B NM App Suit Ref Manual ENG

English
Reference Manual
NM Application Suite on
IntelliSpace Portal
Version 5.0
PHI
NM Application Suite on
IntelliSpace Portal
Reference Manual
Version 5.0
English
4598 001 91041 Rev A
P h i l i p s H e a l t h c a re
PAI
Warranty and Copyright Statement
Philips Healthcare has taken care to ensure the accuracy of this document. However, Philips Healthcare assumes no liability for errors or
omissions and reserves the right to make changes without further notice to any products herein to improve reliability, function, or
design. Philips Healthcare provides this guide without warranty of any kind, either implied or expressed, including, but not limited to,
the implied warranties of merchantability and fitness for a particular purpose. Philips Healthcare may make improvements or changes in
the product(s) and/or program(s) described in this manual at any time.
This document contains proprietary information which is protected by copyright. All rights are reserved. No part of this publication
may be reproduced, transmitted, transcribed, stored in a retrieval system or translated into any human or computer language in any
form by any means without the consent of the copyright holder. Unauthorized copying of this publication may not only infringe
copyright but may also reduce the ability of Philips Healthcare to provide accurate and up-to-date information to users and operators
alike.
Philips Healthcare reserves the right to revise this publication and to make changes in content from time to time without obligation on
the part of Philips Healthcare to provide notification of such revision or change.
Trademarks
ACCESSTM, ADAC®, ALLEGRO®, ARGUS®, AtlasTM, AutoQUANT®, AutoSPECT®, AutoSPECT®Plus, CardiaQ®, CardioMD®,
CardioTM, CardioTM 60, CardioTracTM, CCTTM, ColliMATETM, CPET®, ENsphere®, EPICTM, ExSPECTTM, EZXTM, FlexLOGICTM,
ForteTM, GEMINITM, GENESYS®, GlobalQ®, InStill®, InteLOGICTM, JETSphereTM, JETStream®, MCD/ACTM, MidasTM,
MOSAICTM, P3IMRT®, P3MD®, PegasysTM, PINNACLE3®, PIXELAR®, PrecedenceTM, SENTRYTM, ShadowTM, SKYLight®,
SKYTableTM, SMARTSIM®, SolusTM, TeleLOGICTM, TranscamTM, VantageTM, VersaTableTM, VertexTM, WebViewTM, and X-ACTTM
are trademarks or registered trademarks of Philips Healthcare.
Adobe, the Adobe logo, Acrobat, the Acrobat logo, and PostScript are trademarks of Adobe Systems Incorporated or its subsidiaries and
may be registered in certain jurisdictions.
Sun, SunView, NFS, OpenWindows, Solaris, and SPARCstation are trademarks of Sun Microsystems, Inc. SPARC is a registered
trademark of SPARC International, Inc. SPARCstation is a trademark of SPARC International, Inc. licensed exclusively to Sun
Microsystems, Inc.
UNIX and OPEN LOOK are registered trademarks of UNIX System Laboratories, Inc.
X Window System is a trademark of the Massachusetts Institute of Technology.
Microsoft and Windows are either registered trademarks or trademarks of Microsoft Corporation in the United States and/or other
countries.
Symantec and Symantec pcAnywhere are U.S. registered trademarks of Symantec Corporation. Symantec AntiVirus and Symantec
WinFax PRO are trademarks of Symantec Corporation.
Other brand or product names are trademarks or registered trademarks of their respective holders.
Prescription Device Statement

Caution: Federal law restricts this device to sale by or on the order of a physician (or properly licensed practitioner).
Disclaimer
Neither Philips Healthcare, its parent, nor any of its worldwide affiliates shall be liable or obligated in any manner in respect of bodily
injury and/or property damage from the use of the system/software if such is not in strict compliance with instructions and safety
precautions contained in the relevant operating manuals and in all supplements thereto, in all product labels, and according to all terms
of warranty and sale of the system, or if any change not authorized by Philips Healthcare is made to the software operating the system.
CE Marking
NM Application Suite is CE Marked to the Medical Device Directive 93/42/EEC.
Manufacturer:
Philips Healthcare
Philips Medical Systems Nederland B.V.

Veenpluis 4-6
5684 PC Best
The Netherlands
NM Application Suite Reference Manual
Document number 4598 001 91041 Rev A

© 2012 Koninklijke Philips Electronics N.V.
Philips Medical Systems Nederland B.V., Veenpluis 4-6, 5684 PC Best, The Netherlands
Contents
1 Reconstruction ..................................................................................................... 1
1.1 References for Reconstruction .................................................................................... 1
2 Cardiac ................................................................................................................. 5
2.1 Minimum Requirements for Cardiac ......................................................................... 5
2.1.1 MUGA Processing ....................................................................................... 5
2.1.2 First Pass...................................................................................................... 5
2.1.3 Shunt........................................................................................................... 5
2.2 Acquisition Parameters for Cardiac ............................................................................ 5
2.2.1 First Pass...................................................................................................... 5
2.2.2 Shunt........................................................................................................... 6
2.3 Calculations for Cardiac ............................................................................................ 6
2.3.1 General MUGA Computations ................................................................... 6
2.3.2 General First Pass Computations ................................................................. 7
2.3.3 General Shunt Computations ...................................................................... 7
2.3.4 GBP Multiple Algorithm ............................................................................. 8
2.3.5 MUGA C Algorithm ................................................................................. 12
2.4 References for Cardiac ............................................................................................. 17
3 Whole Body ........................................................................................................ 19

3.1 Minimum Requirements for Whole Body ............................................................... 19
3.1.1 Ileosacrum Ratio ........................................................................................ 19
3.1.2 Three-Phase Analysis ................................................................................. 19
3.1.3 Three Phase + SI Ratio............................................................................... 19
3.2 Calculations for Whole Body and Bone ................................................................... 19
4 Lung .................................................................................................................... 21
4.1 Minimum Requirements for Lung ........................................................................... 21
4.1.1 Lung Ventilation and Perfusion ................................................................. 21
4598 001 91041 Rev A
4.1.2 Washout Analysis....................................................................................... 21

4.1.3 Geometric Oblique Analysis ...................................................................... 21
4.2 Calculations for Lung .............................................................................................. 22
5 Renal ................................................................................................................... 23
5.1 Minimum Requirements for Renal .......................................................................... 23
5.2 Acquisition Parameters for Renal ............................................................................. 24

5.2.1 Simple Renogram ...................................................................................... 24
5.2.2 PrePost Lasix.............................................................................................. 24
5.2.3 Post Renogram Lasix.................................................................................. 24
5.2.4 Patlak......................................................................................................... 24
N M A p p l i c a t i o n S u i t e o n I n t el l i S p a c e Por ta l Version 5.0 iii

5.2.5 Hilson Index ............................................................................................. 25
5.2.6 GFR Gates ................................................................................................ 25
5.2.7 ERPF Schlegel with Void .......................................................................... 25
5.2.8 ERPF Schlegel ........................................................................................... 25
5.2.9 ERPF MAG3 ............................................................................................ 25
5.2.10 DMSA Static Ratio ................................................................................... 25
5.2.11 Deconvolution .......................................................................................... 25
5.2.12 Cortical Analysis........................................................................................ 26
5.2.13 Split T1/2.................................................................................................. 26
5.3 Calculations for Renal ............................................................................................. 26
5.3.1 Generating Multi-phase Dynamic Frames ................................................. 26
5.3.2 Simple Renogram ...................................................................................... 27
5.3.3 Lasix Pre-Post............................................................................................ 27
5.3.4 Post Renogram Lasix ................................................................................. 28
5.3.5 GFR Gates Method................................................................................... 28
5.3.6 ERPF Schlegel Method ............................................................................. 29
5.3.7 ERPF Schlegel with Void .......................................................................... 30
5.3.8 ERPF MAG3 Method............................................................................... 30
5.3.9 Deconvolution Method............................................................................. 31
5.3.10 DMSA....................................................................................................... 31
5.3.11 Hilson Index Method ................................................................................ 32
5.3.12 Patlak Method........................................................................................... 32
5.3.13 Cortical Analysis........................................................................................ 33
6 Endocrine ............................................................................................................ 35
6.1 Minimum Requirements for Endocrine .................................................................. 35
6.1.1 Thyroid ..................................................................................................... 35
6.1.2 Parathyroid Subtraction............................................................................. 35
6.1.3 Parathyroid Dual-Phase............................................................................. 36
6.2 Calculations for Endocrine ...................................................................................... 36
6.2.1 Calculations for Parathyroid ...................................................................... 36
6.2.2 Calculations for Thyroid Uptake............................................................... 36
6.2.3 Thyroid Size Calculation Details ............................................................... 37
6.2.4 Percent Uptake Calculation Details ........................................................... 37
7 Esophagus ........................................................................................................... 39
7.1 Minimum Requirements for Esophagus .................................................................. 39
7.2 Calculations for Esophagus ..................................................................................... 39
Ph ilips Healthcare
8 Gastro Intestinal ................................................................................................. 41

