Добавил:
Sekretar
kiopkiopkiop18@yandex.ru
t.me/Prokururor I Вовсе не секретарь, но почту проверяю
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз:
Предмет:
Файл:Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_5545_Библиотеки_им_академика_М_И_Перельмана.pdf
144
C
AB
B
10 Performance ofGamma Camera
of the camera. Recent developments include high-speed electronics that reduce the
number of misplaced events and improve the image quality signicantly.
10.4 Contrast
Contrast of an image is the relative differences in count densities between adjacent
areas in the image corresponding to the activity distribution in an object. Contrast
(C0) gives a measure of detectability of an area, for example, abnormal tissue relative to the surrounding tissue and is expressed as
0
(10.9)
where A and B are the count densities recorded over the abnormal tissue and surrounding tissue, respectively. The count density B may be from normal tissues or
background counts. It is understood from Eq. (10.9) that the greater the difference
between the abnormal and the surrounding tissues, the better is the image contrast.
Also it can be seen that if additional background is added throughout the image,
contrast is reduced, since the numerator remains the same, while the denominator is
increased. Lesions on the image are seen as either “hot” or “cold” spots with
increased or decreased uptakes of radioactivity in the object meaning that C0 can be
positive or negative and therefore is expressed in absolute values.
Several factors affect the contrast of the image, namely, count density, image
noise due to background activity, size of the lesion, and patient motion, and each
contributes to the contrast to a varying degree and is briey discussed below.
10.4.1 Count Density
Count or information density is an important factor to improve the image contrast
and a minimum number of counts need to be collected for reasonable contrast in the
image. The higher the counts in an image, the better is the contrast. Count density
depends on the amount of activity administered and the uptake in the organ of interest. Contrast is improved with increasing administered activity and also with the
differential uptake between the area of interest and surrounding tissues. The optimum count density is considered to be about 1000counts/cm2.
10.4.2 Image Noise
Noise is an integral part of images in nuclear medicine impairing the detectablity of
lesions and has two forms: systematic or random. Systematic noise pertains to a
xed amount of variation in count rate all across the image distorting the image
contrast. This may be caused by system artifacts or may arise from the radioactivity
distribution in different overlapping organs such as bladder activity adding the noise
B
1
100
C
C
C
AB
B
B
AB
B
1
10.4 Contrast
145
to the pelvic bone image with
99m
TcMDP.Nonuniformities in the gamma camera
and tomographic reconstruction of images also result in systematic noise.
Random noise, also called quantum noise, is a major component of image contrast caused by statistical variations of the background counts in and around the
image and is given by the standard deviation of the background count B. Thus,
Noise = B
We can express it as a fraction of the total background counts contributed to
image contrast, which is called the contrast noise, CN. Thus,
B
C
= =
N
1
B
B
(10.10)
The percent contribution of noise is
C
N
(10.11)
With less number of counts, the percent standard deviation is increased compared to higher counts and hence noise is increased degrading the image contrast.
Noise originates from random variations of the background count density that is
composed of scattered radiations from the patient and the detector, and septal penetration of photons in the collimator. All these counts add to the noise in the image
and worsen the contrast. Narrow pulse-height analyzer reduces the scattered radiations, but it also reduces the counting efciency, that is, sensitivity of the camera.
This in turn increases the percent standard deviation of counts due to low counts
(Eq. 10.11) and hence the noise. An optimum 15–20% window centered on the
photopeak of interest is customarily used in routine imaging. Septal penetration is
mitigated by the choice of appropriate collimators.
