get this from simind, the
default value of density in the simind.ini file needs to be changed. This can be done by adding the
switch /in:x22,5x to the command for the simulation of the projection data. The command for
attenuation is

smc2castor mdp_ect.h00 ect/run/atten

Note that the information about the name of the attenuation is embedded in the mdp_ect.h00
header information. Thus, no information about the attenuation file is needed in the command
line.

If scatter correction also is to be included in the reconstruction, then a file with scatter projections
is needed, which has the same format and is properly scaled. Castor uses this file in the forward
projector as an additive term. The command for scatter and attenuation correction is

smc2castor mdp_ect.h00 scatter_file.h00 ect/run/atten

Version 6.2
 The SIMIND Monte Carlo Program Page 53 of 64

In order to make it easier for a simind user to obtain this file with the requirements, the scattwin
scoring routine has been modified to allow for a direct scaling matching the DW (dual-energy) and
TEW (triple-energy) window scatter correction methods. Please refer to the description of the
parameters ‘dw’ and ‘tew’ in the scattwin routine section.

Smc2Castor program produces some files that may be useful.

1. The projection files needed for the castor reconstruction have the extension *.cdh (header file)
and *.cdf (data file). These are only needed at the reconstruction time and will be safely
removed.
2. In the CASTOR_CONFIG/scanner folder, a file that is needed for the reconstruction is created,
which starts with ‘smc_’. These text files are generated by the smc2castor program based on
the contents in the *.h00 simind projection file. These files can also be removed occasionally.
3. The result of the castor reconstruction is in Interfile format and has the extensions *.hdr and
*.dat. Note that castor adds information about the number of iterations in the file name.
4. The smc2castor program creates a script file starting with the characters ‘castor_’ for the
reconstruction (*.bat for Windows and *.sh for Linux/MacOS). These files are executed when
the /run switch is given. However, these can also be started by a keyboard command. These
are safe to delete occasionally.

FAQ

Installation
Why do I get the error message "simind - Unknown compiler"?
The environmental variable SMC_OS is not set to a proper value. Please read the installation guide
for further help on this issue. This is not an issue from version 4.9.

Phantoms
How do I simulate a map from a patient obtained by CT?
The simulation is based on images, which means that you need to define Index-14 and 15 to -1. You
also need to specify the name of the files in the main menu by the numbers 12,13. simind does not
know about the format except that it has to be an 16-bit integer file for both the phantom and the
source images, so the matrix size is defined by Index-78,79 and 81,82.
The phantom file needs to be scaled to density values * 1000 (voxels with water will have values
equal to 1000). The density file (Option 13 in main menu) needs to have the extension *.dmi and
the source file (option 14) needs the extension *.smi. The number of phantom images is defined by
Index-34 and the start image in the file by Index-33 (usually equal to 1 but you do not need to start
from the beginning of the data set).
To scale the image to real physical dimensions, a pixel size for the phantom image needs to be
defined by Index-31. This scale the phantom in the y- and z-directions. The length (x-direction) of
the phantom in cm (slice thickness) is defined by the half-length of the phantom (Index-5) divided
by the number of images (Index-34). The corresponding pixel size for the activity maps is in the
current version of the program only given by the external switch /PX:
Could you explain how simind decides the #particles/prom? I tried to change the
parameter #26, which is the number of photon history, but it seems to have no effect; the
output file always states a number of 767,555.
When you are using images as a source, Index-26 is not used to maintain a proper sampling of the
desired number of decays described by the images. Instead, the value of Index-26 is replaced by the
sum of pixels in your images (which is equal to 767555). You can increase this number by
multiplying with a factor given by the switch /NN. For example, /NN = 10 gives a simulation by 7
675 550 histories.
Could you explain the geometry in more detail?

Version 6.2
The SIMIND Monte Carlo Program Page 54 of 64

If you picture a real gamma camera patient measurement, the patient axis from the feet to the head
is aligned to the x-axis and the direction from the middle of the patient to the feet corresponds to
negative x-values (simind phantoms are always centred in the origin of the coordinate system).
Please review the image in section ‘Change’. If you view the patient from the feet, a positive simind
y-value goes from the midline to the right and a negative y-value to the left. A positive z-value goes
towards the camera (assuming a camera angle of 0 that is the anterior position) and a negative z-
value goes away from the camera. When using standard phantoms (Index-14,15 is 1,2,3,4,5..), the
source defined by Index-2,3,4 is centred around the origin. You can move the source around by Index-
16, -17, and -18, and this is useful when simulating, e.g. a point source at different depths by
changing Index-18.