8.1 Minimum Requirements for Gastro Intestinal ......................................................... 41
8.2 Calculations for Gastro Intestinal ............................................................................ 42
9 Hepatobiliary ...................................................................................................... 45
9.1 Minimum Requirements for Hepatobiliary ............................................................. 45
9.1.1 Gallbladder................................................................................................ 45
iv N M A p p l i c a t i o n S u i t e o n I n t el l i S p a c e Por t a l Version 5.0

9.2 Calculations for Gall Bladder ................................................................................... 45
9.2.1 Calculations for GBEF .............................................................................. 45
9.2.2 Calculations for Gallbladder Static Analysis ............................................... 46
9.2.3 Calculation for Static GBEF ...................................................................... 46
10 Astonish .............................................................................................................. 47
11 Filters ................................................................................................................. 49
11.1 Laplacian ................................................................................................................. 49
11.2 Spatial (5x5) ............................................................................................................ 49
11.3 Spacial (3x3) ............................................................................................................ 50
11.4 Spatial Temporal ..................................................................................................... 50
12 Keyboard Shortcuts ............................................................................................ 51

12.1 Keyboard Shortcuts for Applications ........................................................................ 51
12.2 Keyboard Shortcuts for Viewers ............................................................................... 51
4598 001 91041 Rev A
N M A p p l i c a t i o n S u i t e o n I n t el l i S p a c e Por ta l Version 5.0 v

Ph ilips Healthcare
vi N M A p p l i c a t i o n S u i t e o n I n t el l i S p a c e Por t a l Version 5.0

1 Reconstruction
This section lists the references for the Reconstruction section of

AutoSPECT Pro.
1.1 References for Reconstruction

The following references are useful supplements to this manual.
1. Almquist H, Arheden H, Arvidsson AH, Pahlm O, and Palmer J.
Clinical implication of down-scatter in attenuation-corrected
myocardial SPECT. J Nucl Cardiol 1999; 6:406.
2. Bacharach SL and Buvat I. Attenuation correction in cardiac positron
emission tomography and single-photon emission computed
tomography. J Nucl Cardiol 1995; 2:246.
3. Baron JM and Chouraqui P. Myocardial single-photon emission
computed tomographic quality assurance. J Nucl Cardiol 1996;
3:157.
4. Bartram P, Toft J, Hanel B, Ali S, Gustafsson F, Mortensen J, and
Hesse B. False-positive defects in technetium-99m sestamibi
myocardial single-photon emission tomography in healthy athletes
with left ventricular hypertrophy. Eur J Nucl Med 1998; 25:1308.
5. Chouraqui P, Livschitz S, Sharir T, Wainer N, Wilk M, Moalem I,
and Baron J. Evaluation of an attenuation correction method for
thallium-201 myocardial perfusion tomographic imaging of patients
with low likelihood of coronary artery disease. J Nucl Cardiol 1998;
5:369–377.
6. Corbett JR and Ficaro EP. Clinical review of attenuation-corrected
cardiac SPECT. J Nucl Cardiol 1999; 6:54.
7. TH Farncombe; HC Gifford; MV Narayanan; PH Pretorius; EC
Frey; and MA King, “Assessment of Scatter Compensation Strategies
for Ga-67 SPECT Using Numerical Observers and Human LROC
4598 001 91041 Rev A
Studies” THE JOURNAL OF NUCLEAR MEDICINE 45:5, 2004.

8. Ficaro EP, Fessler JA, Ackermann RJ, Rogers WL, Corbett JR, and
Schwaiger M. Simultaneous transmission-emission thallium-201
cardiac SPECT: effect of attenuation correction on myocardial tracer
distribution. J Nucl Med 1995; 36:921.
9. Frey EC, Tsui BMW, “A new method for modeling the spatially
Ph ilips Healthcare
variant, object-dependent scatter response function in SPECT”.

IEEE Nuclear Science Symposium, 1996. Conference Record
2:1082-1086.
NM Application Suite on IntelliSpace Portal Version 5.0 1

10. Frey EC, Tsui BMW, and Perry, JR. Simultaneous acquisition of
emission and transmission data for improved thallium-201 cardiac
SPECT imaging using a technetium-99m transmission source. J Nucl
Med 1992; 33:2238–2245.
11. Gallowitsch HJ, Sykora J, Mikosch P, Kresnik E, Unterweger O,
Molnar M, Grimm G, and Lind P. Attenuation-corrected
thallium-201 single photon emission tomography using a
gadolinium-153 moving line source: clinical value and the impact of
attenuation correction on the extent and severity of perfusion
abnormalities. Eur J Nucl Med 1998; 25:220.
12. Gallowitsch HJ, Unterweger O, Mikosch P, Kresnik E, Sykora J,
Grimm G, and Lind P. Attenuation correction improves the detection
of viable myocardium by thallium-201 cardiac tomography in
patients with previous myocardial infarction and left ventricular
dysfunction. Eur J Nucl Med 1999; 26:459.
13. Gilland DR, Jaszczak RJ, Greer KL, and Coleman RE. Transmission
imaging for nonuniform attenuation correction using a three-headed
SPECT camera. J Nucl Med 1998; 39:1105–1110.
14. Glick SJ, Penny BC, King MA, and Byrne CL. Noniterative
compensation for the distance-dependent detector response and
photo attenuation in SPECT imaging. IEEE Trans. MED. EMG.
1994; 13(2):363–374.
15. Graham LS. Quality control for SPECT systems. Radiographics 1995;
15:1471.
16. Hashimoto J, Ogawa K, Kubo A, Ichihara T, Motomura N,
Takayama T, Iwanaga S, Mitamura H, and Ogawa S. Application of
transmission scan-based attenuation compensation to
scatter-corrected thallium-201 myocardial single-photon emission
tomographic images. Eur J Nucl Med 1998; 25:120.
17. Hendel RC, Berman DS, Cullom SJ, Follansbee W, Heller GV, Kiat
H, Groch MW, and Mahmarian JJ. Multicenter clinical trial to
evaluate the efficacy of correction for photon attenuation and scatter
in SPECT myocardial perfusion imaging. Circulation 99 1999; 2742.
18. Robert C. Hendel, MD, James R. Corbett, MD, S. James Cullom,
PhD, E. Gordon DePuey, MD, Ernest V. Garcia, PhD, and Timothy
M. Bateman, MD. The value and practice of attenuation correction
for myocardial perfusion SPECT imaging: A joint position statement
from the American Society of Nuclear Cardiology and the Society of
Nuclear Medicine. J Nucl Cardiol Jan/Feb 2002; 9(1):135–143.
Ph ilips Healthcare
19. Jimenez-Hoyuela JM, McClellan JR, Alavi A, and Araujo LI. Impact
of the correction of the attenuation in myocardial perfusion image
with SPECT. Rev Esp Cardiol 1998; 51(Suppl 1):26.
20. King MA, Tsui BM, and Pan TS. Attenuation compensation for
cardiac single-photon emission computed tomographic imaging: Part
1. Impact of attenuation and methods of estimating attenuation
maps. J Nucl Cardiol 1995; 2:513.
2 NM Application Suite on IntelliSpace Portal Version 5.0

References for Reconstruction 1.1
21. Kluge R, Sattler B, Seese A, and Knapp WH. Attenuation correction

by simultaneous emission-transmission myocardial single-photon
emission tomography using a technetium-99m-labelled radiotracer:
impact on diagnostic accuracy. Eur J Nucl Med 1997; 24:1107.
22. Matsunari I, Boning G, Ziegler SI, Kosa I, Nekolla SG, Ficaro EP,
and Schwaiger M. Attenuation-corrected rest thallium-201/stress
technetium 99m-sestamibi myocardial SPECT in normals. J Nucl
Cardiol 1998; 5:48.
23. Merz CN and Berman DS. Imaging techniques for coronary artery
disease: current status and future directions. Clin Cardiol 1997;
20:526.
24. Moore SC, Brunelle JA, and Kirsch CM. Quantitative multi-detector
emission computerized tomography using iterative attenuation
compensation. J Nucl Med 1982; 23:706–714.
25. Prvulovich EM, Lonn AH, Bomanji JB, Jarritt PH, and Ell PJ. Effect
of attenuation correction on myocardial thallium-201 distribution in
patients with a low likelihood of coronary artery disease. Eur J Nucl
Med 1997; 24:266.
26. Rajeevan N, Zubal IG, Ramsby SQ, Zoghbi SS, Siebyl J, and Innis
RB. Significance of nonuniform attenuation correction in
quantitative brain SPECT imaging. J Nucl Med 1998; 39:1719.
27. Rigo P, Van Boxem P, Foulon J, Safi M, Engdahl J, and Links J.
Quantitative evaluation of a comprehensive motion, resolution, and
attenuation correction program: initial experience. J Nucl Cardiol
1998; 5:458–468.
28. Rogers WL, Clinthorne HN, Harkeness BA, et al. Field-flood
requirements for emission computed tomography with an Anger
camera. J Nucl Med 1982; 23:162–168.
29. Shepp LA and Vardi Y. Maximum Likelihood reconstruction for
emission tomography. IEEE Trans. MED. EMG. 1982;
MI1(2):113–122.
30. Tan P, Bailey SR, Meekly SR, et al. A scanning line source for
simultaneous emission and transmission measurements in SPECT. J
Nucl Med 1993; 34:1752–1759.
4598 001 91041 Rev A
31. Toft J, Hesse B, and Rabol A. The occurrence of false-positive

technetium-99m sestamibi bull’s eye defects in different reference
databases. A study of an age- and gender-stratified healthy
population. Eur J Nucl Med 1997; 24:179.
32. Dawn M. Toma, CNMT, Michael P. White, CNMT, April Mann,
CNMT, Joseph M. Phillips, CNMT, Denise A. Pelchat, CNMT,
Satyendra Giri, MD, MPH, Gonzalo R. Ucrós, MD, and Gary V.
Heller, MD PhD. Influence of arm positioning on rest/stress

technetium-99m labeled sestamibi tomographic myocardial perfusion
imaging. J Nucl Cardiol , Part 1 Mar/Apr 1999; 6(2):163–168.
33. Tsui BMW, Gullberg GT, Edgerton ER, et al. Correction of
nonuniform attenuation in cardiac SPECT imaging. J Nucl Med
1989; 30:497–507.

