Basic Principles of Gamma Camera Imaging

and Quality Control

Sharon L. White, PhD

University of Alabama at Birmingham
July 16, 2015
 Disclosures

No financial disclosures.
Gamma camera images and photographs of
equipment are for illustrating concepts and not
intended to advertise or endorse any particular
manufacturer or vendor.

Page 2
 Learning Objectives

1. Understand basics of operation of conventional

gamma cameras.
2. List performance characteristics of gamma
cameras and features affecting performance.
3. List basic gamma camera calibrations and how
they affect performance.
4. List QC tests for gamma cameras required by
accrediting organizations.
5. Describe how to perform basic QC tests and
assess acceptable performance.
Page 3
Gamma Cameras –Dual Head

4
 Gamma Camera Operation
Array of Photomultiplier Tubes (PMTs):
Localizes the position where the gamma ray
interacts in the crystal

Sodium Iodide crystal:

A gamma ray from the patient interacts
and produces visible light photons

Collimator:
Forms a projection image by
allowing only gamma rays traveling
g in certain directions to reach crystal (for
g a parallel hole collimator, gamma rays
approximately perpendicular to crystal
pass through).
g
Gamma rays emitted from patient

5
 Conventional Gamma Cameras

 Two detectors (heads) most common, although

single head and triple head cameras are used
 Each head has single large NaI (sodium
iodide) crystal, up to 40 cm X 60 cm. Typical
crystal thickness: 3/8 or 5/8 inch
 Array of photomultiplier tubes, typically ~ 50
per head

Page 6
 Position Determination

 The point where the gamma ray hits the crystal

is determined by a weighted average of the
signals from the group of PMTs receiving light
from that event.
 The collimator localizes the origin of the
gamma ray as somewhere along a specific line
through the patient, since only gamma rays
traveling parallel to the holes will go through.
(Except for occasional septal penetration.)

Page 7
 Types of Imaging

 Static Planar
 Dynamic Planar
 Whole body
 Tomographic (SPECT)
 Not all gamma cameras do all types of imaging
– some do only planar, or only SPECT.

Page 8
 SPECT Operation

Camera heads rotate around patient, acquiring a set of

projection images that are reconstructed into slices 9
 Whole
Body
Bone
Scan

Static Planar

SPECT projection images 10

Dynamic
 Performance Characteristics

 Spatial Resolution
 Efficiency/Sensitivity
 Energy resolution

Page 11
 Spatial Resolution
• Intrinsic resolution (Rint) refers to how well the
crystal and PMT system localize an interaction
in the crystal. Affected by crystal thickness,
gamma ray energy, scatter in crystal.
• Collimator resolution (Rcoll) refers to how well
the collimator localizes the gamma ray source
in the patient, affected by hole diameter and
length, distance from collimator to patient.
• System resolution (Rsys) is a combination of
intrinsic and collimator resolution:

12
 Intrinsic Spatial Resolution

 Affected by statistical fluctuations in number of

light photons produced by scintillator.
 More light photons improves statistics, causing
less significant fluctuation in signal size and more
accurate positioning
 Intrinsic spatial resolution improves with
increasing gamma ray energy, up to ~ 250 keV.
 At higher energies scatter in the crystal becomes
more significant. Scatter can cause mispositioned
events, degrading resolution.

Page 13
 Intrinsic Spatial Resolution

 A thinner crystal has better intrinsic resolution

than a thicker one – less spreading of light
and multiple scatter events less likely to be
detected.
 Typical intrinsic resolution is 3.5 to 4.5 mm,
depending on crystal thickness
 Crystal thickness a tradeoff between spatial
resolution and efficiency – thinner crystals
have worse efficiency than thicker ones.

Page 14
 Bar pattern using Thallium, one
peak at a time

Lower energy peak only, 69 keV Upper energy peak only, 167 keV –
Better resolution at higher energy15
 Collimators

 Parallel hole collimators used most commonly

 Different collimators available for different
energy radionuclides – medium energy for
111In and 67Ga, high energy for 131I

 Different choices available for favoring high

resolution vs. high sensitivity
 Parallel hole collimator produces image same
size as object – no magnification or
minification.

Page 16
 Collimators and Scatter

 Gamma rays undergoing Compton scatter in

the patient can pass through collimator holes
as well as unscattered ones.
 A scattered photon has lower energy than the
initial photon. Scattered photons in the image
are reduced by energy discrimination, although
some scattered photons are still included when
their energy loss is small enough that they are
inside the allowed energy window.