When using voxel-based phantoms from images, these are only relative number distributions, e.g.
64×64×64 or 128×128×128. Therefore, you have to first scale to a physical dimension. The scaling is
done by two values, namely the pixel size (y- and z-directions) if transaxial images are assumed, and
the slice thickness (x-direction). For the phantom, Index-31 defines the pixel size for the phantom
images assuming square pixels and Index-5 defines the total axial length of the phantom (= number
of CT-slices (Index-34) times the slice thickness). Thus, simind calculates internally the slice
thickness. Index-6,7 is not in use for image-based phantoms if orientation Index-32 = 0. For known
phantoms, such as the Zubal phantoms (Index-14 = -2,-3,-4) and the NCAT phantoms, the source
maps correspond to the phantom maps; the pixel size for the source maps will be the same, as will
the slice thickness. In these cases, to be safe, you can also set Index-2 equal to the value of Index-5.

Output
What about pixel values in bim file. Are they fractions of one (unity) or what is this
value? How can one obtain counts in each pixel?
The pixel values are normalized in such a way that is the probability for counts in that pixel per
defined activity (Index-25) in the source and for an acquisition time of 1 s. The activity can be
defined in a cylinder (Index-14 = 4) or spread out according to a distribution provided by a 3D image
set. To get realistic counts, one needs to scale the image after a simulation by the appropriate
activity (if not done already by Index-25) and the appropriate time per projection. If this is done for
a long simulation, one can finally add Poisson noise to the data to get a realistic image. Not that
the magnitude of the pixels remains the same even if additional photons are simulated by
increasing Index-26 or using the nn: switch. The statistics will of course be better if more photons
are simulated.

Runtime
Is there a possibility to continue a simulation after a power failure?
The only option one has to continue a simulation that has been terminated is to use the switch /pr:n
to start a simulation at the particular projection number ‘n’. However, the data in the RES and the
energy spectrum will be based on this point and forward, so if you want to simulate an energy
spectrum that corresponds to the average of all projections, you will use this switch for correction.
You must give the command exactly as the initial command and only add the /PR switch

Simulations
How do I create and simulate a new isotope?
1. Create a file in the smc_dir folder, which, e.g. will be called p32.isd with your spectra. The
format is a simple text with energy in keV and the abundance. Look at, e.g. i131.isd in the
smc_dir to understand the format. It is important that you follow this format, i.e. the number
appears in the right place and columns.
2. If you specify a negative energy Index-01 in change, then this indicates the middle energy of
the energy window as before but the negative sign tells simind to use an energy file instead of
the actual number in Index-01.
3. At the command line, when running simind, add the switch /fi:p32 to tell which isotope file
you shall use. Alternatively, copy p32.isd to the default name spectra.isd in the current folder
where you perform your simulations.

Version 6.2
The SIMIND Monte Carlo Program Page 55 of 64

INSTALLATION PROCEDURE
Linux/MacOS
1 Create a directory $HOME/simind
1. Download the distribution you want to install into the $HOME directory and unpack it using
the following commands (examples are for Linux 32 compilation):

gunzip smc_intel_32_vxx.tar.gz
tar -xvf smc_intel_32_vxx.tar

where xx is the version number.

2. Add the following lines to the .cshrc file. This is only needed the first time.

setenv SMC_DIR $HOME/simind/smc_dir (tcsh,csh)

set SMC_DIR=$HOME/simind/smc_dir (bash,sh)

3. Add $HOME/simind to the $PATH search.

4. Execute the command source .cshrc to make the changes effective.

In addition to the simind code, you might need Michael's help to run the following programs.
bis
bim

Both bim and bis are described in the manual but discrepancies can occur. Included is a data file
'bench.smc'. Read in this file by the command
change bench
Simulation
simind bench

Windows
1. Create a directory c:\simind
2. Unpack smc_win_intel_zz_vxx.zip (zz denotes the version, i.e. either 32- or 64-bit, and xx is
the version number) in c:\simind! Keep folder information while unzipping and be sure that
a directory c:\simind\smc_dir with some files of extension *.cr3 have been created.
3. In the System/Advanced/Environmental Variables/User Variables/ window add (by clicking
"new") the following environmental system variables by filling in proper fields:

Variable SMC_DIR set to value C:\simind\smc_dir\

4. Add the path C:\simind to the environmental variable PATH in the "System Variables".
5. Restart the command line window.

