where T,S, and Rare the true, scatter, and random coincidences, respec-
tively. It serves as a parameter to compare the performances of different
PET scanners. Image noise can be minimized by maximizing NECR.
Quality Control Tests for PET Scanners
Daily Tests
Sinogram Check
This is normally carried out by using a standard 20-cm long cylindrical
phantom containing a positron emitter (normally a 1 to 3 mCi [37–111
MBq]^68 Ge source). It is mounted on the patient table and centered both
vertically and horizontally in the field of view. This arrangement allows
uniform exposure to radiations of all detectors to provide a uniform sino-
gram. Initially, a reference blank sinogram is obtained during the last setup
of the scanner. Subsequently, a blank sinogram is taken daily before the
patient study is started on each day. This daily sinogram is then compared
with the initial blank sinogram. The difference between the two sinograms
is characterized by a quantity called average variance, which is calculated
by the square sum of the differences of the relative detector efficiencies
between the two scans weighted by the inverse variances of the differences.
To obtain the average variance, the sum is then divided by the number of
detectors. When this value exceeds 2.5, recalibration is needed, whereas
for a value greater than 5.0, manufacturer’s service is warranted (Buchert
et al., 1999). Normally, in modern PET cameras, this is carried out by the
manufacturer’s menu-driven software.
Weekly Tests
Normalization
As mentioned before, normalization corrects for nonuniformities in the
acquired PET data. It is accomplished by using a standard 20-cm long cylin-
der containing 1 to 3 mCi (37 to 111 MBq)^68 Ge activity accurately posi-
tioned at the center of the FOV. All detectors are uniformly exposed to
radiations in the absence of any object in the FOV. From the acquired 2-D
or 3-D data, the correction factors are calculated for each detector by divid-
ing the average counts of all detector pairs by each individual detector pair
count [i.e., along the LOR; (Eq. 13.3)]. These factors are saved and later
applied to corresponding detector pairs in the acquired emission data of
the patient (Eq. 13.4). Usually, normalization factors need to be taken
weekly or monthly, although some manufacturers recommend quarterly.
For better statistical accuracy, long hours of counting (6 to 8 hrs) are nec-
essary, and often, overnight data acquisition is made. Data acquisition and
Quality Control Tests for PET Scanners 205