Page 17
 Scatter in Patients

A scattered or Scattered photon

non-scattered may be accepted as
gamma may good event if energy
be emitted at within window.
such an angle to Results in mis-
be absorbed by positioned event
septa and not g
g
enter crystal

If scattered photon energy Scatter in patient.

sufficiently low, it will be rejected Scattered photon
by energy discrimination – it will be passes through
outside energy window. collimator hole18
 Parallel Hole Collimator
Resolution
d = hole diameter
L = hole length
X = distance from collimator
lead to source
septa,
thickness
t L Collimator Resolution

d
d x Rcoll  ( L  x)
L

19
At collimator surface 5 cm from surface

Collimator resolution gets

worse as source moves away
from collimator
surface. Important to position
patient as close as
possible to collimator
20
10 cm from surface
 Collimator Specifications
Type Hole Septal Hole Coll. System
Diameter Thickness Length Res. At Res at 10
(mm) (mm) (mm) 10 cm cm (mm)
(mm) 9.5 mm
crystal
LEGP 1.40 0.180 24.7 8.0 8.8

LEHR 1.40 0.152 32.8 6.3 7.4

MEGP 2.95 1.143 48.0 10.7 11.3

HEGP 3.81 1.727 60.0 12.0 12.5

HEHR 3.06 1.95 60.0 9.6 10.4

21
 Other collimator types

 Pinhole – forms magnified view of small

object, such as thyroid. Image is inverted.
 Diverging – produces minified image, for
imaging large object (e.g. lungs) on smaller
detector area. No longer common.
 Converging – produces non-inverted,
magnified view of small object. Not commonly
used.
 Fanbeam – hybrid of parallel hole and
converging, sometimes used in brain SPECT

Page 22
 Efficiency or Sensitivity

 Refers to fraction of emitted gamma rays

detected and used to form image
 Efficiency has intrinsic component based on
the thickness of the crystal and the attenuation
coefficient of the scintillation material (how
likely that a gamma ray is absorbed and
detected rather than just pass through)
 Thicker crystal will have higher efficiency, at a
cost of decreased spatial resolution.

Page 23
 Efficiency or Sensitivity

 System efficiency is a combination of intrinsic

efficiency and collimator efficiency.
 Collimator efficiency related to diameter and
length of holes, and thickness of septa.
 Tradeoff between collimator spatial resolution
and efficiency.

Page 24
 Efficiency or Sensitivity
• Parallel hole collimator efficiency
proportional to:

ædö
2
d 2

ç ÷ ´
è L ø (d + t) 2

d=hole diameter
L=hole length
t=septal thickness 25
 System Sensitivity

 System sensitivity relatively low, <≈ .02 %,

due to necessity of absorptive collimation.
 System sensitivity usually specified in
cpm/mCi at 10 cm for a specific radionuclide.
 Typical values on the order of 150-170
cpm/mCi for Tc-99m for a low energy high
resolution collimator.

Page 26
 Energy Resolution

• Good energy resolution important :

• scatter rejection
• separating multiple photopeaks
 Depends significantly on statistical
fluctuations in events in the imaging chain,
such as number of light photons produced in
scintillator, and number of photoelectrons
produced in PMT photocathode, although
other factors contribute

Page 27
 Energy Resolution

 Defined as FWHM of photopeak divided by

photopeak energy, expressed as percentage
 Since it is energy dependent, for a gamma
camera usually specified for Tc-99m, typically
9-10% for conventional gamma cameras.

Page 28
 Image Acquisition Options

 Matrix size (examples are 64 X 64, 128 X 128,

256 X 256, 512 X 512)
 Zoom factor (field of view)
 Combination of matrix size and zoom factor
determines pixel size. Pixel size affects
resolution and noise in image, as well as slice
thickness in SPECT
 Total counts and imaging time

Page 29
 Image Acquisition Options

 The following slides shows the effect of

different image acquisition options, such as
matrix size, zoom factor and total counts.
 Planar images of four quadrant bar pattern and
SPECT phantom standing on end are used to
illustrate these options.

Page 30
 SPECT Phantom
• Jaszczak Phantom
for SPECT quality
control.
• Approved by ACR
for SPECT ACR
accreditation images
• Standing on end,
used for evaluation
of planar spatial
resolution with
scatter – rod sizes:
12.7, 11.1, 9.5, 7.9,
31
6.4 and 4.8 mm
 Image Acquisition Options

 The following slides shows the effect of matrix

size options ranging from 64 X 64 to 512 X
512
 Total counts the same in each – 500K for
SPECT phantom and 5 M for bar pattern

Page 32
64 X 64 128 X 128

256 X 256 512 X 512

33
64X64 128 X 128
Pixel 8.8 mm Pixel 4.4 mm

256 X 256 512 X 512

Pixel 2.2 mm Pixel 1.1 mm

34
Bar spacings 2.0, 2.5, 3.0, 3.5 mm
 Image Acquisition Options

 The following slide shows zoom options.