The “Castor” programs

On the simind web-page, there is a link to a zip file containing a distribution for the castor software
version 1.0 with pre-build executables for Linux, MacOS and Windows (32 and 64-bit). This zip-
file can be seen as a starting point to use the castor reconstruction. It is, however, strongly
recommended that users of this program register themseft to the community list at the
www.castor-project.org in order to receive information about future updates and to get other
important information. On the page, there is also information about to configure and compile the
builds to particular preferences since it is not the intention that this website should be regarded
as a mirror-site.

Version 6.2
The SIMIND Monte Carlo Program Page 56 of 64

To get started and use the smc2castor program for reconstruction of simind-genereated SPECT
projections then following installation/preparation is needed.
1. Download the zip file castor.zip from the web-page.
2. Extract the whole file structure to a convenient place on your harddrive.
3. Define the environmental variable CASTOR_CONFIG to the absolute path of the config
folder within the castor directory structure.
4. Add the relevant bin folder to you PATH environmental variable (that is, either bin_osx,
bin_linux, bin_win32 or bin_win64).

Version 6.2
The SIMIND Monte Carlo Program Page 57 of 64

VERSION HISTORY
V 6.2
• An aligned roi file can now be saved. This ROIs are created from the image-based phantom
and resampled to the image resolution for the simulated projection. The purpose is to have
available the geometric ROIs to be used, for example, to evaluate partial volume effects. Since
this is not predicted to be used often the creation is initated from a parameter in the simind.ini
file. However, since each value in this file can be changed by the switch /in the invocation of
this option can be made by the switch /in:x54,1x. A 2-byte integer file with the source name
base name and the extension *.iro will be created. If the switch is defined as /in:x54,999x then
the simulation is terminated after the *.iro creation. If code-based images are used, then it is
the code that is stored. If it is a *.smi file (index15=-1) then it will be the values in that file
that are used. This means that if a user för this particular file type wants to separate sources,
they need to have different values (that is for the kidneys specify a source values unique to
each kidney. This may of course be a limitation but is the only way to get segmentation to work
on this type of source maps.
• Several small bugs have been fixed and slight changes in the res file layout have been made..

V 6.1
• Bug fixes and a new print-out format to the screen (6.1.1)
• A value of the keV/chanel is not more written to the last cell in the energy spectrum (6.1.1).
• The smc2castor program has been written to make it easy to reconstruct simind projection
data using the castor reconstruction software http://castor-project.org/.
• The scattwin routine has been updated to make is easier to simulate data for the dual-energy
window scatter correction methods or the triple-energy window method. This is mainly done
to get data that can be directly used in the castor reconstruction program.
• A new switch is defined /IN that allows the user to change the default value in the simind.ini
file. These default values are now numbered and thus possible to change at runtime.
• The spectrum BIS format for the scattwin routine has been changed. The 1st spectrum in the
bis file is now the total spectrum, the 2nd is the spectrum for all primary photons unscattered
in the phantom, the 3rd is the spectrum for all photons scattered in the phantom and the 4rd,
5th, 6th, etc. spectra are for scatter order 1,2,3,…max scatter order.
• If a bremsstrahlung spectrum is used, the values are not printed anymore in the res file owing
to the length of the spectrum.
• An issue with index 59 för parallel hole collimators (collimator movement) has been fixed.
V 6.0
• CZT detectors are now possible to simulate, and there is a new menu for this in the change
program. The methods have been validated and described in reference (5).
• Dynamic planar imaging is possible to simulate using user-specific TAC files.
• Pinhole imaging has been improved and validated. Results from this validation are published
in reference (17).
• The simulation of the scatter has been corrected because the first-order coherent scatter was
not simulated correctly. The impact of this correction is mainly seen in the lower energy region
of the pulse-height distribution where the coherent scatter becomes important.
• The change program now displays the context of the collimator-specific parameters in the
menu. The menu is change when a different collimator type is change. Thus, the menu for
pinhole collimators is different from that for the parallel-hole ones.
• simind is now supporting MPI parallel computing through the use of software from
https://www.open-mpi.org/. A binary version of simind called simind_mpi is included.
Currently, only Linux and MacOS are supported. Note that this required that you install
openmpi accordning to instructions. No static builds for simind are yet available.