Although it is conceptually understandable that reducing noise or background
counts minimizes the image contrast, a more appropriate parameter to consider for
detectibility of a lesion is the contrast-to-noise ratio (CNR), which is given by
NRN=
0
C
(10.12)
Combining Eqs. (10.9) and (10.10),
NR
(10.13)
Cherry etal. (2012) have given a detailed discussion of this parameter and its
effect on the image contrast. The larger the CNR, the better is the contrast. It is
reported that a lesion should have a CNR of >3–5in order to be detected. For a given
imaging setting, a minimum number of counts needs to be collected for reasonable
image contrast. Even with adequate spatial resolution of an imaging device, lack of
sufcient counts may give rise to poor contrast due to increased noise such that the
lesion may be missed. It can be seen from Eq. (10.13), increasing the amount of
146
administered activity or counting for a longer period of time in the case of low
administered activity increases the count density and in turn, improves the CNR for
the area of interest. However, due consideration should be given to the radiation
dose to the patient from large amounts of administered activity.
Noise is more problematic in SPECT than in planar imaging, because in SPECT,
counts per pixel are less than in planar imaging and also reconstruction in SPECT
amplies the noise. Reconstruction lter has a signicant effect on noise and an
appropriate lter should be chosen in SPECT.However, in planar imaging, overlapping structures cause more contrast degradation than in SPECT imaging, because in
the latter, only images of slices are collected excluding the interfering activities
from adjacent slices.
10 Performance ofGamma Camera
10.4.3 Choice ofMatrix
Noise is also affected by the choice of a matrix and so is the contrast. For example,
for a given time of data acquisition from a source, the count density per pixel will
be lower in a 128×128 matrix than in a 64×64 matrix. This will cause an increase
in noise and hence a loss of contrast in a matrix of smaller pixels, which can be mitigated by counting for a longer time. This effect is more prominent in SPECT imaging because of the low count density per projection and hence per pixel, relative to
planar imaging. So the choice of an acquisition matrix of appropriate size is
essential.
10.4.4 High Count Rate
At high count rates, pulse pileup can degrade the image contrast. Two Compton
events occurring simultaneously may add up to form the photopeak, but the event
will be mispositioned somewhere between the two events and hence the distortion
of the image.
10.4.5 Size ofLesion
Image contrast to distinguish a lesion depends on its size relative to system resolution and its surrounding background. Unless a minimum size of a lesion larger than
system resolution develops, contrast may not be sufcient to appreciate the lesion
even at higher count density. The lesion size factor depends on the background
activity surrounding it and whether it is a “cold” or “hot” lesion. A relatively small
size “hot” lesion can be well contrasted against a lower background, whereas a
small “cold” lesion may be missed against surrounding tissues of increased
activities.
10.5 Quality Control Tests forGamma Camera
147
10.4.6 Patient Motion
Patient motion during imaging reduces the image contrast. This primarily results
from the overlapping of adjacent areas by the movement of different organs. It is
somewhat alleviated by restraining the patients or having them in a comfortable
position.
10.5 Quality Control Tests forGamma Camera
To ensure high quality of images produced by imaging devices, several quality control tests must be performed routinely on gamma cameras. The frequency of tests is
daily, weekly, and, for some tests, monthly, quarterly, or even annually. The most
common tests are the positioning of the photopeak (peaking), uniformity, spatial
resolution, and linearity of the camera. These tests can be carried out with the collimator attached to the camera (extrinsic) or without the collimator (intrinsic), and
should be performed for each radionuclide used in a specic clinical study. Various
quality control tests and their frequency are given in Table10.3.
In the intrinsic method, the source of a particular radionuclide containing approximately 100–200μCi (3.7–7.4MBq) in a syringe is normally placed at a distance of
four to ve times the detector eld of view to ensure uniform irradiation of the
detector (Fig.10.11a). Because the collimator is removed, the integrity of the collimator cannot be assessed by this method.