 Matrix size 512 X 512 on each, but smaller
field of view used with Zoom 1.46 on second
one, resulting in smaller pixel size.

Page 35
512 X 512 matrix,
1 M counts

512 X 512 matrix,

Zoom 1.46, 1 M counts

36
 Image Acquisition Options

 The following slides show three images with

the same matrix size, 512 X 512
 Total counts different in each
 Counts per pixel higher with higher total
counts, causing images to be less noisy,
affecting visibility of rods or bars.

Page 37
500K 1M

All 512 X 512 Matrix

Total counts varies
2M

38
1.25M 5M

20M 512 X 512 matrix

Total counts varies
From 1.25 M to 20 M

39
 Gamma Camera Calibrations

 PMT gains must be balanced

 Correction Tables:
Energy
Linearity
Uniformity (Flood)
 Center of Rotation (COR) offset calibration for
SPECT-capable cameras.

Page 40
 Correction Tables

 Energy correction table corrects for variations

in measured energy across the detector
 Linearity table corrects for non-uniform light
collection efficiency across face of
photomultiplier tubes, which causes straight
lines to appear wavy without correction
 Flood table corrects for remaining non-
uniformities

Page 41
 Effects of Correction Tables

No Energy
corrections And
Linearity

Energy,
Energy Linearity,
only Uniformity
(all corrections)

42
99mTc Intrinsic Flood Images
 Center of Rotation
Pixel number Offsets between physical center of
recording signal
rotation and center of image matrix must
128x128 matrix
be corrected for.
1 63

COR error

Source

65 1 Image
Image acquisition Backprojection Matrix
43
 What is the primary function of a
collimator in a gamma camera?
A. Protect the crystal 96%

B. Define direction of
incoming gamma rays
entering crystal
C. Substantially eliminate
scatter
D. Reduce count rate to
prevent dead time 0%
4%
0% 0%

E. Shield the electronics A. B. C. D. E.

44
What is the primary function of a
collimator in a gamma camera
A. Protect the crystal
B. Define direction of incoming gamma rays
entering crystal
C. Substantially eliminate scatter
D. Reduce count rate to prevent dead time
E. Shield the electronics

Reference: The Essential Physics of Medical

Imaging, JT Bushberg, JA Seibert, EM Leidholdt
Jr, JM Boone, 3rd edition, 2012, 680-681. 45
 Which would improve spatial
resolution in gamma camera images?
A. Choose a camera with a 88%

thicker crystal
B. Use a 64X64 matrix
rather than 256X256
C. Image with lowest energy
gamma rays available
D. Position patient as close 8%
4%
as possible to collimator 0%
face A. B. C. D.
46
 Which would improve spatial resolution in
gamma camera images?
A. Choose a camera with a thicker crystal
B. Use 64 X 64 matrix rather than 256 X 256
C. Image with lowest energy gamma ray available
D. Position patient as close to collimator face
as possible
Reference: The Essential Physics of Medical
Imaging, JT Bushberg, JA Seibert, EM Leidholdt
Jr, JM Boone, 3rd edition, 2012, 686-691.
Physics in Nuclear Medicine, SR Cherry, JA
Sorenson, ME Phelps, 4th edition, 2012,365-366.
47
 Quality Assurance

 Routine QC tests are performed daily and

weekly, typically by technologists
 Physicists should perform annual
assessments.
 Accrediting bodies provide standards for
annual tests

Page 48
 Joint Commission Requirements

 Effective July 1, 2015, at least annually,

assess:
Image uniformity/system uniformity
High contrast resolution/system spatial
resolution
Sensitivity
Energy Resolution
Count Rate Performance
Artifact evaluation

Page 49
 American College of Radiology Accreditation

 At least annually:
Intrinsic uniformity
System uniformity
Intrinsic or System spatial resolution
Relative sensitivity
Energy Resolution
Count Rate Parameters
System performance for SPECT: tomographic
uniformity, contrast and spatial resolution

Page 50
 Intersocietal Accreditation Commission
(IAC) Recommendations

 Guidelines for annual tests include:

