Page 50 - ARNM-2-2

Advances in Radiotherapy
& Nuclear Medicine PET and Compton Camera CZT based system

detectors by implementing a single-layer CZT design to atomic number (Z), atomic weight (A), ionic charge (Q),
capture scattering and absorption interactions. This is and excitation energy (E). This configuration enables the
made possible by our detectors’ ability to resolve multiple simulation of radioactive decay and atomic de-excitation
photon interaction events (MIPEs) in the cross-strip design physics, enabling the emission of positrons and their
of CZT crystals, in which algorithms have been developed kinematics, as well as positron annihilation. In addition, all
to pair separately, detected intra-crystal scattered MIPE. electromagnetic interaction physics involving annihilation
31
Thus, simplifying the problem of implementing a separate photons and prompt gammas with the CZT detector
or combined imaging modality of PET and CC in software crystals are specified using the em_standard_opt4 physics
components. Users can then select between separate PET list provided in GATE.
mode, CC mode, or joint PET-CC mode based on the In this study, we simulated As with an activity of
72
application–whether standard PET, multi-isotope, or triple 2 MBq over a 1 s acquisition time. The visualization of the
gamma coincidence imaging. These findings build upon experiment in GATE is depicted in Figure 4. The source,
previous work that quantified increases in sensitivity of represented as a 0.1 mm radius sphere, is positioned
similar single-layer CZT detector systems for dual PET-CC centrally to the panels at the origin of a Cartesian coordinate
imaging purposes, making a significant advancement as space (x, y, z) specified at (0, 0, 0). To demonstrate positron
45
no CC reconstruction had been provided until this point. range and radiotracer behavior in soft-tissue equivalent

2. Methods material, the source was placed centrally within a spherical
water phantom of 2 cm diameter.
The study utilized Monte Carlo simulation of the dual-panel
CZT PET detector imaging system, employing the well- 2.3. Image reconstruction
established Geant4 application for tomography emission We performed PET image reconstruction utilizing an
(GATE) software. 46,47 The simulated isotope, As, with its in-house list-mode maximum likelihood expectation
72
large positron range (5.19 mm), enables comparison of PET maximization (LM-MLEM) iterative reconstruction
and CC reconstruction methods. Furthermore, its prompt- code, implemented in the compute unified device
gamma emission (834 keV at 81%) allows us to demonstrate architecture (CUDA) software. This method follows the
increased system sensitivity by detecting scattering for standard formulation of LM-MLEM as described by
COR projection data. Reconstruction was performed using Equation I. 48,49
ground truth information from the Monte Carlo simulation,
i.e., discarding random and scatter coincidences from f j k p
prompt-gamma down scattering. In addition, PET and f j k ( 1 ) s a ij i k (I)
af
CC reconstruction were performed without energy, time, j i j ij j
or spatial blurring, demonstrating the best-case scenarios
for the performance of PET and CC modalities within the CC image reconstruction follows a similar manner,
system. Thus, no regularization or filters were applied in employing an open-source LM-MLEM iterative
our MLEM PET and CC image reconstruction methods. To reconstruction CUDA code, formulated as in Equation I.
account for reported energy resolutions of 5.85% and 4.40% The initial image was initialized to uniformity, represented
at 511 keV and 622 keV, respectively, for flexible circuit- by f ()0 = 1 . The system matrix, a was constructed in our
j
ij
bonded cross-strip CZT detectors, we introduced a 51 keV PET reconstruction based on orthogonal distance-based
energy uncertainty to CC projection data. 41 ray-tracer (OD-RT) projectors and a fixed Gaussian kernel
for the tube of response (TOR). In the CC reconstruction,
50
2.1. System geometry
the system matrix was constructed based on a ray-tracing
The dual-panel CZT PET detector, as constructed in GATE, method where the surfaces of the cone projections are
is presented in Figure 3. It consists of 4 × 4 × 0.5 cm CZT sampled as a set of line samples with energy-based
3
detector crystals. Each panel comprises 150 CZT detector Gaussian kernels for the volume of response (VOR). 51,52
crystals arranged in five columns of 30 edge-on stacked We assumed uniformity for the detector sensitivity in
CZT detectors. The panels boast a detector surface of both methods, denoted as s =1. Since LM is implemented
j
20 × 15 cm , with a thickness of 4 cm and a distance of for both reconstructions, only the captured projections
2
20 cm between the faces of the two panels. were considered, set as p =1. Implementation of angular
i
blurring in the form of energy resolution and doppler
2.2. Radioisotope definitions broadening was taken into account in the system matrix
In our simulations using GATE, radioisotopes were construction of the CC reconstruction. 53,54 The 3D image
specified with the ion source definitions, including reconstruction was done on a 40 × 40 × 40 voxel grid

Volume 2 Issue 2 (2024) 4 doi: 10.36922/arnm.3330