In the extrinsic method, a sheet source made of plastic containing the radionuclide
of interest is used. Because
99m
a
Tc sheet source can be prepared by adding several millicuries of
a water-lled plastic sheet container. The source should be thoroughly mixed and free
of air bubbles. Because of the inconvenience of daily preparation of the
source and radiation exposure to the technologist during preparation, an alternative
57
Co sheet source is used, which is commercially available in rectangular or
solid
circular forms. 57Co has a longer half-life (∼270d) and emits photons of 122keV and
136keV, which are equivalent to the 140keV photons of
cally made with 10mCi (370MBq) of 57Co, are also called ood sources and most
99m
Tc is most commonly used in nuclear medicine studies,
99m
Tc. These sources, typi-
99m
Tc activity to
99m
Tc sheet
Table 10.3 Recommended
quality control tests and their
frequency
Test Frequency
Peaking of photopeak Daily
Intrinsic or extrinsic uniformity Daily
Spatial resolution and linearity Weekly
Center of rotation (see SPECT
section later)
High-count uniformity calibration Monthly
Collimator integrity Semiannually
Overall system performance (see
SPECT)
Monthly
Semiannually
148
ab
Fig. 10.11 Arrangement
of detector, collimator and
57
Co ood source. (a)
Intrinsic method without
collimator. (b) Extrinsic
method with collimator
10 Performance ofGamma Camera
commonly used for over a period of one to two years. The source is placed on the
collimator attached to the detector (Fig.10.11b) and an image is taken. The use of
these sources in the extrinsic method provides information on PM tubes as well as
any structural imperfections in the collimator. Because 57Co activity decays over
time, counting time increases with time to accumulate sufcient counts for the image.
An additional problem with 57Co source is the contamination with small amounts of
56
Co and 58Co (t
=70–80days) emitting high-energy γ rays (>500 keV). This con-
1/2
tamination is more prominent in the new sources, because of the short half-life but
fades away over time and can be mitigated by using a medium- energy collimator.
10.5.1 Daily Checks
10.5.1.1 Positioning ofPhotopeak
Positioning, often called peaking, of the photopeak must be done daily or as needed
for each radionuclide used in the clinical study to center the PHA window on the
center of the photopeak. In older analog cameras, a source of radioactivity of interest is placed on the collimator attached to the detector (extrinsic) and the high voltage on the PM tube is adjusted to center the energy window on the photopeak. For
99m
Tc, typically 1mCi (37MBq) of the activity in a syringe is used as a source for
peaking and a 20% window is set around 140keV.Peaking for
and so on must be done separately, as needed.
In modern cameras, peaking is performed automatically by menu-driven
protocol- based software provided by the manufacturer. Initially at the time of the
camera set-up, the photopeak window is set with a
method. Subsequently the daily check of the position of the photopeak is performed
with a 57Co ood source by the extrinsic method using a low-energy high-resolution
collimator. If the peak shift is more than 10%, the camera must be tuned. Tuning is
performed by the computer program by repeaking of the camera with a
placed at least 30cm away from the detector and without a collimator (intrinsic
method). The same method is applied for other radionuclides.
111
In, 67Ga,
99m
Tc source using the intrinsic
123
I,
99m
Tc source
201
Tl,
10.5 Quality Control Tests forGamma Camera
149
10.5.1.2 Uniformity
The uniformity of the detector response is checked daily by either the intrinsic
method using a ∼100–200μCi (3.7–7.4MBq)
99m
Tc source or the extrinsic method
using a 10mCi (370 MBq) 57Co ood source, although for convenience the latter
method is routinely employed. The ood source is placed on the low-energy highresolution collimator attached to the detector and an energy window of 20% is set
around the photopeak. An image is acquired with about ve million counts and then
assessed for uniformity and linearity (tube pattern), integrity of the PM tubes, artifacts and so on by visual inspection (Fig.10.12). Nonuniformity exceeding 5–10%
is detectable by human eyes. Integral uniformity and differential uniformity are then
calculated (see below) for both the useful eld of view (UFOV) and central eld of
view (CFOV). The UFOV is the area of the eld of view obtained after masking the
edge of the detector to minimize edge packing effect, and the CFOV is the area
whose linear dimensions (length×width for rectangular detector, or diameter for
circular detector) are reduced to 75% of those of the UFOV, as illustrated in
Fig.10.13.
