Objective: To develop a diffusion MRI-based neurodegeneration index using free-water imaging and assess its ability to quantify neurodegenerative burden in Lewy body dementia (LBD) and Alzheimer’s disease (AD).

Background: Quantifying neurodegeneration in parkinsonism and dementia remains challenging. Current markers of neuronal atrophy lack standardized thresholds and vary across populations. Individual markers are also influenced by aging or other neurodegenerative processes. Free-water diffusion MRI (dMRI) measures extracellular and tissue microstructural changes linked to neurodegeneration and neuroinflammation.

Method: A support vector machine (SVM) model was used to generate the Automated Index of Neurodegeneration for Dementia (AIND). Data were obtained from the Alzheimer’s Disease Neuroimaging Initiative, Parkinson’s Progression Markers Initiative and Parkinson’s Disease Biomarkers Program. Data from 1Florida Alzheimer’s Disease Research Center served as an independent test set. We included 497 dMRI scans from controls and people with mild cognitive impairment or dementia due to AD, dementia with Lewy bodies or Parkinson’s disease. Participants were stratified by Montreal Cognitive Assessment scores into high, intermediate and low impairment groups. Free-water (FW) maps were generated from an automated pipeline. Model inputs included FW, FW-corrected fractional anisotropy, age and sex. The SVM classified high vs. low impairment groups. Associations between AIND scores and markers of AD pathology and neurodegeneration were examined using linear regression.

Results: The training/validation cohort (n=303) showed an AUC of 0.979 (95% CI: 0.966-0.991). The test cohort (n=76) showed an AUC of 0.927 (95% CI: 0.864-0.991), and the independent cohort (n=119) showed an AUC of 0.913 (95% CI: 0.845-0.981). AIND scores correlated positively with amyloid-PET Centiloid (r=0.456), glial fibrillary acidic protein (r=0.366), neurofilament light chain (r=0.315), pTau181 (r=0.418) and pTau217 (r=0.481) (all p≤0.01), and negatively with Aβ42/40 (r=-0.356), cortical gray matter and hippocampal volumes (all p≤0.02).

Conclusion: AIND showed strong performance in quantifying neurodegenerative changes in LBD and AD. Its associations with molecular and imaging biomarkers support its biological relevance and suggest potential utility as a standardized, clinically interpretable measure of neurodegenerative burden across common dementia syndromes.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/a-diffusion-mri-neurodegeneration-index-for-lewy-body-dementia-alzheimers-disease/