Effective count rates for PET scanners with reduced and extended axial field of view

Abstract

We investigated the relationship between noise equivalent count (NEC) and axial field of view (AFOV) for PET scanners with AFOVs ranging from one-half to twice those of current clinical scanners. PET scanners with longer or shorter AFOVs could fulfill different clinical needs depending on exam volumes and site economics. Using previously validated Monte Carlo simulations, we modeled true, scattered and random coincidence counting rates for a PET ring diameter of 88 cm with 2, 4, 6, and 8 rings of detector blocks (AFOV 7.8, 15.5, 23.3, and 31.0 cm). Fully 3D acquisition mode was compared to full collimation (2D) and partial collimation (2.5D) modes. Counting rates were estimated for a 200 cm long version of the 20 cm diameter NEMA count-rate phantom and for an anthropomorphic object based on a patient scan. We estimated the live-time characteristics of the scanner from measured count-rate data and applied that estimate to the simulated results to obtain NEC as a function of object activity. We found NEC increased as a quadratic function of AFOV for 3D mode, and linearly in 2D mode. Partial collimation provided the highest overall NEC on the 2-block system and fully 3D mode provided the highest NEC on the 8-block system for clinically relevant activities. On the 4-, and 6-block systems 3D mode NEC was highest up to ∼300 MBq in the anthropomorphic phantom, above which 3D NEC dropped rapidly, and 2.5D NEC was highest. Projected total scan time to achieve NEC-density that matches current clinical practice in a typical oncology exam averaged 9, 15, 24, and 61 min for the 8-, 6-, 4-, and 2-block ring systems, when using optimal collimation. Increasing the AFOV should provide a greater than proportional increase in NEC, potentially benefiting patient throughput-to-cost ratio. Conversely, by using appropriate collimation, a two-ring (7.8 cm AFOV) system could acquire whole-body scans achieving NEC-density levels comparable to current standards within long, but feasible, scan times.

Figure 1

PET scanners with 2-, 4-, 6-, and 8-block rings axially, and the position of the anthropomorphic object, and cylinder object. The dashed lines show the axial extent of the solid angle subtended by the center of the scanner and the end-shielding in 3D mode for the 2-, 4-, and 8-ring scanners (6-ring scanner omitted for clarity). The figure is roughly to scale.

Figure 2

Voxelization and segmentation of the attenuation coefficients of the anthropomorphic phantom. The linear attenuation coefficients (A) were calculated from a CT image via conventional multi-linear conversion on the DSTE. The indicated regions were assigned to SimSET attenuation indexes according to the best match. (B) Segmented attenuation object; central sagittal slice. SimSET material indexes are shown.

Figure 2

Voxelization and segmentation of the attenuation coefficients of the anthropomorphic phantom. The linear attenuation coefficients (A) were calculated from a CT image via conventional multi-linear conversion on the DSTE. The indicated regions were assigned to SimSET attenuation indexes according to the best match. (B) Segmented attenuation object; central sagittal slice. SimSET material indexes are shown.

Figure 3

NEC rates on the DSTE (15.5 cm AFOV) scanner in 3D mode for the 200 cm long cylinder phantom used in this work, the NEMA standard 70 cm long cylinder phantom, and the anthropomorphic phantom.

Figure 4

Peak NEC (A) and NEC at 370 MBq (B) vs. maximum axial acceptance in SSRB for the 200 cm long cylinder scanned with the 8-block ring system.

Figure 5

(A) True, (B) random, and (C) NEC event rates vs. scanner AFOV for each collimation mode. (D) 3D mode true, random, and NEC rates were normalized to a value of one at 7.8 cm AFOV and displayed as a function of scanner AFOV to show relative increases for each count type. Rates shown correspond 260 MBq within the 200 cm cylinder phantom.

Figure 6

(A): NEC rates as a function of activity in the anthropomorphic phantom, for each AFOV system, and each collimation mode. Note the differences in vertical scales.

Figure 6

(A): NEC rates as a function of activity in the anthropomorphic phantom, for each AFOV system, and each collimation mode. Note the differences in vertical scales.