V 5.0c

Version 6.2
The SIMIND Monte Carlo Program Page 58 of 64

• In the listmode, the indicator for a new projection is now defined as a whole listmode event
record where all parameters except for the scatter order are set to the value -99. The previous
version to indicate a new projection angle used only a negative value for the deposit energy,
but in some cases, it was not possible to set the indicator for the new projection because no
energy was deposited.
V 5.0
• The code is now built with 64-bit float variables as default in order to increase the precision.
However, the output of the images *.bim and *.a00 and the energy spectrum *.bis remain
stored as 32-bit float data.
• Simind now uses the RANLUX random number generator. Here, it is possible to configure the
accuracy of the random numbers by a value in the simind.ini file ranging from level 0-4,
where 4 is the most accurate but also the slowest. The default value is set to 2. Additional
information on RANLUX can be found on Google.
• A new set of isotope files is available. The data has been taken from the following web site
http://www.nucleide.org/DDEP_WG/DDEPdata.htm. Owing to the limitation in the cross-
section ranges, the simulated photon energies are restricted to be within the range of 20 keV
to 2500 keV.
• The code has been clean regarding the definition of the COR. This means that for no change,
the values in the simind.ini file and Index-36,37 and 39 should be zero.
• I have encountered some problems in using the 'RR' switch when running on clusters. Identical
projections appeared even if different values for the RR switch were given and I have not found
the reason for this. However, I have developed a work-around for this problem in that the
random number generator is called RR times before starting each history (if the switch is
given). This means that there is no advantage of giving high values for the RR switch as long
as they are not the same.
• The activity values and concentrations for the voxel-based phantom, as defined from the *.zub
file, can now be written to a CSV file.
• There is now a possibility to simulate the shape of the tumours based on the Butterworth filter
equation. The cut-off and order are given in the tumour input file.
• Now, four collimator-dependent variables are defined, which are specific to a particular
collimator.
• The source distribution generated by the jaszak routine can now be changed in their different
directions using the switch /DIR:
• A cone source shape has been added to the "Multiple Sphere" simulation. If the cone number
is negative then the cone direction is reversed.
• A new scoring routine "penetrate" is included, which separates the image into components.
• The "Save Aligned Density Map" (Flag15) now also supports geometrical phantoms (Index-14
= 2-4).
V 4.9
• Four new shapes for the "Multiple Sphere" simulation. A horizontal hexagonal rod shape, a
rectangular rod, and horizontal and vertical cylinder shapes are defined by an additional value
in the input file. If these values are omitted, then the default will be spheres.
• Index-30 now is the rotation angle with the sign defining the direction. However, the old values
0, 1, 2, and 3 that define the rotation mode remain.
• A new scoring routine has been added. It produces a binary list mode file. Additional
information and the format for this is found under "Scoring Routines".
• If there is a file named simind.loc in smc_dir, then this will be used instead for the
simind.ini file. This is to avoid local settings being overwritten each time a new version of
simind.ini installed. However, note that an error will occur if the content of the simind.ini
file have increase due to an update.
• An option for an output to a CSV file has been added by the change Index-85. This makes it
possible to easily import results to, e.g. Excel.
• The default values for the centre have been changed to better reflect the common definitions
in reconstructions. This means that for a 64×64 matrix, the midpoint point x = 0, y = 0 will be
in between pixel 32 and 33 but (0,0) will be scored in pixel 33. The mid-point for a DMI and

Version 6.2
The SIMIND Monte Carlo Program Page 59 of 64

SMI file with 64 images will similarly be between 32 and 33. These default values can be
changed in the smc_dir\simind.ini file.
• Some bugs and error-handling problems have been fixed.