34. Vidal R, Buvat I, Darcourt J, Migneco O, Desvignes P, Baudouy M,
and Bussiere F. Impact of attenuation correction by simultaneous
emission/transmission tomography on visual assessment of 201Tl
myocardial perfusion images. J Nucl Med 1999; 40:1301.
35. Frey EC, Tsui BMW. “A new method for modeling the spatially
variant, object-dependent scatter response function in SPECT”.
IEEE Nuclear Science Symposium, 1996. Conference Record
2:1082–1086
36. Ye J. 1992 Quantitative Tc-99m Myocardial Perfusion SPECT with
180 Acquisition, Ph.D Dissertation, Georgia Institute of Technology
37. Hudson HM and Larkin RS, 1994 “Accelerated image reconstruction
using ordered subsets of projection data” IEEE Trans. Med. Imag. 13
601-609
Chang’s Attenuation Correction
1. Budinger TF, Gullberg RT, Nuesman RH: 1979. Emission computer
tomography. Image Reconstruction from Projections:
Implementation and Applications, ed G. T. Herman, Springer-Verlag.
2. Budinger TF, Gullberg RT: 1981. The use of filtering methods to
compensate for constant attenuation single-photon emission
computer tomography. IEEE, Transactions in Biomedical
Engineering BME-28:42-157.
3. Chang LT: 1978. A method for attenuation correction in
radionuclide computer tomography. IEEE, Transactions in Nuclear
Science NW-25:638-643.
Ph ilips Healthcare

2 Cardiac
This section describes the minimum requirements, calculations, and

references for Cardiac.
2.1 Minimum Requirements for Cardiac
2.1.1 MUGA Processing

• 1 gated dataset, LAO view
2.1.2 First Pass

• 1 dynamic dataset
For dynamic first pass acquisitions, use the following parameters:
• time per frame: 30-40 msec
• acquisition time: 30–40 sec
• image resolution: 64 x 64
Important: Longer acquisitions or higher image resolutions may result in
poor system performance due to the resulting high volume of data.
2.1.3 Shunt
2.2 Acquisition Parameters for Cardiac

4598 001 91041 Rev A
2.2.1 First Pass

The application requires a dynamic, rapid framing rate study. Frame rates
of 0.03–0.04 (25–35 frames per second) for 40–50 seconds are usually
employed.
Ph ilips Healthcare
Important: For the ventricular ROI, include only the clearly identifiable
ventricle. Do not include the atrium or any superimposed blood pool area.
For the whole heart ROI, include only the left and right ventricles. Do not
include the atria or any other structures.

2.2.2 Shunt
The application requires a fast dynamic study.
• Acquire a dynamic study of the first transit of a bolus through the
pulmonary system. At least 30 seconds must be acquired at 0.5 seconds
per image.
• Images may be acquired in a 64x64 or 128x128 matrix.
2.3 Calculations for Cardiac

This section contains the calculations used in the Cardiac methods.
2.3.1 General MUGA Computations

Note: The difference between the MUGA C Preference and the MUGA
GBP Preference is in the edge detection algorithm. Both Preferences are
based on the Odyssey GBP implementation, but MUGA C uses the MUGA
C edge detection algorithm.
The following table captures the results generated by MUGA:

EF = (((ED - Bkg) - (ES - Bkg)) / (ED - Bkg)) * 100
where
ED = Mean of raw pixel values within ED region
ES = Mean of raw pixel values within ES region
Bkg = Mean of raw pixel values within Bkg region
PFR = Max(d/dc(ROI)) / (Max(ROI) * frameTime / 1000)
where
ROI = Mean counts within an ROI
d/dc = First derivative of ROI
Max = Maximum value in array
Max(ROI) = ED volume
Ph ilips Healthcare
PER = Min(d/dc(ROI)) / (Max(ROI) * frameTime / 1000)
where
ROI = Mean counts within an ROI
d/dc = First derivative of ROI
Min = Minimum value in 2/3 part of the array
Max = Maximum value in the array
Max(ROI) = ED volume

Calculations for Cardiac 2.3
TPF = Time[Max(d/dc(ROI))] - Time[ES]
where
Time[x] = Time in seconds of occurrence
Filling at 25%, 33%, 50%, 75% = 100 * (fractionalEDVolume -
ESVolume) / totalEDVolume
2.3.2 General First Pass Computations

The whole-heart smoothed curve appears with reference lines. For
analysis, these reference lines can be adjusted to define the points to be
extracted from the ventricular curve. The extracted ventricular curve is
then used to identify the frame pairs (1-5) either automatically (by the
program) or manually by adjusting the reference line pairs for each frame
pair individually. The identified frame pairs are used to calculate the
ejection fraction values and for the generation of a 16 frame gated study.
Ejection Fraction for a frame pair i (EF(i)) is calculated using the counts
at the diastolic and the systolic frames, as shown below:
• EF(i) = ((EDCounts - ESCounts) / EDCounts) * 100;
• EDCounts is the end-diastolic counts and ESCounts is the
end-systolic counts of an identified frame pair.
Average EF = Summation(EF(i)) / N
• N is the number of identified frame pairs.
The identified frame pairs are then used to generate and display the 16
frame gated study.
• The time activity curves for all the user defined organs are also
computed and displayed.
2.3.3 General Shunt Computations

The right pulmonary raw curve is presented with reference lines. The
reference lines on the curve may be adjusted to define the points to fit
4598 001 91041 Rev A
within the gamma curve. The gamma fit is interactively regenerated to

match every movement of the reference lines. Next, the system generates a
difference-curve which is again gamma fitted.
The pulmonary-to-systemic flow ratio (Qp/Qs) is then calculated as the
ratio of the areas of the first-transit curve and the abnormal recirculation
area.
• Qp/Qs = A1 / (A1 - A2)

• A1 is the fitted lung area and A2 is the fitted re-circulation area.
SVC quality control time is used to evaluate the bolus quality
• SVC Control Time = T1/2 - P1/2

• P1/2 is time range from time zero to half the peak counts of SVC curve
and T1/2 is the half time value for the SVC Curve.
• Lung Raw Curve, Lung Fitted Curve, Lung Diff Curve, and Lung Diff
Fitted Curve are all displayed along with the SVC Curve, including the
curves for all user-defined organs.
Pulmonary Transit Time = T1/2 - P1/2, where P1/2 is time range from
time zero to half the peak counts of Lung curve and T1/2 is the half time
value for the Lung Curve.
2.3.4 GBP Multiple Algorithm

This section describes the algorithm used for calculating GBP Multiple.
Input
The input for the algorithm is a gated sequence of images. The images
have a resolution of either 64x64 or 128x128. Images of other resolutions
are resized.
Preprocessing
Spatial and temporal filtering are applied next.
Spatial filtering by applying a filter in the frequency domain
The input images are filtered in the frequency domain, so an FFT is

performed. A pre-defined window in the frequency domain is loaded and
the contents are multiplied with the FFTed image (this being equivalent to
convolution in the spatial domain). The kernels used for this stage are
defined in two binary files , "mgakernel128" and "mgakernel64", and are
chosen based on the resolution of the input images. After the kernel is
applied, the image is IFFTed and rescaled to get a smoothed image.
Temporal filtering
A 1-2-1 temporal filter is applied on the smoothed images. Only the first
and last images are filtered using a 3-1/ 1-3 kernel.
Ph ilips Healthcare
Setting the Bounding Box

The algorithm seeks edges within a bounding box that is defined by the
user. The location of the bounding box itself is estimated by the
algorithm. The initial size is pre-defined. However, both could be altered
by the user. The bounding box is only used for the first image in the image
array and the detected edge of image n is used as the bounding box for
image n+1.

