Abstract
Ziel/Aim: The ProVision scanner is a dedicated PET system for prostate imaging with a dual-head setup, advanced detector technology, and depth-of-interaction (DOI) and time-of-flight (TOF) capabilities. We have assessed the image performance using a flexible image reconstruction framework for limited-angle TOF-DOI-PET with an initial investigation of point spread function (PSF) correction for truncated angular sampling artifacts.
Methodik/Methods: The ProVision detector [1] array has 4 crystal layers, each with six 30-mm axially oriented crystals with a pitch of 2.05x4.40 mm², providing distinct DOI data. A dual-ended shared readout achieves a timing resolution of 170 ps and intrinsic resolutions of 3.53 mm (Z) and 1.88 mm (X) FWHM. Each scanner head has a 3x4 detector matrix, with top and bottom heads spaced 27.4 cm apart and limited translation movement. A 22-step scanning protocol expands the FOV and reduces truncation artifacts. For image reconstruction, we optimised list-mode Maximum-A-Posteriori, including TOF data, detector attenuation, and positional uncertainty using multiray sampling [2]. Scanner characterisation includes point source data and a NEMA NU4-2008 IQ phantom, with an initial assessment of post-reconstruction PSF correction using Richardson-Lucy deconvolution.
Ergebnisse/Results: The system spatial resolution (FWHM) for an 18F-FDG point source (1 mm diameter, 2 mm length) is 1.5 mm, 3.2 mm, and 2.2 mm in the x, y, and z directions, respectively. Reconstructed images of the IQ phantom yield recovery coefficients of 0.77, 0.71, 0.60, 0.37 and 0.31 for rods from 5 mm to 1 mm, respectively, with a 15.74% non-uniformity and spill-over ratios of 0.21 (air) and 0.31 (water). A post-reconstruction, space-invariant PSF correction reduces elongation by 41.2% in the y-direction and 51.1% in the z-direction for an array of 6 x-axis point sources. A pelvis phantom was scanned, and its reconstructed image superimposed on a CT scan.
Schlussfolgerungen/Conclusions: The ProVision detector concept shows promising results for cost-effective stand-alone prostate PET imaging. Planned improvements include better reconstruction methods, space-variant PSF correction, enhanced timing resolution, and AI integration, aiming to further improve image quality.
Methodik/Methods: The ProVision detector [1] array has 4 crystal layers, each with six 30-mm axially oriented crystals with a pitch of 2.05x4.40 mm², providing distinct DOI data. A dual-ended shared readout achieves a timing resolution of 170 ps and intrinsic resolutions of 3.53 mm (Z) and 1.88 mm (X) FWHM. Each scanner head has a 3x4 detector matrix, with top and bottom heads spaced 27.4 cm apart and limited translation movement. A 22-step scanning protocol expands the FOV and reduces truncation artifacts. For image reconstruction, we optimised list-mode Maximum-A-Posteriori, including TOF data, detector attenuation, and positional uncertainty using multiray sampling [2]. Scanner characterisation includes point source data and a NEMA NU4-2008 IQ phantom, with an initial assessment of post-reconstruction PSF correction using Richardson-Lucy deconvolution.
Ergebnisse/Results: The system spatial resolution (FWHM) for an 18F-FDG point source (1 mm diameter, 2 mm length) is 1.5 mm, 3.2 mm, and 2.2 mm in the x, y, and z directions, respectively. Reconstructed images of the IQ phantom yield recovery coefficients of 0.77, 0.71, 0.60, 0.37 and 0.31 for rods from 5 mm to 1 mm, respectively, with a 15.74% non-uniformity and spill-over ratios of 0.21 (air) and 0.31 (water). A post-reconstruction, space-invariant PSF correction reduces elongation by 41.2% in the y-direction and 51.1% in the z-direction for an array of 6 x-axis point sources. A pelvis phantom was scanned, and its reconstructed image superimposed on a CT scan.
Schlussfolgerungen/Conclusions: The ProVision detector concept shows promising results for cost-effective stand-alone prostate PET imaging. Planned improvements include better reconstruction methods, space-variant PSF correction, enhanced timing resolution, and AI integration, aiming to further improve image quality.
| Original language | English |
|---|---|
| Title of host publication | Nuklearmedizin 2025 |
| Volume | 64 |
| Publisher | Georg Thieme Verlag KG |
| Publication date | 12.03.2025 |
| Edition | 01 |
| Pages | L52 |
| DOIs | |
| Publication status | Published - 12.03.2025 |