Overall system performance may be
evaluated with a fillable phantom with cold
inserts of different sizes and visually inspect
resulting images
Collimator integrity, comparing intrinsic and
extrinsic floods, should be performed as well
as visual inspection of collimators

Page 51
 Routine QC - Uniformity

 Uniformity must be checked every day that

gamma camera is used, before the first patient
 Uniformity (flood) image may be acquired with
collimator on for system (extrinsic) uniformity or
collimator off for intrinsic uniformity
 5 million counts adequate for daily QC for large
FOV camera, use 256 X 256 or 512 x 512 matrix
(manufacturer may have specific
recommendations)

Page 52
 System Uniformity
With collimator on, use planar sheet source:

57Co sheet source Water filled sheet source

10-15 mCi when new Add 10-15 mCi 99mTc
122 keV g 140 keV g
Half life 270 days Half life 6 hours
53
 Intrinsic Uniformity

 General method – use ~ 500 mCi 99mTc point

source, placed at a distance of five times the
length of the camera field of view
 Some cameras have a special source holder
and vendor specific procedure which allows the
source to be closer

Page 54
 Uniformity- Annual Testing

 Acquire intrinsic uniformity images with Tc-

99m at low and high count rates – often the
daily QC is only done with Co-57.
 Low count rate, typically 20-40 kcps, high
count rate, 65-80 kcps, but refer to
manufacturer’s recommendations.
 The high count rate acquisition provides
assessment of camera’s function at higher
count rates – modern cameras should still
have good uniformity.

Page 55
Good uniformity images

56
Poor uniformity

57
 Uniformity - Quantification

Integral Uniformity should be < 5% for 5M count extrinsic flood

for camera following NEMA method for calculation. Refer to vendor
58
specifications.
59
60
 Spatial Resolution and Linearity
 Routine QC - Image bar pattern at least weekly,
extrinsically or intrinsically, to check spatial
resolution and linearity
Confirm that smallest resolvable bar pattern
remains the same with no abrupt changes
Ensure that bars do not appear significantly
wavy, and no abrupt change in appearance
 2.5 M counts adequate for weekly QC
 Annual test do intrinsic bars. 5 M counts is
required for ACR accreditation submission. Be
very careful using bar pattern with collimator off.

Page 61
Spatial Resolution – Four Quadrant
Bar Pattern

62
Intrinsic Bars – Linearity Correction
Off

63
 Extrinsic Resolution (FWHM) with Line
Source

 Image a thin line source (plastic or glass tube)

filled with 99mTc, (at least ~ 1mCi/ml) 10 cm
from collimator
 Use matrix size so that pixel size less than
about 1/5 expected FWHM
 Draw profile across image to produce curve of
counts vs. pixel
 Determine FWHM with available software, or
other means

Page 64
 Line source profile and curve

FWHM= 7 pixels =
7.7 mm

65
 System Resolution with Scatter

Tc-99m Tl-201
Static images of SPECT phantom standing on end on top of
collimator. Provides a measure of planar system resolution with
66
scatter.
 Measuring Sensitivity
– Place ~1-2 mCi 99mTc,
and small volume of
water in plastic flat-
bottomed vial on top of
Styrofoam cup 10 cm
from collimator face.
– Record exact activity
and time
– Count for 1 min, also
count and subtract
background
– Use total counts in
image, not an ROI
drawn around image 67
 Measuring Sensitivity

• Compute cpm/mCi and compare with vendor

specifications, also check that both heads
have comparable sensitivity (within about 5%)
• If camera is off peak it will affect results, also
ensure window width is same as
manufacturer’s specification
• A syringe will give comparable results, use a
small volume spread out through syringe
rather than a tiny point source

Page 68
Energy Resolution Measurement

69
 Estimating Energy Resolution

• Estimate ~ 9% energy resolution – photopeak width is

approximate width of 9% window at half the peak height70
 Count Rate Performance
 Measure maximum count rate with point
source and collimator off. Approach camera
with point source and observe count rate go
up, until close enough that it decreases due to
dead time. Note maximum count rate (a good
estimate). Quickly move source away.
 Acquire uniformity images at high count rate
and ensure uniformity is still reasonable (65-80
kcps for most newer cameras, older cameras
may have performance degradation at lower
count rates)

Page 71
 Low to High Count Rate Intrinsic
Floods

19 kcps 79 kcps

109 kcps
(too high
without high
count rate
mode
corrections)
72
 SPECT Image Quality

