The study presented in this article was carried out using GATE (Geant4 Application for Tomographic Emission) software. GATE is a validated simulation tool based on the Monte-Carlo method dedicated to nuclear medicine applications [30, 31]. Alongside similar applications of the uEXPLORER and J-PET-based Total-Body PET scanners, they use different materials and designs in their scanner. The two main parameters that distinguish these technologies are the scintillator material (organic plastics or inorganic crystals) and the arrangement of detection units. For each of the groups, several tomographs have been simulated to be compared based on all parameters which influence their sensitivity and cost.

The uEXPLORER-based scanners are the first group of tomographs that were simulated. For these cases, we simulated 8 PET scanners based on the crystal configuration of the uEXPLORER. These groups stand for current clinical tomographs that have a radial arrangement of scintillation crystals. These eight configurations, combined from 2 different types of crystals (LYSO and BGO), two different AFOV (194.8 and 97.4 cm), and two different geometrical configurations (full and sparse) were simulated [8, 9, 32]. The reasons for performing an investigation over the scanners mentioned above are the proposed solutions to reduce the construction cost of TB PET scanners. Sparse geometry has been introduced as a cost-effective solution to extend AFOV [33].

In this study, we simulated 8 TB J-PET scanners with various configurations of panels to evaluate the effect of plastic scintillator dimensions and multiple layers of modules on tomograph performance. For the scintillator with a cross section of 4 mm \(\times\) 20 mm, a two-layer geometry with 200 cm and 250 cm lengths was simulated. For the plastic strips with a 6 mm \(\times\) 30 mm cross section, two, three, and four layers configurations with 200 cm and 250 cm lengths were simulated. Plastic scintillator strips have been equipped with SiPM at each end in all panel configurations. These specific arrangements of SiPMs and plastic scintillators allow for extending the AFOV without incrementing the number of SiPMs or electronics but only by increasing the length of plastic strips [7].

The detection panels of TB J-PET are equipped with wavelength shifters (WLS) which are utilized to improve axial resolution [20, 26, 34]. The WLS with dimensions of 3 mm \(\times\) 108.15 mm \(\times\) 6 mm is located perpendicular to the plastic strips, as shown in Fig. 1. The WLS layers are read out by the SiPMs, coupled to them from one side.

In the TB J-PET, the expected axial spatial resolution for the registration of gamma photons is equal to 2.1 mm. Spatial resolutions of the image are estimated to be 3.7 mm in transversal and 4.9 mm in axial direction [20]. The TOF resolution for the J-PET was estimated as a function of the lengths of the applied scintillator strips, and it varies from CRT = 140 ps to CRT = 240 ps when the length is increased from 50 cm to 200 cm [20]. For the scatter fraction reduction, the energy loss threshold of 200 keV will be used, resulting in the scatter fraction of 36.2 % [20]. For the two-layer Total-Body J-PET solution, the noise equivalent count rate NECR peak was estimated to be 630 kcps at kBq/mL [20], which is in between the values obtained by uEXPLORER (1524 kcps at 17.3 kBq/mL) [8] and Biograph Vision (306 kcps at 32 kBq/mL) [35].

In order to compare the results of TB PET scanners to standard ones, the Biograph Vision from Siemens was simulated [35,36,37]. Biograph Vision is composed of 8 rings, where each ring consists of 38 panels. Each panel is built from a 20 \(\times\) 10 array of 3.2 \(\times\) 3.2 \(\times\) 20 mm LSO crystals, providing 32 mm in axial direction [35]. In total, Biograph Vision spans 26.3 cm AFOV.

For each one of the geometries, as mentioned earlier, two types of simulations have been performed, (i) with a line source of the diameter of 1 mm and a length of 250 cm located along the central axis of the tomograph and (ii) with a line source surrounded by a cylindrical water-filled phantom with a diameter of 20 cm and a length of 183 cm. The sensitivity for a slice (S\(_i\)) was calculated according to the following formula:

$$\begin S_i=\frac}}}} , \end$$

(1)

where \(L_}\) is the source length, d is the width of the slice, and A\(_}\) is the initial activity. The rate R\(_i\) of each slice in counts per second is determined by dividing the counts collected in the slice by the duration of the measurement.

TB PET provides extended AFOV that considerably improves sensitivity compared to the current clinical PET. Still, it is required to determine new event selection criteria to achieve optimum results [13].

TB PET scans, thanks to the larger AFOV, are capable of detecting more oblique coincidences. While these events contribute positively to the increase in system sensitivity, they deteriorate the axial resolution of the tomograph [38]. Since sensitivity and spatial resolutions are the main characteristics of PET, making a trade-off between these two parameters will enhance the quality of the final reconstructed image [39]. The optimization is performed as a function of the acceptance angle (shown in Fig. 3), which is used for pre-selecting those maximum azimuthal coincidences contributing to the reconstructed image.

Schematic visualization of uEXPLORER (I) (blue) [8, 10, 13]) and sparse geometry (transparent yellow) with 194.8 cm of AFOV and dual layers. TB J-PET (II) with axially arranged plastic scintillators (gray) coupled with SiPMs (black) at each end and arrays of WLS (green) between each layer. The oblique LORs (with large values of \(\theta\)) have a negative contribution in the spatial resolutions due to the parallax error. To avoid it, uEXPLORER uses a ring-based cut that accepts the coincidences within a maximum of 5 rings. As shown in figure (II), TB J-PET uses continuous plastic scintillators (gray), \(\theta _\) denotes the acceptance angle applied for it to cut oblique LORs. \(\theta _\) demonstrate the largest angle of detectable oblique coincidences

Due to the essential difference between the geometrical configurations of TB J-PET and other TB PET scanners, it is necessary to define distinct logic for acceptance angle cuts for them. The TB J-PET provides 2.5 m AFOV with one detection ring. At the same time, the rest of the configurations described in the “The uEXPLORER” sub-section are constructed with several rings along the axial axis. Acceptance angle cut (\(\theta _\)) is a suitable choice for the case of TB J-PET (Fig. 3), but for other configurations, the results will be presented based on maximum ring difference [40].

In this study, two sets of simulations have been performed, first only with a line source axially located at the center of the scanner, then with the same source while surrounded by a cylindrical water-filled phantom. Despite many advantages in performing TB imaging, applying cuts to suppress the oblique coincidence detection is essential to avoid their destructive effect on the spatial resolution of tomographs. Accordingly, the sequences of such cuts on the sensitivity of dedicated scanners have been investigated. For all the scanners, 57° of acceptance angle cut or its equivalent has been performed. For scanners that utilize crystal scintillators, the equivalent is represented by a ring-wise cut.

The acceptance cut has a significant effect on the reduction of the sensitivity of TB PET scanners. Tomographs such as Biograph Vision and 97.4 cm uEXPLORERs’ were neutral against this cut, which can be explained by their smaller AFOV, which fits inside the cut region. Moreover, the effect of this cut has been investigated in the presence of the phantom to mimic clinical scenarios.

The Total-Body sensitivity is defined as the average of rate of detected annihilations originating from within the 183 cm long phantom (which mimics a human body) divided by the total activity \(A_}\) present within it:

$$\begin S_}=\frac^}} \times N} , \end$$

(2)

Where N is the number of slices within the body range, in the case of scanners with an axial field-of-view shorter than 183 cm, the empty slices are outside of the scanner but still within the body range and are also taken into averaging.