Performance evaluation of a sparse detector rings PET scanner with extended axial field of view

Objectives: To evaluate the NEMA performance of a novel PET scanner prototype based on the Siemens Biograph mCT model, with an extended axial field of view (AFOV) PET scanner by means of a sparse detector rings configuration (ES-PET).

Methods: The Siemens Biograph mCT PET/CT encompasses 4 rings of 48 detector blocks of 13x13 LSO compact crystals (4x4x20mm³) each, thus a total AFOV of ~22 cm. We propose a modified model of the mCT, ES-PET, with 8 detector block rings, thus ~44cm AFOV. Each block includes 13 (transaxial) x 7 (axial) LSO crystals interleaved with equal axial gaps of 4mm (equivalent to the LSO axial dimension), thus resulting in a detector block size equivalent to that of the mCT. Monte Carlo simulations for both scanner models, i.e. mCT and ES-PET, were performed using the Geant4 Application for Tomography Emission (GATE). Spatial resolution, system sensitivity, and image quality (IQ) phantom (with 4:1 target-to-background ratio and 6 hot spheres) were evaluated according to NEMA NU 2-2007 guidelines. IQ phantom images were reconstructed using the Software for Tomographic Image Reconstruction (STIR) OSEM algorithm (3 iterations, 24 subsets, 245x245 matrix, and 2 mm slice thickness). A total of 9min simulation time was acquired for each of the two PET models. A maximum ring difference of 49 and 104 was used for the mCT and ES-PET respectively.

Results: ES-PET showed a ~2% difference in the tangential spatial resolution and up to ~8% reduction in the axial spatial resolution. The system sensitivity of the ES-PET was 10.41 cps/kBq, compared 9.5 cps/kBq for the mCT. No image artifacts were identified, and all spheres were detectable in the ES-PET images of the IQ phantom. Contrast recovery coefficients of the 10mm, 13mm, 17mm, 22mm, 28mm and 37mm diameter spheres were 42.8%, 50.04%, 63.17%, 65.23%, 72.86% and 70.77% respectively for the ES-PET model (44.26%, 58.53%, 69.21%, 68.34%, 74.65% and 74.09% respectively for the mCT). For ES-PET, background variabilities of 11.63%, 8.93%, 7.49%, 4.7%, 0.37% and 0.14% were measured for the 10mm, 13mm, 17mm, 22mm, 28mm, and 37mm diameter regions of interest (ROI) respectively (7.46%, 5.93%, 3.56%, 1.61%, 0.44% and 2.03% respectively for the mCT).

Conclusions: The performance of a Monte Carlo model, ES-PET, of the Siemens Biograph mCT PET with an extended sparse detector rings configuration, was validated. ES-PET allowed doubling the PET axial FOV with no increase in total number of scintillators, and without jeopardizing the PET image quality. ES-PET with sparse detector rings has yielded an improved axial spatial resolution and system sensitivity over the compact mCT model. Yet, an increased background variability was observed, which is currently under investigation.