Estimating the location of the bounding box
The bounding box is placed with its center coinciding with the center of
the left ventricle. To compute the center of the left ventricle, first the
center of the heart in the image is determined. This is done using a center
of mass algorithm which defines the center of any image to be the
weighted average of all the pixels within the image. The weight in this case
is the location of each pixel. When the center of the heart has been
identified, a search is performed in the near vicinity (limited to a 4x4
square around the center of the heart) for the maximum pixel. The
location of the maximum pixel is assumed to be the center of the left
ventricle.
Only pixels to the right of the heart center (x values vary from Xc to end of
the image and y varies from Yc - 4 to Yc + 4) are considered for
determining the LV center. The RV center is also calculated in a similar
fashion but pixels to the left of the heart center are considered: (Xc ,Yc) :
Heart center.
Estimating the size of the bounding box
The size of the bounding box is set with a minor axis length of 80 and a
major axis length of 105. This is assumed to an interpolated display where
the resolution of images is 512x512.
Detecting Ventricle Edges

Once the user sets the bounding box such that it completely surrounds the
left ventricle, a search for the edges is done within this region. Broadly, the
principle used here is to apply a second derivative filter on each image and
search for an appropriate zero crossing. The steps to accomplish this are
described below.
Filtering the images
This step is done on the raw data and is not to be confused with the
filtering done in the frequency domain as part of pre-processing.
4598 001 91041 Rev A
• The raw data is filtered temporally using a 1-2-1 kernel. Images are
picked cyclically for applying the kernel.
• A 3x3 averaging kernel is applied spatially to smooth the image
obtained from the previous step. This step could be repeated several
times. However, by default this averaging is performed only once.
• A hyperbolic tangent transform is applied on the resulting image.

Edge detection is performed on the result of this step. All parametric
calculations are done on the data resulting from the filtering in the
frequency domain.

Resizing of images
Images are resized to 128x128. All subsequent calculations on images are

done on these resized images.
Convolving each image with a second derivative filter
Second derivative kernels of two sizes (5x5 and 7x7) are available. The 7x7
kernel is the default choice and is used to convolve the filtered image. The
filters used are:
Convolving each image with a second derivative filter
Second derivative kernels of two sizes (5x5 and 7x7) are available. The 7x7
kernel is the default choice and is used to convolve the filtered image. The
filters used are shown below.
Figure 1 Filters for convolving
Determining the seed points
The bounding box defined is a continuous region. However, edges are

sought for along a discrete number of radials. The seeds identify the end
points of those radials beginning at the centre and terminating on an edge
of the bounding box.
• The number of radials is fixed at 64. So 64 points on the elliptical
bounding box have to be identified at angles of 2Pi/64 from each
Ph ilips Healthcare
other.
• First the end points of each segment in the bounding box are obtained
and the segments are ordered in a trigonometric sense.
Note that the number of segments in the polygon is determined by the
number of control points needed to specify the geometry, and hence
could be less than or greater than the number of radials. This means
that we need to find points on these segments at specified angles to the
center.

• For every angle at which a radial end point is to be determined, the

corresponding edge on the ellipse/bounding box that spans the angle
is extracted.
• The equation of this line is determined using its extremities.
• The equation of the line passing through the center of the ellipse and
at the required angle (the required radial) is determined.
• The "seed point" or the extremity of the radial is the point of intersec-
tion of these two lines.
Creation of polar map
To facilitate easy lookup of a pixel value on a radial, the values in the

second derivative image are stored as a polar map. However, as an
optimization measure the entire image is not converted from Cartesian
coordinates to polar coordinates. Only the section of the image that
completely covers the ellipse is worked on. Each row of this polar map is
associated with an angle. A map that determines the search boundary for
this angle is created by computing the distance of every seed point from
the center of the bounding box.
Seeking the edge
The pixel with maximum value on the specified radial is sought for in the
second derivative image. This search is done only in the vicinity of the
corresponding seed point. The initial search range is determined by the
derivative sensitivity parameter. This is set to a value of 7 by default. This
means that a search for the max pixel is made on 7 pixels on either side of
the seed point along the radial.
• If no pixel in this range has a value greater than 0, then the edge point
is set to the seed point.
• If a pixel with value greater than 0 is found, then we traverse back
along the radial from that point on until a zero-crossing is found.
When this is found, this is set as the edge point.
4598 001 91041 Rev A
Smoothing the edges
The edges found at this stage can be staggered and not necessarily smooth.
So some basic corrections are applied. A correction is applied to ensure
that the variation in the distance of the edge from the centre is limited to
the standard deviation of the distances from the centre.

The edges thus corrected are further averaged. This step could be done
many times but by default is performed twice. A nearest neighbor
algorithm is used to smooth every edge point based on the values of its 5
neighboring edge points. The edge points are picked in a cyclic fashion.

Conversion back into Cartesian coordinates
The operations performed in the last few steps were in the polar
coordinates. So the edges determined are now converted back to Cartesian
coordinates and stored for display/analysis.
Determining the background region

Once the left ventricle has been identified on all the images in the gated
sequence, the algorithm proceeds to the identification of the background
region. The background region is identified on the end-systole image.
Estimate the end-diastole and end-systole intervals
The pixels that lie within each of the identified regions are determined and
the total count in each of the regions is computed. The interval with the
maximum pixel count in the left ventricle is said to be the end-diastole
and the one with the minimum counts is assumed to be the end-systole.
Identifying the background region
The region identified in the end-systole is taken. It is then shifted

horizontally by a preset factor (the initial horizontal offset is set to twice
the width). Then the top quarter of this region is chopped off. Finally a
strip of a specified width is peeled off from the right most edge of this
region and this is used as the background region. The initial width of the
strip is set to 5 pixels wide on a region that is interpolated to map to a
resolution of 128x128.
Applying morphological corrections on the background region
Initially a 3x3 binary erosion kernel is applied on the region.

Subsequently a 3x3 binary dilation kernel is applied twice in succession to
dilate the region.
Conclusions
The GBP Multiple algorithm identifies the left ventricle following the
identification of the region of the left ventricle by the user. Subsequently a
Ph ilips Healthcare
radial search is performed on the second derivative image starting from the
centre to detect the edges of the ventricle.
2.3.5 MUGA C Algorithm

This section describes the algorithm used for calculating MUGA C.

Preprocessing
The data is smoothed both spatially and temporally.
Spatial Smoothing
For spatial smoothing the Wiener filter is used.
Temporal Smoothing
Then a simple 1-2-1 kernel in wrap mode is used to temporally smooth

the image data.
Normalization
The smoothed data obtained from the previous step is then normalized.
The scale factor is the ratio between the mean pixel value of the raw image
to the mean pixel value of the smoothed image.
Detecting ventricle edges

Conceptually, radials are drawn outward from the centre of the bounding
box towards its edges. Edge detection is performed along the radials. The
bounding box technique is common to both GBP Multiple and MugaC.
1-1-1 Temporal Smoothing
Before detection of an ROI for a given frame, the frame is averaged with
the previous frame and subsequent frame. For the first (i.e., 0th) frame,
the previous frame will be N-1 and the next frame will be 1 .This is done
to smooth the image and avoid spurious fluctuations.
Converting from Cartesian to Polar Coordinates

4598 001 91041 Rev A
With the centre of the bounding box as the origin for the polar coordinate
system, the image is converted to a polar format. With this the radials in
the Cartesian coordinate system now become horizontal lines, making
traversal and indexing much simpler. The conversion between the two
coordinate systems is simple and is obtained by the following:
Polar to Cartesian: x= r * cos (theta), y = r * sin (theta)
Cartesian to Polar: theta = (2 * PI * column value of the pixel) /

Maximum column value
During conversion from Cartesian to Polar a spatial 3x3 filter is applied in
the vicinity of the radial pixels to get the closest fit.

Spatial filtering of the polar image
The polar image is filtered by averaging each column with 15 pixels if the
image matrix is larger than 64x64, else 7 pixels are used.
Second derivative filter applied
The horizontal second derivative of the pixels in polar domain is

computed. The essence of converting to polar domain can be understood
here. We have to compute the derivative in only the horizontal direction,
which is equal to finding the second derivative of any radial along its
direction. The second derivative is computed by convolving with the
kernel [1, 0, -2, 0, 1]
Seeking the edge
The maximum pixel along each row (radial) of the 2nd derivative is
determined. Three neighboring points are averaged before finding the
maximum. The positions of the maximum could be as shown in figure
below.
Figure 2 Example maximum positions
Each of these points corresponds to a coordinate of the ROI that shows

the edge of the ventricle. The distance of the point on each horizontal line
from the first column in the figure corresponds to the radius of the edge
from the centre in Cartesian system. Joining all these points gives the
contour of the edge in the Cartesian system.
Ph ilips Healthcare
Applying a Median Filter
A 5 point median filter is applied on these positions to eliminate outliers.
Ensuring the contour stays spherical by averaging
A simple averaging technique is used to ensure that there are no jagged

edges or protrusions on the ROI. This is done with the knowledge that the
ventricles are usually ovoid or spherical.