 Acquire images with SPECT phantom to

evaluate contrast, resolution and uniformity,
including artifact evaluation

Page 73
 SPECT Phantom Imaging

• Deluxe version has

spheres of
diameters: 31.8,
25.4, 19.1, 15.9,
12.7, 9.5 mm
• Rods of diameters:
12.7, 11.1, 9.5, 7.9,
6.4 and 4.8 mm

74
 SPECT phantom imaging procedure

 Make sure largest sphere lined up with largest

rod section (rotate if needed)
 Fill phantom with ~20-25 mCi 99mTc for high res
collimator. Count rate should be < 30kcps
 Use 99mTc sodium pertechnetate. Some
radiopharmaceuticals may stick to the plastic or
nylon screws and cause artifacts

Page 75
 SPECT phantom imaging procedure

 Center phantom in field of view

 For cardiac cameras with 180 deg orbit, align
largest sphere and rod section with center of
leading detector for first frame.
 ACR protocol is for 32 M total counts. Check
count rate, adjust time per stop to achieve this.

Page 76
 SPECT phantom imaging procedure

 Use 128 X 128 matrix, 120 or 128 views over 360

degrees (180 degrees for a cardiac camera that
cannot do 360 degree rotation) Adjust zoom factor
as needed to achieve pixel size close to 3 mm.
(ACR says 2.7 to 3.3 mm)1.33 to 1.46 are
common zoom factors for large FOV camera
 Use a radius of rotation as close to 20 cm as
possible (an elliptical orbit is helpful)
 Apply attenuation correction during image
reconstruction.

Page 77
SPECT phantom reconstructed
slices

78
SPECT phantom reconstructed slices –
no attenuation correction

79
 SPECT phantom image quality

 Phantom images visually inspected for:

 Resolution – smallest size of rods visible
 Contrast – number of spheres visible
 Uniformity – look for ring type artifacts or other
artifacts

 For guidelines on acceptable image quality, refer

to ACR website for accreditation scoring criteria.
Criteria vary according to type of collimator and
radionuclide used (although currently it is only
required to submit SPECT images for Tc-99m)

Page 80
Ring Artifacts

Ring artifacts
visible

81
Severe Ring Artifacts

82
 SPECT Ring Artifacts

 Caused by non-uniformities such as:

 Visible non-uniformities in flood image due to
camera being off peak, PMT gain imbalance, or
need for new correction tables
 Shift in photopeak as camera head rotates
 Collimator defect or damage (not visible in intrinsic
flood image)

 Even small non-uniformities can cause ring

artifacts

Page 83
Phantom filled with 99mTc Sestamibi
rather than Sodium Pertechnetate

Artifact

84
 Which of these is not specifically listed as an
annual physics test required by TJC or
ACR?
13% A. Sensitivity
74% B. Center of Rotation
0% C. Uniformity
13% D. Energy Resolution

85
Which of these is not specifically listed as an
annual physics test required by TJC or
ACR?
A. Sensitivity
B. Center of Rotation
C. Uniformity
D. Energy Resolution
References: The Joint Commission Revised Requirements
for Diagnostic Imaging Services,
http://www.jointcommission.org/assets/1/6/HAP-
CAH_DiagImag_Prepub_July2015release_20150105.pdf
ACR Nuclear Medicine Accreditation Program
Requirements, http://www.acr.org/Quality-
Safety/Accreditation/Nuclear-Med-PET
86
 What is the primary cause of ring
artifacts in SPECT phantom images?

86% A. Non-uniformities
14% B. Center of Rotation error
0% C. Phantom off center in field of view
0% D. Using the wrong matrix size

87
What is the primary cause of ring artifacts in
SPECT phantom images?
A. Non-uniformities
B. Center of Rotation error
C. Phantom off center in field of view
D. Using the wrong matrix size

Reference: The Essential Physics of Medical

Imaging, JT Bushberg, JA Seibert, EM Leidholdt
Jr, JM Boone, 3rd edition, 2012, 718-719.

88
 Imaging and recording counts for a known
amount of activity in a small flask for 1 min is
a method of measuring _____?

13% A. Uniformity
4% B. Spatial Resolution
79% C. Sensitivity
4% D. Energy Resolution

89
 Imaging and recording counts for a known
amount of activity in a small flask for 1 min is
a method of measuring _____?
A. Uniformity
B. Spatial Resolution
C. Sensitivity
D. Energy Resolution

Reference: AAPM Virtual Library, 2012 Spring Clinical

Meeting, talk by James Halama: Nuclear Medicine-Testing
of Gamma Camera, SPECT and SPECT/CT Systems in a
Clinical Environment,
http://www.aapm.org/education/VL/vl.asp?id=125 90
The End

Page 91