According to the NEMA protocol, the intrinsic method is employed to calculate
the integral and differential uniformities using a 100–200μCi (3.7–7.4MBq)
99m
Tc
source. Before calculating these parameters, the acquisition matrix is adjusted so
a
b
c
Fig. 10.12 Images of a 57Co ood source. (a) Intrinsic uniformity image showing less activity
distribution at the periphery due to greater distance traveled by the photons. (b) Extrinsic uniformity image. (c) Nonuniform extrinsic image showing a defective PM tube
150
IU
CC
maxmin
100
DU
high count lowcount
100
Fig. 10.13 Denation of
UFOV and CFOV
10 Performance ofGamma Camera
that the pixel size is about 6±1.9mm and then smoothed with a ninepoint lter.
Integral uniformity (IU) is dened by
where C
max
and C
maxmin
CC
are the maximum and minimum pixel counts across the image.
min
(10.14)
Differential uniformity (DU) is given by
high count lowcount
(10.15)
where “high” and “low” are the maximum and minimum differences in counts over
ve contiguous pixels in all rows and columns of the matrix. The computer program
calculates both IU and DU values, which should be less than 4–6%; otherwise, the
camera needs to be tuned. Note that the IU is a global parameter, whereas the DU is
a regional parameter indicating the nonuniformities in pixel counts.
10.5.2 Weekly Checks
10.5.2.1 Spatial Resolution andLinearity
The spatial resolution and linearity of the gamma camera is checked weekly by
using a bar phantom (Fig.10.3). The bar phantom is placed on the detector head
with a low-energy high-resolution collimator attached (extrinsic), and a ood source
of ∼10mCi (370MBq) of 57Co is placed on the top of the bar phantom. An image
is taken with approximately ten million counts and visually inspected to check the
linearity and separation of the smallest bars (Fig.10.4). This is a qualitative method.
For a quantitative method, a line spread function must be determined and then
FWHM measured as discussed earlier. Although extrinsic tests are done for convenience, intrinsic tests are preferable for better accuracy.
10.6 Questions
151
10.5.3 Monthly Checks
10.5.3.1 High-Count Uniformity Calibration
A high-count uniformity ood image (30 million counts) is acquired extrinsically
using a 10mCi 57Co ood source or intrinsically using 100μCi (3.7MBq)
99m
Tc. It
is performed monthly or quarterly as recommended by the manufacturer. A very
high-count uniformity calibration with 100 million counts is carried out annually.
These high-count ood images are stored and used for nonuniformity correction, as
needed, and should be done with each radionuclide used and each collimator
employed. Because of the longer time required for high counts to be acquired, data
are often collected overnight.
10.5.3.2 Collimator Integrity
Although collimators made of lead are heavy and robust, they are prone to dent and
damage due to the ductile nature of lead. Any dent or damage on the collimator will
appear as an artifact on the image taken using it. Collimator is normally encased in
a frame and there should be no gap between the frame and the collimator. Although
most collimator defects can be seen on extrinsic images, all collimators should be
checked by visual inspection at least monthly for their structural integrity as to the
hole parameters, bolts, frames, and proper t to the detector head.
10.5.4 Annual, Semiannual, or As-Needed Checks
Tuning of the camera is performed monthly or quarterly by the protocols described
earlier. Other essential parameters such as energy resolution, high count rate
response, multiwindow registration (e.g.,
111
In and 67Ga with multiple photons) and
sensitivity should be evaluated at least annually or as needed after extensive modication or repair of the camera.
In addition, tests on accessories such as computers, multiformat cameras, scanning tables, rotation of gantry, and so on should be performed periodically.
Furthermore, all tests must be documented in a record book with pertinent information, such as the date, time, total counts, window settings, the type of radioactive
source, the type of camera, and initials of the technologist performing the tests.
10.6 Questions
1. (a) Dene the spatial resolution of a gamma camera.
(b) What are the different components of the spatial resolution?
(c) A system with a spatial resolution of 5mm is better than a system with a
spatial resolution of 8mm. True or false?
152
10 Performance ofGamma Camera
2. The intrinsic resolution of a gamma camera depends on:
(a) The thickness of the NaI(Tl) detector. True or false?
(b) The energy of the γ-ray. True or false?
(c) The width of the pulse-height window. True or false?