The Contour indices (Note: index here is the column value of the contour
in a given row) are processed as below. Each contour index is compared
with the contour index on the previous row and subsequent row. The
comparison between the three contour indices can lead to any of the below
cases. The principal idea here is to reduce variance among the points that
form the ROI to get a smooth shape.
Case 1: If the current contour index is equal to the previous contour
index and also equal to the next contour index as shown in the figure
below, then compare the current contour index with the smoothed second
derivative at this point, which is the output of the method in the Section
"Second derivative filter applied". If the value of the second derivative is
bigger, the current contour index is moved to the right by one; otherwise it
is moved to the left, as shown in the figure below.
Case 2: If the next contour index and previous contour index are equal,
the current contour index is less than both of them, and the second
derivative is more than the current contour index, then the current
contour index value is increased by one or moved to the right, as shown in
the figure below.
Case 3: If the next contour index and previous contour index are equal,
the current contour index is more than both of them, and the second
derivative is less than the current contour index, then the current contour
4598 001 91041 Rev A
index value is decreased by one or moved to the left as shown in the figure
below.

Case 4: If the current contour index and previous contour index are equal,
the next contour index is less than both of them, and the second derivative
is less than the current contour index, then the current contour index
value is decreased by one or moved to the left as shown in the figure below.
Case 5: If the current contour index and next contour index are equal, the
previous contour index is more than both of them, and the second
contour index value is increased by one or moved to the right, as shown in
the figure below.
Case 6: If the current contour index and previous contour index are equal,
the next contour index is more than both of them, and the second
contour index value is increased by one or moved to the right as shown in
the figure below.
Case 7: If the current contour index and next contour index are equal, the
previous contour index is less than both of them, and the second derivative
is less than the current contour index, then the current contour index
Ph ilips Healthcare
value is decreased by one or moved to the left, as shown in the figure

below.

References for Cardiac 2.4
This smoothing process is iterated multiple times. The number of

iterations depends on the size of the image and whether the image is the
first frame of an image array.
Converting polar to Cartesian coordinates
The edges obtained in the polar coordinates are converted back to polar
coordinates.
2.4 References for Cardiac

“Improved Detection of anterior Left Ventricular Aneurysm with
Multiharmonic fourier Analysis” Heric B. Valette, Michel H. Borguignon,
et al. J Nucl Med 1990; 31:1303-1306
“Filters and Fourier Analysis of Gated Blood Pool Studies: A Search for
the Optimal Combination” H. Valette, M H Bourguignon, et al. Phys Med
Biol 1990; 35(1):1-11
“A fully automated determination of the left ventricular region of interest
in nuclear angiocardiography” ML Goris, JH McKillop, PA Briandet
Cardiovasc Intervent Radilo 1981;4(2):117-123
4598 001 91041 Rev A

Ph ilips Healthcare

3 Whole Body
This section describes the minimum requirements and calculations for

Whole Body.
3.1 Minimum Requirements for Whole Body
3.1.1 Ileosacrum Ratio

• 1 static dataset
3.1.2 Three-Phase Analysis

• 1 dynamic dataset (bone flow sequence)
3.1.3 Three Phase + SI Ratio

• 1 dynamic dataset (bone flow sequence) + 1 spot
3.2 Calculations for Whole Body and Bone

This section describes the calculations performed within the Whole Body
and Bone application.
The ratio between two ROIs are calculated as:
• Ratio = Numerator ROI / Denominator ROI
The difference between two ROIs is as follows:
• Difference = ROI1 - ROI2 (where ROI = sum of counts within iden-
4598 001 91041 Rev A
tified region of interest).

Ph ilips Healthcare

Ph ilips Healthcare

4 Lung

Lung.
4.1 Minimum Requirements for Lung
4.1.1 Lung Ventilation and Perfusion

For Lung Perfusion analysis:
• 1 static lung perfusion dataset -Anterior (ANT) or Posterior (POST)
view
Optionally, you can load up to 7 additional static lung perfusion datasets,
for a total of 8 (1 each in the following views: LPO, POST, RPO, RLAT,
RAO, ANT, LAO, and LLAT)
For Lung Ventilation analysis:
• 1 static lung ventilation dataset- Anterior (ANT) or Posterior (POST)
view
Optionally, you can load up to 7 additional static lung ventilation datasets,
for a total of 8 (1 each in the following views: LPO, POST, RPO, RLAT,
RAO, ANT, LAO, and LLAT)
The Perf + Vent preference requires the datasets for both the Perfusion and
Ventilation preferences.
4.1.2 Washout Analysis

• 1 dynamic washout dataset
• 1 breath hold
4598 001 91041 Rev A
• 1 equilibrium
4.1.3 Geometric Oblique Analysis

• 1 anterior static
Ph ilips Healthcare
• 1 posterior static
• 1 left posterior oblique static
• 1 right posterior oblique static

4.2 Calculations for Lung
Calculations for posterior and anterior images are as follows:
• Total counts of the left lung and the right lung regions.
• Percentage counts of each of the segments in a lung.
• Percentage comparison counts of the right and the left lung regions.
If both posterior and the anterior images are available, the geometric
mean value for both the left lung and the right lung are calculated. The
calculation applies even if the anterior and posterior ROIs are different
(there is no area correction):
Geometric mean counts = SQRT(POST Counts * ANT Counts)
Values for lung images are calculated as follows:

1. Calculate upper, middle, and lower segment percentages for left lung.
2. Calculate total percentage for left lung (LTotal%). This matches the
sum of all the segment percentages calculated above.
3. Calculate right lung total percentage (RTotal%):
RTotal% = 100 - LTotal%
4. Calculate right upper segment percentage and right middle segment

percentages (RUpper% and RMiddle%).
5. Calculate lower right segment percentage (RLower%), not from
counts:
RLower% = RTotal% - RUpper% - RMiddle%
In addition to these results, the total counts under the region for all the
user defined organs are also calculated.
Ph ilips Healthcare

5 Renal
This section describes the minimum requirements for Renal.
5.1 Minimum Requirements for Renal

The following table specifies (with an ‘X’) the datasets required by all the
methods.
Pre- Post- Pre- Post- Multi-

Method Flow Function
syringe syringe void void phase
Simple Reno-
gram X1 X1
PrePost Lasix X2 X2 X2
Post Renogram X (diure-

Lasix sis)
Patlak X1 X1
Hilson Index X2 X2 X2
GFR Gates X2 X2 X X X2
ERPF Schlegel
with Void X1 X X X X X1
ERPF Schlegel X1 X X X1
ERPF MAG3 X2 X2 X X X2
poste-
DMSA Static
rior stat-
Ratio
ics
X1 X1
4598 001 91041 Rev A
Deconvolution
Cortical Analy-
sis X1 X1
Split T1/2 X (diure-

sis)
1: Function or Multi-phase
Ph ilips Healthcare
2: [Flow and Function] or Multi-phase

Important: The application uses the Pre Injection Syringe and Post
Injection Syringe images, if loaded, to compute the injected dose. Use the
following guidelines when acquiring pre- and post-injection syringe images:

• Place each syringe on the table 30 cm from the collimator face. Under
each syringe, include all other items typically used between a patient
and the table (e.g., sheet, pad, restraining strap).
• If the radiopharmaceutical activity of the pre-injection syringe is
greater than 3 mCi, use a syringe shield when acquiring images for
both the pre- and the post-injection syringes. Pre-injection syringe
activity less than 3 mCi does not require shielding.
You must use the same shield for both images.
• If you typically image patients in the supine position, position the
detector under the table; otherwise, if you use a different patient posi-
tion, position the detector accordingly.
Important: For GFR and ERPF calculations, the normalized and corrected
kidney counts are determined from 60 seconds of data collected 2 to 3
minutes following tracer arrival in the kidney. To obtain correct results, the
maximum frame duration for a dynamic must be 1 minute or less.
5.2 Acquisition Parameters for Renal
5.2.1 Simple Renogram

• Matrix 64x64 or 128x128
• Each stage must be at least 60 sec in duration
5.2.2 PrePost Lasix

• Acquisition must continue 20 minutes post lasix injection
5.2.3 Post Renogram Lasix

• Requires a single framing rate dynamic (1 phase)
Ph ilips Healthcare
5.2.4 Patlak
• 128x128 matrix
• 120 frames at 15 sec/fr

Acquisition Parameters for Renal 5.2
5.2.5 Hilson Index

• Each stage must be at least 60 sec in duration
5.2.6 GFR Gates

• Pre/Post Syringes- up to a 512x512 matrix
• Dynamic up to 128x128 matrix, minimum 6 min acquisition
5.2.7 ERPF Schlegel with Void

• Pre/Post syringes: 60 sec static, up to 512x512 matrix
• Dynamic up to 128x128 matrix, time per frame must divide evenly
into 60 sec
• Less then 300 uCi of Hippuran
5.2.8 ERPF Schlegel

• Pre/Post syringes: 60 sec static, up to 512x512 matrix
• Dynamic up to 128x128 matrix, time per frame must divide evenly
into 60 sec
• Less then 300 uCi of Hippuran
5.2.9 ERPF MAG3

• Pre/Post Syringes: 60 sec static, up to 512x512 matrix
• 2-segment variable framing rate study (must be at least 6 minute acqui-
sition)
4598 001 91041 Rev A
5.2.10 DMSA Static Ratio

• (No minimum requirements, just requires static images)
5.2.11 Deconvolution
• Dynamic: 64x64 or 128x128 matrix

