Comparison of Partial Volume Correction methods for Huntington’s disease integrated PET/MR imaging analysis

Partial volume effects (PVEs) bias PET imaging, particularly in neurodegenerative diseases with regional atrophy such as Huntington’s disease (HD). Despite their relevance, partial volume correction (PVC) methods remain inconsistently applied in HD PET studies. This study compared three MRI-assisted voxelwise PVC approaches to identify the most robust method for quantifying phosphodiesterase 10A availability using ¹¹ C-IMA107 PET in manifest HD. ¹¹ C-IMA107 PET and T1-weighted MRI data were analyzed from 10 people with manifest HD and 10 age- and sex-matched healthy controls (HCs). Parametric binding potential (BP _ND ) maps were generated and corrected using three voxelwise MRI-assisted PVC methods: Multiresolution-Multimodal Resolution-Recovery (MMRR), Van Cittert (VC), and Zhu’s deconvolution-based approach. Regional BP _ND values were extracted from basal ganglia regions. Group comparisons, correlations between binding and regional volumes, and effect sizes were assessed before and after PVC. Reliability between baseline and 1-year follow-up was evaluated in HCs using intraclass correlation coefficients (ICC). All PVC methods increased BP _ND relative to uncorrected estimates, with larger effects in HD, consistent with stronger PVEs due to atrophy. Group differences between HD and HCs were preserved and enhanced after PVC, particularly in striatal regions. VC and Zhu produced highly consistent quantitative results, while MMRR showed greater variability. Zhu’s method provided the best balance between contrast recovery, noise control, and longitudinal reliability (ICC>0.82 in striatal regions). Voxelwise PVC improves sensitivity and reliability of ¹¹ C-IMA107 PET quantification in HD. Among evaluated approaches, Zhu showed the most favorable balance between stability and sensitivity, supporting its use in longitudinal and interventional HD PET studies.