V 4.8
• The "Multiple Sphere" routine can now have any number of spheres, and combinations of
spheres and cylinders can be made in a single simulation.
• Several important bug fixed.
V 4.7
• A MacOS version is now available
• The change program has been modified to include help for each index. When adding a ´?´
character after or before an index number, you get help on this topic. The help text is the same
as in this on-line manual.
• The SMC_OS environmental variable is now obsolete and not used anymore.
• Bug fixes when calling the Multiple Source routine.
• Old Index-39 and -40 have been removed because the need for calculating a pure sinogram for
a specific slice has been regarded as obsolete owing to the availability of large disk space in
modern C. Index-39 has been replaced by a possibility to shift the phantom relative to the x-
direction.
• The file name for the aligned transmission map (Flag-15) is now defined by the base name of
the density map. Previous versions set the name from the base name of the input file.
• The default cross-section table for the collimator is now pb_sb.cr3 (set in
smc_dir\simind.ini). This is because discussions with several collimator vendors have
indicated that antimony is generally added to the lead in order to make the collimator harder
and more resistant to damage. The amount of antimony may vary among vendors but the
percentage amount seems to be of the order of 2%. The Pb_sb cross-sectional table therefore
has 2% antimony. A clean lead cross-section table is pb.cr3.
V 4.6
• The Switch /PR has now been modified such that a positive number indicates the start
projection number (as before) and a negative number indicates the stop projection number.
The order does not matter!
• If you have changed the simind.ini file and want to store this on a special folder to avoid
problems with installing a new version, you can assign the full path and file name to the
environmental variable SMC_INI.
• The program is also compiled with the Intel Fortran Compiler for Windows version 10. The
executables from this compiler are faster than Lahey but I will keep compiling versions for
Lahey.
• The scattwin routine is now included in the main program by setting Index-84 = 1. The
dummy scoring routine used earlier is invoked when Index-84 = 0. The possibility for linking
user-written scoring routines remains the same.
• The two different collimators (the analytical collimator developed by Frey et al. and the
collimator simulating explicit septal penetration) have now also been included in the main
program. Thus, there is now no need for the simindc and scattwinc programs. The default
collimator is the earlier version based on analytical weighting and is used by Index-53 = 0. The
collimator simulating penetration and scatter is used when setting Index-53 = 1. In a later
version of simind, a pinhole collimator will also be included.
• To reduce the number of programs, the former jaszak routine simulating multiple elliptical
sources is now included in the main program and is selected by Index-15 = 6. Additional
information about how to define the sources can be found on the page regarding "Multiple
Spheres".
• A simple myocardial shaped distribution can be defined by selecting Index-15 = 7. Information
about this distribution and how to set it up can be found on the page regarding "Myocardial
source".

Version 6.2
The SIMIND Monte Carlo Program Page 60 of 64

• Index-26 in change is now in units of 1 million histories. This is because the figures in
change will otherwise become quite large. It is rare that one wants to run only 1000 histories
and this can be represented by the value 0.001 in Index-26.
• Feature: A new switch /RR for initiating the random number generator when running on
clusters has been made.
• Minor bug fixes and changes in the contents of the *.res file.
V 4.5
• Feature: Additional information about the registered events penetration events as function of
acceptance angle (Index-19) is provided in the *.res file when running the simindc and
scattwinc programs. This information can make it easier to verify that a proper value of Index-
19 is defined.
• Feature: Simulations of scintillation cameras with COR as a function of the projection angle
can be simulated. For further information, refer to Index-42 in the change program
description. I think this feature needs some more validations with different phantoms, so
please let me know about success or failure when you are using this.
• Feature: If Index-19 is set to 3, then the acceptance angle is calculated based on the detector
size and phantom size and not only as a function of the collimator hole dimensions (as is the
case when Index-19 is set to 2). This ensures that the acceptance angle is not too small. The
used of this value may however not lead to an optimal value regarding the calculation time.
This is only important when running the simindc and scattwinc programs.
• Bug fix: There was some problem with the tutorial files in earlier versions. In order to avoid
conflicts between versions, the tutorial files and phantoms are now included in the main
distribution file.
• Feature: A density value for each tumour in the *.inp can now be defined. Further information
will be found in the "source routine" page.
• Feature: For this version, a Windows Installer script is used to install the program on Windows
computers. The script asks for a location under which the folders simind and smc_dir are
installed. Most likely c:\ is OK. The script adds the environmental variables and adds the
simind directory to the path. At present, you might need to uninstall a previous version of a
"Windows Installer" installed program. This is still under development, so for those who are
happy with the zip/tar files, these files will remain available.
• Feature: A possibility to use a cut-off energy to terminate photon histories with extremely low
energy compared to the energy window has been added and when phantom interactions are
simulated. The cut-off energy value is defined by Index-83 in the change program. Please
refer to the main page for the change program for further information about this feature.
• Feature: A description about the activity and the volumes of each of the organs has been added
to the results file when simulating a Zubal-like phantom (value -2 to -6 for Index-14 and -15 in
the change program). This information is useful when setting up a simulation when the total
activity in the organ rather than activity concentration is known.
• Feature: Sometimes one would like to have a fixed date and time in the interfile header instead
of the actual computation time. Using the Switch /TS (Time shift) supported with a value, one
can define a time shift in hours relative to a fixed initial time (that can be regarded as the
injection time). The value of the switch is in units of hours. The fixed time and date are defined
in the simind.ini file.
• Some more new switches have been added. Please refer to the page on "Running simind" for
a complete description.
• The error messages have been revised.
• Some revisions of this web page.
V 4.4
• Feature: A new collimator routine has been developed, which is included in the pre-compiled
versions. Those programs ending with 'c' have this new collimator. The method is described in
the paper Ljungberg, M., Larsson, A., and Johansson, L. A New Collimator Simulation in
simind based on the Delta-Scattering Technique IEEE Trans Nucl Sci 52:5; 1370-1375,
2005. What is important is that the acceptance angle is sufficient to allow for penetration