• Single phase: 90 frames, 20 sec/fr
• 2 phase: 60 frames, 1sec/fr; 57 fr, 20 sec/fr

5.2.12 Cortical Analysis
• (Dynamic 120 must be divisible by the time per frame)
5.2.13 Split T1/2

• Requires a single framing rate dynamic (1 phase)
5.3 Calculations for Renal

This section describes the calculations performed within the Renal
application.
For renal applications that calculate it, T1/2 is reported only if a peak
occurs before the last frame of the acquisition, even if it is only one frame
before. Some curves may not have an obvious peak because of small
changes in counts that are difficult to discern in the display. But even if the
change is small, T1/2 is still reported, since the peak does exist.
5.3.1 Generating Multi-phase Dynamic Frames

When the Renal application generates all the frames in a multi-phase
dynamic study, it uses the following algorithm.
MaxFrameRate = the highest frame rate in the loaded
image
PhaseFrameRate = the frame rate for an individual phase
PhaseFrames = the number of frames in the phase
1. For each phase, first calculate: MaxFrameRate/PhaseFrameRate =

Rate
2. If the result is not an integer, round down.
3. Then calculate: PhaseFrames/Rate = Frames
4. If the result is not an integer, combine the remaining frames into 1
frame. The frames must be the first frames in the phase.
5. Add all the Frames results to get the total frames generated for the
Ph ilips Healthcare
study.
Consider an image with a first phase consisting of 30 frames at 2sec/frame,
and a second phase consisting of 60 frames at 15sec/frame. For the first
phase, we get:
15/2 = 7 (rounding down)
30/7 = 4 Frames + 1 Frame from the remaining 2 = 5
Frames
For the second phase we get:

Calculations for Renal 5.3
15/15 = 1
60/1 = 60 Frames
The total frames is then:

5 + 60 = 65
5.3.2 Simple Renogram

Peak Counts, Peak Time, and the T 1/2 values are calculated for each
kidney from the background subtracted kidney time activity curves.
Differential Perfusion of each kidney is obtained by taking the ratio of the
counts of each individual kidney with the total kidney counts between 2-3
mins of the renal study. Renal Retention percentage of each kidney is the
percentage residual activity remaining in the kidneys at the residual
activity time.
5.3.3 Lasix Pre-Post

The results displayed for the Lasix Pre-Post technique are time related
parameters calculated from the kidney regions during various stages of the
renal study like the differential phase, lasix injection time. There are also
some count related statistics shown for these stages along with the ratio
comparisons.
The following computations are displayed:
• 20 min/peak ratio: (Counts at user specified minutes (default value =
20) / peak counts) * 100.
• Diff time in minutes: Displays starting and ending times is specified,
in minutes for the differential calculation.
• T1/2 from peak (min): Time in minutes when each kidney curve
reaches ½ count starting at the time of peak counts.
• Peak Post Lasix (min): Time in minutes when each kidney curve
reaches peak. It is normally 0.
• T1/2 from Lasix (min): Time in minutes when each kidney curve
4598 001 91041 Rev A
reaches ½ counts starting at the time of Lasix injection.

• xM/Pre-L: The ‘x’ in the name is the value of the “Time of post Lasix
percent remaining” input in the Define Regions workstep. By default,
this value is 10. If the sum of the values for “Time of Lasix Injection”
and “Time of post Lasix percent remaining” exceeds the total study
time, the result is reported as a dash (‘-’). The result is computed as:
counts for each kidney at the time specified/

counts in frame before Lasix injection

• xM/PK-L: The ‘x’ in the name works the same way as above. The
result is computed as:
counts for each kidney at the time specified/
peak counts
5.3.4 Post Renogram Lasix

The TACs have adjustable marker values to allow you to manually set the
start and end point of the respective curve fits. The default placement is to
Start at Lasix Injection Time and End at end of curve. The fits are then
used to compute and display T1/2 values for the kidneys.
5.3.5 GFR Gates Method

The following table captures results generated by the GFR Gates method:
* If a syringe image is acquired for less than 60 sec, then a correction factor
is used to extrapolate the counts to 60 sec:
60/frame rate of syringe image Ph ilips Healthcare

5.3.6 ERPF Schlegel Method

The following table captures the results generated by ERPF Schlegel
method:
4598 001 91041 Rev A

5.3.7 ERPF Schlegel with Void
ERPF schlegel with Void is an extension of the ERPF schlegel technique.
Please refer to the section on ERPF Schlegel for those calculations.
However, there are some extra results that are computed by this technique
which are explained below:
5.3.8 ERPF MAG3 Method

The following table captures the results generated by ERPF MAG3
method:
Ph ilips Healthcare

5.3.9 Deconvolution Method

The following table captures the results generated by the Deconvolution
method:
5.3.10 DMSA
The following table captures the results generated by the DMSA method:
4598 001 91041 Rev A

5.3.11 Hilson Index Method
The following table captures the results generated by the Hilson Index
method:
5.3.12 Patlak Method

Time markers are identified on the left and the right kidney TACs to be
used for the linear fits. The relative functions for the individual kidneys
are calculated by taking the ratio of the individual linear fit gradients to
the gradient sum. Uptake ratios at different intervals of time (2, 3, and 20
minutes) with respect to the peak value are also calculated. A more
detailed discussion follows.
The intrarenal tracer concentration B(t), in the initial circulation from
blood to kidney, according to Patlak's model, is:
where Ku is a renal influx rate constant, A(t) is the arterial tracer

concentration at time t, and Vn is an initial nonspecific distribution
volume in the kidney.
The above equation divided by A(t) yields:
A Patlak plot is created by plotting B(t)/A(t) on the y-axis and

Ph ilips Healthcare
on the x-axis. Fitting the curve with a straight line. The slope of the line is
Ku and the y-intercept is Vn.
Next we create the linear regression curves. We use the following to find
the marker positions from the left background subtracted renal curve if
available, or the right one if not:
LeftKidneyDepth = 153.1 * W / H + 0.22 * A + 0.77
RightKidneyDepth = 161.7 * W / H + 0.27 * A - 9.4,

where W, H, and A are respectively patient's weight (kg), height (cm) and
age (years).
Next, perform linear regression on the background subtracted renal curves
between the marker positions. Then create the fitted curves.
Use the same marker positions found before to create the linear regression
curve for the cardiac curve.
For the Patlak transformation, create the Patlak curve PatCurve(x[i], y[i])
where f is the kidney curve, c is the cardiac curve.

The resulting curve is then scaled:
The Patlak slope is the slope of the rising part of the above curve. The user
is able to adjust the limits for the slope calculation using regression.
Left and right Patlak slope percentages relative to the sum of the slopes are
calculated and presented.
5.3.13 Cortical Analysis

Various curve related parameters like the Peak Time, Peak Counts, T 1/2,
4598 001 91041 Rev A
etc. are calculated on the background subtracted time activity curves

similar to the other renal techniques.

Ph ilips Healthcare

6 Endocrine
This section describes the minimum requirements, calculations, and

references for Endocrine.
6.1 Minimum Requirements for Endocrine
6.1.1 Thyroid
• 1 Uptake (static Thyroid image)
Optionally, you can also load the following:
• a thyroid marker image
• a pair of syringe images: Full and Empty
• Injection site planar images
• LAO
• RAO
• Anterior
Important: When obtaining the Syringe Full and Syringe Empty data, use
acquisition times appropriate to avoiding rollover. Tailoring the acquisition
duration to the dose activity avoids pixel overflow in the resulting images;
this is important because the application requires precise counts from
these images to calculate valid thyroid uptake values from the injected
dose. The application normalizes the counts in the syringe images based on
the acquisition time of the thyroid image.
When acquiring thyroid images, the images need to contain a minimum

of scatter contribution; otherwise, the uptake based on dose computed
from pre- and post-syringe images may be underestimated.
4598 001 91041 Rev A
6.1.2 Parathyroid Subtraction

• 1 Thyroid (Tc or Early)
• 1 Parathyroid (Tl or Late)
Ph ilips Healthcare
The two statics must have the same matrix size.

Optionally, you can also load the following:
• LAO
• RAO

• Dual Phase
6.1.3 Parathyroid Dual-Phase

• 1 Dual-phase
6.2 Calculations for Endocrine

This section describes the Parathyroid and Thyroid calculations
performed within the Endocrine application.
6.2.1 Calculations for Parathyroid

• Subtracted Image = Blood Pool Image - subtraction factor* Thyroid
Image
• Total Count of the Thallium Image = Shown on the image viewer.
• Total Count of the Technetium Image = Shown on the image viewer.
6.2.2 Calculations for Thyroid Uptake

Thyroid's result computation is based on the following Data Bucket
selection (in addition to Thyroid image data):
• Full Syringe Image
• Full + Empty Syringe images
• Full + Empty + Injection site images
Calibration Factor: If Syringe data is used from a Dose Calibrator then a
calibration factor must be determined that establishes the relation between
counts per minute and kiloBequerel (or uCi). The calibration factor
should be determined using a known amount of activity in MBq or mCi
(e.g., 37MBq or 1 mCi).
By default, the EBW system is set to use Bq. To change to Ci, open the
EBW Preferences and change it in the PET Preferences Viewing page. If
Ph ilips Healthcare
you use mCi, convert uCi for kBq in this example.