(d) The number of counts collected. True or false?
3. (a) What is the best photon energy for imaging with a gamma camera?
(b) Why is a thinner NaI(Tl) detector used in a gamma camera?
(c) Intrinsic resolution improves with higher γ-ray energy. True or false?
(d) Spatial resolution of a gamma camera improves as the number of photo-
multiplier tubes is increased. True or false?
4. For a gamma camera with a parallel-hole collimator,
(a) The spatial resolution increases with decreasing detector thickness. True
or false?
(b) The collimator efciency decreases with increasing collimator length. True
or false?
(c) The spatial resolution increases with decreasing collimator length. True
or false?
(d) High-energy collimators have higher efciency and resolution than low-
energy collimators. True or false?
(e) The best resolution is obtained at the face of the parallel-hole collimator.
True or false?
5. What are the effects of the following factors on the spatial resolution and sensi-
tivity of a gamma camera?
(a) Photomultiplier (PM) tubes with higher quantum efciency
(b) A wider “window” on the pulse-height analyzer (PHA)
(c) Increasing the activity of
99m
Tc from 5mCi (185MBq) to 15mCi (555MBq)
(d) Increasing the diameter of the collimator hole
(e) Adding more tissue between the collimator face and the patient’s organ
(f) Using a diverging collimator
(g) Increasing the source-to-collimator distance for a parallel-hole collimator
(h) Using a γ-ray of higher energy, which penetrates the septum of the
collimator
6. (a) In routine practice, how is the spatial resolution of a gamma camera
checked?
(b) The full width at half maximum of the line spread function of a gamma
camera does not give a true picture of spatial resolution. Why?
(c) What is the modulation transfer function (MTF) of a system?
(d) A system gives the best spatial resolution when its MTF is equal to 1. True
or false?
(e) If PM tubes and the PHA of a gamma camera have MTFs of 0.5 and 0.7 at
a certain spatial frequency, what is the overall MTF of the camera?
(f) As the sensitivity of a gamma camera increases, its spatial resolution
decreases. True or false?
References and Suggested Readings
153
(g) The collimator efciency of a parallel-hole collimator is not affected by the
source-to-detector distance. True or false?
7. (a) What is the primary cause of nonuniformity in an image?
(b) What is edge packing?
(c) How is the nonuniformity in an image corrected?
8. (a) What is the contrast of an image?
(b) What are the different factors that affect the contrast of an image?
(c) How does pulse pileup affect the contrast?
(d) Is contrast or spatial resolution affected by increasing the administered
activity?
9. (a) What are the daily and weekly tests performed for gamma cameras?
(b) What is meant by extrinsic and intrinsic tests?
10. If count density in a lesion is 77,000 and the adjacent background is 35,000,
calculate
(a) The contrast
(b) The noise, and
(c) The contrast-to-noise ratio
References and Suggested Readings
Bushberg JT, Seibert JA, Leidholdt EM Jr, Boone JM. The Essential Physics of Medical Imaging.
3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2011.
Cherry SR, Sorensen JA, Phelps ME. Physics in Nuclear Medicine. 4
W.B.Saunders; 2012.
Erickson J.Imaging systems. In: Harbert J, da Rocha AFG, ed. Textbook of Nuclear Medicine.
Volume I.Basic Science. Philadelphia: Lea & Febiger; 1984:105.
Groch MW, Erwin WD. Single-photon emission computed tomography in the year 2001:
Instrumentation and quality control. J Nucl Med Technol. 200; 29: 12.
Murphy PH.Acceptance testing and quality control of gamma cameras, including SPECT. J Nucl
Med 1987; 28:1221.
Rollo FD.Evaluating imaging devices. In: Rollo FD, ed. Nuclear Physics, Instrumentation and
Agents. St. Louis: Mosby; 1977:436.
Rollo FD, Harris CC. Factors affecting image formation. In: Rollo FD, ed. Nuclear Physics,
Instrumentation and Agents. St. Louis: Mosby; 1977:387.
th
ed. Philadelphia: