Abstract
241473
Introduction: The NeuroEXPLORER (NX) is an ultra-high-performance brain-dedicated PET system with a design focused on ultra-high sensitivity and resolution, with continuous head motion correction. The NX has a cylindrical design with a diameter of 52.4 cm and long axial field-of-view of 49.5 cm. The detector micro-block consists of a 4x2 array of 1.56x3.07x20 mm LYSO crystals decoded by 2x2 3-mm SiPMs including depth-of-interaction (DOI) measurement. Performance tests showed time-of-flight (TOF) resolution of 236 psec, sensitivity of 46.8 kcps/MBq (NEMA), and point source transverse resolution of 1.8-2.2 mm with FBP and <1.4 mm with OSEM. The goal of this study was to demonstrate the performance of the NX with human brain studies in 2 ways: 1) using targeted radiopharmaceuticals, and 2) comparing scan quality to the HRRT (previous state-of-the-art) in the same subjects.
Methods: NX scans were performed under IRB-approved protocols with tracers targeting SV2A (18F-SynVesT-1, 177 MBq), dopamine D2/D3 receptors (11C-PHNO, 265 MBq), M1 muscarinic receptors (11C-LSN3172176, 455 MBq), and dopamine transporters (DAT, 18F-FE-PE2I, 170 MBq). In addition, paired scans were performed on different days on the NX and HRRT with 18F-FDG (~ 10 mCi) and 18F-SynVesT-1 (4-5 mCi). A CT (NX) or transmission scan (HRRT) were performed for attenuation correction. List mode data were acquired for up to 2 h. Reconstructions were performed with OSEM (NX: 7 iter., 10 subsets, 0.5-mm voxels; HRRT: 2 iter., 30 subsets, 1.2-mm voxels) with corrections for point spread function, attenuation, scatter, randoms, and deadtime. The NX reconstruction also included TOF, depth-of-interaction, and inter-crystal scatter correction. For the NX, head motion was assessed using a stereovision camera and correction was applied post-reconstruction using the average position during each scan frame. HRRT motion correction was performed event-by-event. Images were registered to MR for anatomical localization. Parametric images of binding potential (BPND) and relative delivery (R1) were obtained using SRTM2 with no filtering and population-based reference region k2.
Results: Fig. 1A shows SV2A images at early and late times, demonstrating exceptional resolution in cortex and subcortical structures. A differential pattern of uptake between early "flow" images and late SV2A images is seen. For example, inferior colliculi are visualized only in the early images, while the activity distribution in the thalamus changes from early to late frames. Fig. 1B shows D2/3 BPND images, with clear delineation of the substantia nigra (left, arrow), high binding along the nigrostriatal tract, and focal medial bilateral binding in a thalamic nucleus (right, arrow, likely anteroventral nucleus). Fig. 1C and D show M1 BPND and R1 images, respectively, with dramatic distribution changes particularly in the cerebellum. Fig. 1E and F show DAT BPND and R1 images, respectively, where the nigra is clearly visualized, albeit with lower contrast than for 11C-PHNO. Optimization of SRTM2 parameters is needed to minimize outliers in BPND images (Fig. 1E). Fig. 2 shows a comparison of NX and HRRT scans performed in the same subjects for glucose metabolism (A) and synaptic density (B). The striking improvement in image contrast and quality of the NX compared to the HRRT is evident; this makes clear the magnitude of partial volume averaging that has been present in HRRT images.
Conclusions: Human NX images with targeted tracers demonstrate low noise and exquisite resolution, showing focal uptake in specific brain nuclei and providing the opportunity for research upstream of neuronal activity. The dramatic improvement in resolution and overall quality of the NX images compared to the HRRT is clear. Ongoing work includes full implementation of continuous motion correction, optimization of reconstruction parameters, and a comprehensive comparison to the HRRT. We conclude that the NX will dramatically expand the scope of brain PET studies.
Funding:U01EB029811