Measure this activity on the gamma camera using the same collimator and
distance as has been used for syringe measurements. Acquire a static image
for 10 or 20 seconds. Determine the total counts in the image. Convert
the counts to counts per minute, then divide the counts per minute by
activity as expressed in kiloBequerels or microCuries, e.g. 37000 kBq or
1000 uCi. This calibration factor will have to be determined again if
different collimators are used or when the gamma camera has been tuned.

Calculations for Endocrine 6.2
NOTE: If the Full Syringe Image or the Empty Syringe Image is not
selected by the user as one of the bucket data, then the user has to provide
Calibration Factor and the Activities for each of them as input, using
which the counts are calculated as below. Finally, these count values are
used in the calculation of thyroid's percent uptake which is described in
the following section.
Full Syringe Count = (Calibration Factor * Unit Conversion Factor) * Full
Syringe Activity
Empty Syringe Count = (Calibration Factor * Unit Conversion Factor) *
Empty Syringe Activity
6.2.3 Thyroid Size Calculation Details

• Thyroid size in sqcm = (Thyroid’s Polygon region’s area) / 100
• Cubic volume = (sqrt(Thyroid size))3
• Thyroid weight = Thyroid volume = (0.326 * Cubic volume)
6.2.4 Percent Uptake Calculation Details

• Applied Counts = Full Syringe Count - Total Empty Syringe Count
- Injection Site Count (if selected as bucket data by the user).
• Decay correction on Thyroid data: The half time for Technetium is
21720 seconds or 6.03 hr. For I-123 this is 46800 seconds or 13
hours.
• The decay factor that must be used to take care of the difference in
original activity and the activity at the time of the Thyroid measure-
ment can be calculated as follows:
Decay factor = e (t / 21720) x ln2
where t is the time difference in seconds between the time of acquisi-
tion of the thyroid image and the measurement time of the syringe.
• Net Uptake Counts = (Thyroid Counts - (Background Counts /
4598 001 91041 Rev A
Background Pixels)*Thyroid Pixels) * Decay Factor

• Thyroid Percent Uptake = (Net Update Counts/Applied Counts) *
100

Ph ilips Healthcare

7 Esophagus

Esophagus.
7.1 Minimum Requirements for Esophagus

• 1 dynamic for either the Transit or Reflux methods
or
• 2 dynamic datasets (if the Preferences are combined)
7.2 Calculations for Esophagus

The calculations for Esophagus are as follows:
Peak Ratio = Curve’s Peak Time Value (in sec)
The Empty Ratio is calculated thus:
1. Generate the time-activity curve of the region that encloses the
wholeEsophagus and smooth it with a five-point smooth.
2. Find the peak activity value, P.
3. Determine the activity at 10 seconds after peak time, V10.
4. Determine the background as the average counts in the first five
frames of the curve, B.
5. ER = ((P -V10)/(P - B)) x 100%
If the Background value is higher then the Value at 10 seconds after the
Peak-time the Emptying rate will be higher than 100%. In other words, if
B > V10, ER > 100%. This may happen when activity of earlier swallows
has remained in the Esophagus.
4598 001 91041 Rev A
Ph ilips Healthcare

Ph ilips Healthcare

8 Gastro Intestinal

Gastro Intestinal.
8.1 Minimum Requirements for Gastro Intestinal

Minimum requirements for the Gastro Intestinal application are as
follows:
• 2 static images
Optionally, you can load any of the following sets of data into this
application:
• 2 series of static frames
These can be either 2 sets of static images corresponding to Anterior and
Posterior views for Tc-99m, or 2 sets of static images corresponding to
Tc-99m and In-111 in a single view.
• 4 series of static frames
These are 4 sets of static images corresponding to Anterior and Posterior
views for each isotope (Tc-99m and In-111).
• 2 or 4 dynamic sequences
Note: When using statics for analysis, you must load an immediate image
(an image at T = 0) for the analysis results to be reported for the
appropriate times.
When you load static datasets, the application automatically sorts them by
view and then by ascending time of acquisition. The resulting arrays of
static images are used to build dynamic images. The remainder of the
static image processing is therefore the same as for images from dynamic
4598 001 91041 Rev A
datasets. For the sorting to work correctly, the datasets must contain
specific case-sensitive strings: anterior data must contain “ANT”, posterior
data must contain “POST”; additionally, the technetium data must not
contain the substring “In”, and the indium data must not contain the
substring “Tc”.
By default, this application is configured to recognize and automatically
load static datasets with the following names:
Ph ilips Healthcare
• Tc-99m ANT (technetium isotope, anterior view)

• Tc-99m POST (technetium isotope, posterior view)
• In-111 ANT (indium isotope, anterior view)

• In-111 POST (indium isotope, posterior view)
To facilitate manual loading of multiple static datasets for different
isotopes and views, make sure your exam names follow the above
conventions. Label all the datasets for the same isotope and make sure you
view them using the appropriate name from the list above. This ensures
that when you select all the datasets you want to use, the application
auto-loads the selected datasets correctly.
8.2 Calculations for Gastro Intestinal

The counts and pixels for each stomach region shown in the Anterior and
Posterior images are obtained. Conjugated counts for Anterior and
Posterior are calculated for 0, 30, 60, 90, and 120 minutes.
The square-root is taken of the Anterior multiplied by the Posterior
images: sqrt(Anterior x Posterior).
The acquisition times of each image is obtained. Averages of the Anterior
and Posterior acquisition times are calculated for each timed set, while the
calculated conjugated counts are corrected for decay correction for each
specific isotope.
A curve is generated from the identified organ, including linear fit curves
for solid and and exponential fit curves for liquid options. T 1/2 of the
curve and percent retention (at 30, 60, 90, 120, and 240 minute times)
are also calculated.
For static data, the T1/2 curve is created using interpolations at 1 minute
intervals. If T1/2 is found at a point in the curve, the value is reported in
the result table. Otherwise it is extrapolated and the reported value is
appended with ‘*’ in the result table.
Absolute values for % retention are reported only for acquired time when
the raw or residual curve is used. However, if the fit curve is used, the %
retention values are reported for all times in the results table. Retention
values for time for which there is no data are displayed as '-' in the table.
If data is found at a time that is within ±15 minutes of 30, 60, 90, 120,
180, or 240 minutes, it is reported in the result table as the closest of those
values. For example, when using statics, data at minute 46 or 74 would
both be reported as the value for 60.
Ph ilips Healthcare
Other curve behaviors include:

• The curve is not smoothed if there are fewer than 16 points.
• For dynamic data, points on the curve are displayed at the end of the
frame duration. For example, if the time per frame is 60 seconds, the
first point on the curve appears at 1 minute, not 0.
• The timing markers snap to the closest data point when released after
clicking and dragging.

Calculations for Gastro Intestinal 8.2
• Curves are normalized and possibly decay corrected for Anterior and
Posterior.
• The geometric mean is calculated and the resulting curve is then
normalized.
4598 001 91041 Rev A

Ph ilips Healthcare

9 Hepatobiliary

Hepatobiliary.
9.1 Minimum Requirements for Hepatobiliary
9.1.1 Gallbladder
or
• Up to 12 static images
For dynamic acquisitions, the suggested time per frame is 30–120 seconds.
Make sure the number of frames is large enough to accommodate your
clinical procedure’s total time.
Important: While you can load a mixture of statics and dynamics into the
application, you must process each one individually using a different
preference, as if you were loading two patients into the application.
By default, this application is configured to recognize and automatically

load static datasets by applying the filters Pre * stimulus and Post
* stimulus. For dynamic datasets, the filter is EF.
To facilitate manual loading of multiple static datasets for different views,
ensure that your exam names follow the above conventions. This way, the
application loads the selected datasets automatically and correctly.
Otherwise, you must load each dataset one by one.
9.2 Calculations for Gall Bladder

4598 001 91041 Rev A
9.2.1 Calculations for GBEF

The Raw Gallbladder curve and the Decay Corrected Raw curve are
presented with reference lines. Initially, the reference line points to the
maximum and minimum counts on the curve. If needed, these reference
lines can be adjusted to redefine the time range.
Ph ilips Healthcare
These maximum and minimum points are used in the calculation of the
GBEF:
GBEF = (100 * (Maximum counts - Minimum counts)) /
Maximum counts
where

Maximum Count = Reference Line’s maximum value
Minimum Count = Reference Line’s minimum value
Ejection Period = Maximum count time - Minimum count
time
Latent Period = CCK infusion start time - Maximum count
time
Ejection Rate = Ejection Fraction / Ejection Period
9.2.2 Calculations for Gallbladder Static Analysis

Calculations for Gallbladder Static Analysis are shown in the table below:
Counts Pre-Meal = Gallbladder counts - Background
counts
where
Gallbladder counts = Total pixel sum of the ROI covering the gallbladder
region
Background counts = Total pixel sum of the ROI covering the background
region
Counts Post Meal = (Gallbladder counts - BKG counts) *
e[-(0.693/361.2) * diff_time(min)]
Ejection Fraction = ((Counts Pre-Meal - Counts Post
Meal)/Counts Pre-Meal) * 100
9.2.3 Calculation for Static GBEF