Version 6.2
The SIMIND Monte Carlo Program Page 61 of 64

effects. Index-19 defines this acceptance angle in the change program. By giving a negative
number, the acceptance angle is then defined as the magnitude defining the degrees of the
angle. There is no need to provide a value greater than 90 degrees (that is -90). By default, X-
ray generation in the lead is turned off but you can add this by changing the simind.ini file
in the smc_dir directory or add the switch /XR:1 to the command line.
• Feature: You can now specify a Zubal file as the third file name in the command line, which
has have a different name from the input file name and the standard name smap.zub. You can
also specify an ini file as the third and fourth names (the order does not matter). You can also
specify the Zubal file name and the isotope file name using the switch /FI and /FZ, respectively.
For example, simind input/01:- 140/FI:in111 indicate that the in111.isd file should be used.
Note that you may still need to define /01 to a negative number because the actual value of
the number defines the centre energy for the energy window.
• Feature: The density and source map file names can now be set by the switches /FS (source)
and /FD (density). This makes it easier to set up scripts and avoid potential problems because
different phantoms can be simulated using the same SMC input file.
• Feature: You can change the cross-section tables by the switch /X1-/X6 where the X1-X5
refers to the Index-9,10,11,12, and 15 in the main change menu and X6 is the material for
the collimator when using simindc, scattwinc. Do not specify the extension.
• Feature: You can specify a different collimator by the switch /CC. Do NOT add the character
'*' that needs to be present according to how to select a collimator in the change program.
For example, >simind input output/cc:ge-legp gives the collimator ge-legp as listed in the
collim.col file in the smc_dir directory.
V 4.3
• Feature: The tumour definitions have been refined by including two types of tumours based on
the information in the *.inp file. Tumour type 1 has the centre in the pixel defined by value
4,5,6 in the *.inp file. If valued 1,2,3 are equal to zero, then the tumour will be a single pixel.
Tumour two has the centre at the surface of the pixel that is between i,j,x and i-1,j-1,k-1. The
minimum size for this tumour is 8 pixels.
• Bug fixes: The program for Linux did not read integer maps (30 Oct/2005). The inserted voxel
tumours with the *.inp file did not appear symmetrical (26 Oct 2005).
• Feature: The definitions of the densities for the different organs in the Zubal type of phantoms
have been moved from the simind.ini file to the *.zub. This makes it easier for the user to
modify the densities and it does not require a modification of simind.ini.
V 4.2
• Feature: The format for the Zubal file *.zub has been changed in that additional phantoms
(RSD) have their own section after the voxman and voxbrn sections. Before, some of the code
was also used for the RSD phantoms.
• Feature: The parameter "photons/decay" (Index-24) is now automatically updated based on the
probabilities in the isotope files *.isd if this are used.
V 4.1
• Feature: The generation of X-ray photons has been improved and made more accurate. Now,
sampling is made from Kalpha1, Kalpha2, Kbeta1, Kbeta2, and Kbeta3 as has been defined in
the cross-section tables. The cross sections are better described in this manual.
• Feature: Cross-section tables have been modified, so previous *.cr3 files do not work.
• Feature: The simind.ini file has been modified and now densities can be specified for each
of the structures also in the Zubal brain phantom (Index-15 = -3). Before, this was only possible
for the Voxelman phantom. Further, some additional flags have been added to the ini file that
turn off certain effects.
V 4.0

