Longitudinal Deformation-Based Morphometry (DBM) Maps and Regional Results Associated with Cognitive Conversion in Parkinson's Disease

This dataset contains voxel-wise statistical maps and regional results derived from deformation-based morphometry (DBM) analyses investigating longitudinal structural brain changes associated with cognitive conversion in Parkinson's disease (PD). The analyses included newly diagnosed, untreated PD participants from the Parkinson's Precision Medicine Initiative (PPMI) who were cognitively normal at baseline and had at least three years of longitudinal cognitive follow-up. Participants were classified as cognitively stable if they remained cognitively normal throughout follow-up, or as PD converters if they transitioned from normal cognition to mild cognitive impairment (MCI) or dementia and remained cognitively impaired thereafter. Participants with fluctuating cognitive status were excluded. The final imaging cohort comprised 350 participants, including 292 cognitively stable participants and 58 converters. T1-weighted MRI scans were processed using PELICAN, an in-house, open-source pipeline. Images were linearly and nonlinearly registered to the MNI-ICBM152 template. Deformation-based morphometry (DBM) was used to quantify regional gray matter volume changes based on the Jacobian determinant of the nonlinear deformation field. Regional DBM measures were extracted from 102 cortical and subcortical regions defined by the CerebrA atlas. Voxel-wise linear mixed-effects models were used to examine differences in longitudinal DBM trajectories between PD converters and cognitively stable participants. Models included cognitive conversion status (DX), time from baseline, the DX-by-time interaction, baseline age, and sex as fixed effects, with participant-specific random intercepts to account for repeated measurements. Cognitive conversion status was coded as DX = 0 for cognitively stable PD participants and DX = 1 for PD converters. Accordingly, the DX effect represents the difference between converters and cognitively stable participants, while the DX-by-time interaction represents the difference in longitudinal DBM trajectories between the two groups. The dataset includes beta coefficient maps for the following model effects:- Baseline age (Age_Baseline)- Cognitive conversion group (DX_1)- Sex (Sex_Male)- Time from baseline (Years_from_Baseline)- Cognitive conversion group-by-time interaction (Years_from_Baseline_DX_1) For each effect, both unthresholded beta coefficient maps and corresponding FDR-significant beta maps (q < 0.05) are provided. The file LME_DBM_AllRegions_WithEffectSize.csv contains regional statistical results for the same analysis across all 102 CerebrA regions, including effect-size estimates, T-statistics, P-values, and FDR-corrected Q-values for the group, time, and group-by-time effects. All statistical analyses were corrected for multiple comparisons using the False Discovery Rate (FDR) procedure. The MRI-derived statistical maps are provided in MINC (.mnc) format in MNI-ICBM152 space. The regional results are provided in CSV format. The original participant-level PPMI data are not redistributed as part of this dataset. They can be requested through the PPMI data-access portal. For additional information regarding the participants, imaging methods, and statistical analyses, see: Moqadam R, Azizi H, Raeesi S, Zeighami Y, Dadar M. Vascular and neurodegenerative contributions to cognitive decline and multidomain progression in de novo Parkinson's disease. medRxiv. 2026. doi:10.64898/2026.09.09.26362673.