The Static GBEF analysis is for the calculation of the percent emptied
from a pre-intervention CCK Gallbladder static and up to 12
post-intervention statics:
Percent Emptied = ((Pre-CCK counts - Post CCK counts) /
Pre-CCK counts) * 100
Ph ilips Healthcare

10 Astonish
Because Astonish uses specific information from your camera to model the
collimator performance, it expects acquisition-specific information in the
DICOM header for each image. Astonish supports only the following
cameras with the latest acquisition software:
• SKYLight version 3.1 or later
• Forte version 2.0 or later
• CardioMD version 2.0 or later
• Precedence (all versions)
If your Forte, SKYLight, or Forte camera does not have the acquisition
version specified, you must call your Field Service Engineer to have your
system upgraded before you can perform Astonish reconstruction.
For Astonish to work properly, make sure that the collimator is correctly
specified in the data. If you need to change the header information, you
can use the EBW Patient Browser to change it.
Astonish is only available as a licensed option within AutoSPECT Pro. To
verify that you have the appropriate license, click Preferences in the EBW
NM Patient Directory and in the Preferences dialog, select Licensing to
check your licensing options.
If you have the correct version of the acquisition and processing software
and the Astonish license, but Astonish is still not enabled on your system,
part of your header information may be corrupt. Call Customer Support
to have them examine your datasets to ensure that the correct information
is present.
4598 001 91041 Rev A
Ph ilips Healthcare

Ph ilips Healthcare

11 Filters
The filtering process is similar for static and dynamic images. Available
filters for static images are Laplacian, Spatial (5x5), and Spatial (3x3).
Additionally, the Spatial Temporal filter is available for dynamic images.
Filtering is accomplished by assigning each pixel in an image, one at a
time, a new count value. The new count value is determined by the
specific filter weighting arrangement applied to the image.
In other words, each pixel is multiplied by a number determined by the
pixel’s position relative to the center pixel. The products of these
multiplications are added together, but only the pixel in the center is
assigned the resulting “weighted average” count value. The filter is then
moved and the process repeated, until every pixel in the image has been
treated as the center and assigned a new, weighted average count value.
Filtering an 8-bit deep image will convert it to a 16-bit deep image.
11.1 Laplacian
The Laplacian filter is available for both static and dynamic images.
11.2 Spatial (5x5)

The Spatial-5x5 filter is available for both static and dynamic images.
4598 001 91041 Rev A
Ph ilips Healthcare

11.3 Spacial (3x3)
The Spatial (3x) filter is available for both static and dynamic images.
11.4 Spatial Temporal

The Spatial Temporal Filter is available only for dynamic images.
Ph ilips Healthcare

12 Keyboard Shortcuts
12.1 Keyboard Shortcuts for Applications
Shortcut Description
F4 Next intensity map
Shift + F4 Previous intensity map
F5 Next colormap
Shift + F5 Previous colormap
F9 Triangulation on/off
12.2 Keyboard Shortcuts for Viewers

Note: Many of the following shortcuts allow you to scroll using the
keyboard. However, in the following contexts you can scroll using only the
mouse (scroll wheel or left button), not the keyboard: the General Review
application, the AC Map workstep in Autospect PRO, and the Review
workstep in other applications.
ArrowDown Scroll to the next image in the vertical (Y) dimension
Ctrl + ArrowDown Add row to the bottom in a tiled layout
ArrowLeft Scroll to the previous image in the horizontal (X) dimension
Ctrl + ArrowLeft Remove column from the right in a tiled layout

4598 001 91041 Rev A
ArrowRight Scroll to the next image in the horizontal (X) dimension
Ctrl + ArrowRight Add column to the right in a tiled layout
ArrowUp Scroll to the previous image in the vertical (Y) dimension (same as
Right/Left for 2D viewer)
Ctrl + ArrowUp Remove row from the bottom in a tiled layout

Ph ilips Healthcare
End Go to the last image in the horizontal (X) dimension
Home Go to the first image in the horizontal (X) dimension
PageDown Scroll to the next page of images, if there is only one image dimenstion;
scroll to the next page of images in the vertical dimension, if there are
more than one image dimensions

Ctrl + PageDown Scroll to the next page of images in the horizontal dimension, if there
are more than one image dimensions
PageUp Scroll to the previous page of images, if there is only one image dimen-
stion; scroll to the previous page of images in the vertical dimension, if
there are more than one image dimensions
Ctrl + PageUp Scroll to the previous page of images in the horizontal dimension, if
there are more than one image dimensions
Pause Start/pause/resume movie
Ctrl + Tab Move focus cyclically forward through a tab container? tab pages (brings
the newly focused tab page to the front)
Ctrl + Shift + Tab Move focus cyclically backward through a tab container?tab pages
(brings the newly focused tab page to the front)
Ph ilips Healthcare

Philips Medical Systems is part © Koninklijke Philips Electronics N.V. 2012
All rights are reserved. Reproduction or transmission
of Royal Philips Electronics in whole or in part, in any form or by any means,

electronic, mechanical or otherwise, is prohibited
without the prior written consent of the copyright
owner.
www.medical.philips.com
[email protected]
0197
This Medical Device meets the provisions of the
Manufacturer’s address transposition of the Medical Device Directive
93/42/EEC within the country of origin of the
Philips Medical Systems Nederland B.V. Notified Body concerned with the device.
Veenpluis 4-6 Printed in The Netherlands.
5684 PC Best 4598 001 91041 A * 10/2012

The Netherlands

B NM App Suit Ref Manual ENG

Uploaded by

Copyright:

Available Formats

B NM App Suit Ref Manual ENG

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

B NM App Suit Ref Manual ENG

Uploaded by

Copyright:

Available Formats

English

Prescription Device Statement

NM Application Suite is CE Marked to the Medical Device Directive 93/42/EEC.

Philips Medical Systems Nederland B.V.

NM Application Suite Reference Manual

Document number 4598 001 91041 Rev A

3 Whole Body ........................................................................................................ 19

4.1.2 Washout Analysis....................................................................................... 21

5.2 Acquisition Parameters for Renal ............................................................................. 24

N M A p p l i c a t i o n S u i t e o n I n t el l i S p a c e Por ta l Version 5.0 iii

8 Gastro Intestinal ................................................................................................. 41

iv N M A p p l i c a t i o n S u i t e o n I n t el l i S p a c e Por t a l Version 5.0

12 Keyboard Shortcuts ............................................................................................ 51

N M A p p l i c a t i o n S u i t e o n I n t el l i S p a c e Por ta l Version 5.0 v

vi N M A p p l i c a t i o n S u i t e o n I n t el l i S p a c e Por t a l Version 5.0

This section lists the references for the Reconstruction section of

1.1 References for Reconstruction

Studies” THE JOURNAL OF NUCLEAR MEDICINE 45:5, 2004.

variant, object-dependent scatter response function in SPECT”.

NM Application Suite on IntelliSpace Portal Version 5.0 1

2 NM Application Suite on IntelliSpace Portal Version 5.0

21. Kluge R, Sattler B, Seese A, and Knapp WH. Attenuation correction

31. Toft J, Hesse B, and Rabol A. The occurrence of false-positive

Heller, MD PhD. Influence of arm positioning on rest/stress

NM Application Suite on IntelliSpace Portal Version 5.0 3

4 NM Application Suite on IntelliSpace Portal Version 5.0

This section describes the minimum requirements, calculations, and

2.1 Minimum Requirements for Cardiac

2.1.1 MUGA Processing

2.1.2 First Pass

2.2 Acquisition Parameters for Cardiac

2.2.1 First Pass

NM Application Suite on IntelliSpace Portal Version 5.0 5

2.3 Calculations for Cardiac

2.3.1 General MUGA Computations

The following table captures the results generated by MUGA:

PER = Min(d/dc(ROI)) / (Max(ROI) * frameTime / 1000)

6 NM Application Suite on IntelliSpace Portal Version 5.0

TPF = Time[Max(d/dc(ROI))] - Time[ES]

2.3.2 General First Pass Computations

2.3.3 General Shunt Computations

within the gamma curve. The gamma fit is interactively regenerated to

• Qp/Qs = A1 / (A1 - A2)

NM Application Suite on IntelliSpace Portal Version 5.0 7

2.3.4 GBP Multiple Algorithm

Spatial filtering by applying a filter in the frequency domain

The input images are filtered in the frequency domain, so an FFT is

Setting the Bounding Box

8 NM Application Suite on IntelliSpace Portal Version 5.0

Estimating the location of the bounding box

Estimating the size of the bounding box

Detecting Ventricle Edges

Filtering the images

• A hyperbolic tangent transform is applied on the resulting image.

NM Application Suite on IntelliSpace Portal Version 5.0 9

Images are resized to 128x128. All subsequent calculations on images are

Convolving each image with a second derivative filter

Figure 1 Filters for convolving

Determining the seed points