Version 6.2
The SIMIND Monte Carlo Program Page 62 of 64

• Feature: A new sampling technique, called delta-scattering sampling, has been implemented
for the simulation of non-homogeneous voxel phantoms. This method roughly improves the
speed by a factor of 2-3.
• Feature: The sampling of the photon in the phantom has been changed so that it now uses true
photon splitting instead of sampling a particular scatter order. This means that in each
interaction point and from the decay point, a virtual photon is sent to the detector. The number
of photons actually simulated will therefore be the number of photon histories times the scatter
order minus some photons that escape the phantom. It may seem that the simulation time has
increased but more photons are simulated and the efficiency therefore is increased.
• Feature: The form factors and scattering functions for bounded electrons. These data are now
taken from the EPDL97 documentation. A new cross-section file format has been constructed
with the new extension *.cr3. The energy grid for the cross sections is now 0.5 keV instead of
1.0 keV. The data ranges from 1 keV to 1 MeV.
• Several new materials have been added. A user cannot create his own table because the
generation is based on an IDL program. However, a request to me for a new compound or
materials can be made by e-mail.

V 3.9 (including some of the version 3.8 changes that were not documented)
• The number of detector hits per second has been added to the result file as a rough indicator
of the efficiency. Note that it is not the true CPU time but rather the live time for the
simulation.
• A bug in the scattwin routine resulted in wrong scatter order values and percentage values in
the result file. This has now been fixed. Further, the results by adding data from two separate
energy windows were different when compared with the situation where the two windows are
added within the simulation. This should of course be the same .
• The presentation of the time for a simulation has been improved.
• The jaszak source routine has been updated so it now allows cylinders as the source shape in
addition to spheres. Note that the direction of the cylinders is along the x-axis.
• Updated the manual and especially a) the description of available switches for the simind
program b) the features for the scattwin routine
• The lesion code in the Zubal phantom vox_man.dat (located in the smc_dir) has been removed
and replace by the liver code. This is because lesions are handled separately in simind.
• The order in the main menu was not consistent between the Windows version and change
and the UNIX version.
V 3.7
• A Windows version of change was written because on the Windows 2000/ME system, the
ansi.sys drivers have been removed.
• The location of the smc_dir directory is now handled by environmental variables both in
Windows and in UNIX.
• The NCAT file binary format has been included as a phantom number. This is useful because
the image direction is opposite to the MCAT phantom.
V 3.6
• The scattwin scoring routine now calculates the 'air' images at the same time as the images
with scatter and attenuation. This means that only one simulation is necessary.
• Implementation on the MCAT3 file format and naming convention.
V 3.5
• simind has been re-written so it now allows for parallel computing using the Message Parsing
Instruction (MPI) standard. The calculations have been validated on a IBM SP2 computer
machine at the University of Michigan Computing Centre
V 3.2

Version 6.2
 The SIMIND Monte Carlo Program Page 63 of 64

• Densities can now be individually assigned to each organ in the Zubal phantom by editing the
simind.ini. Moreover, a flag control of the earlier average densities for bone, soft tissue, and
lungs shall be used.
V 3.1
• The weight is now multiplied with the Activity*PhotonsPerDecay and divided by the number
of simulated photons. The total count in the projections and the spectra are therefore
corresponding to a measurement of 1 s for the defined activity for the current radionuclide.
• Minor changes in the output file. One of these is that you get the command line, which can be
useful, if you have to overwrite the values in the SMC file using a switch.

V 3.0
• A new format for the cross-section files that replace the interpolation step in the previous
version.
• The implementation of direct access to MCAT phantom and source files.
• Implementation of an input file in smap.fso that generates tumours in a template source map
file.
• External scattwin.fsc scoring routine that makes it easy to simulate multiple energy windows.
• Added shift in the y-direction for the camera relative to the phantom
V2.x
• The major change is the movement to Fortran 90, which allows dynamically allocated arrays.
This greatly increases the flexibility of the program because the matrix size can now be selected
from a menu in change.
• An INI file has been implemented, which stores some of the default values. You can have local
INI files stored in different directories and have simind read these instead of the default file,
stored in the smc_dir directory.
• A /WB switch has been added, which easily makes an A/P simulation with, e.g. the Zubal
phantom. The output Interfile takes this into account.
• Minor improvements.

REFERENCES

1. Ljungberg M, Strand SE. A Monte Carlo program for the simulation of scintillation camera characteristics.
Comput Methods Programs Biomed. 1989;29(4):257-72.
2. Ljungberg M. The SIMIND Monte Carlo program. In: Ljungberg M, Strand SE, King MA, editors. Monte
Carlo Calculation in Nuclear Medicine: Applications in Diagnostic Imaging - Second edition. Florida: Francis &
Taylor; 1998. p. 111-28.
3. Ljungberg M. The SIMIND Monte Carlo Code. In: Ljungberg M, Strand SE, King MA, editors. Monte
Carlo Calculation in Nuclear Medicine: Applications in Diagnostic Imaging - second edition. Florida: Francis
{&} Taylor; 2012. p. 315-21.
4. Ljungberg M, Larsson A, Johansson L. A new collimator simulation in SIMIND based on the delta-
scattering technique. Ieee T Nucl Sci. 2005;52(5):1370-5.
5. Pretorius PH, Liu C, Fan P, Peterson M, Ljungberg M. Monte Carlo Simulations of the GE Discovery
Alcyone CZT SPECT Systems. Ieee T Nucl Sci. 2015;62(3):832-9.
6. Guerra P, Santos A, Darambara DG. Development of a simplified simulation model for performance
characterization of a pixellated CdZnTe multimodality imaging system. Phys Med Biol. 2008;53:1099-113.
7. Ramo S. Currents induced by electron motion. Proc IRE. 1939;27(9):584-5.
8. Shockley W. Currents to conductors induced by a moving point
charge. J Appl Phys. 1938;9(10):635–6.
9. Chen L, Wei YX. Monte Carlo simulations of the SNM spectra for CZT and NaI spectrometers. Appl Radiat
Isot. 2008;66:1146-50.

Version 6.2
 The SIMIND Monte Carlo Program Page 64 of 64

10. Gros d’Aillon E, Tabary J, Glière A, Verger L. Charge sharing on monolithic CdZnTe gamma-ray detectors:
A simulation study. Nuclear Instruments and Methods in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment. 2006;563(1):124-7.
11. Zubal IG, Harrell CR. Voxel Based Monte Carlo Calculations of Nuclear Medicine Images and Applied
Variance Reduction Techniques. Image and Vision Computing. 1993;10:342-8.
12. Zubal IG, Harrell CR. Computerized three-dimensional segmented human anatomy. Med Phys.
1994;21:299-302.
13. Zubal IG, Harrell CR, Esser PD. Monte Carlo determination of emerging energy spectra for diagnostically
realistic radiopharmaceutical distribution. NuclInstMethinPhysRes. 1990;A299:544-7.
14. Segars WP, Lalush DS, Tsui BMW. A Realistic Spline-Based Dynamic Heart Phantom. IEEE Trans Nucl
Sci. 1999;46:503-6.
15. Segars WP, Lalush DS, Tsui BMW. Modeling respiratory mechanics in the MCAT and spline-based MCAT
phantoms. IEEE Trans Nucl Sci. 2001;48:1192-5.
16. Segars WP. Development of a new dynamic NURBS-based cardiac-torso (NCAT) phantom. Universiy of
North Carolina; 2001.
17. Peterson M, Strand S-E, Ljungberg M. Using the Alloy Rose’s Metal as Pinhole Collimator Material in
Preclinical Small Animal Imaging: A Monte Carlo Evaluation. Medical Physics. 2015;42:1698-709.

Version 6.2