Frontotemporal dementia characterization using neurite orientation dispersion and density imaging.

Area of Science:

• Neuroimaging
• Neurology
• Biomedical Engineering

Background:

• Frontotemporal dementia (FTD) encompasses several neurodegenerative disorders characterized by progressive changes in brain white matter (WM) and grey matter (GM).
• Understanding the specific microstructural alterations in different FTD variants is crucial for accurate diagnosis and management.
• Diffusion MRI techniques offer sensitive measures of brain microstructure, but their application in differentiating FTD subtypes requires further refinement.

Purpose of the Study:

• To assess WM and GM microstructure in FTD variants using the neurite orientation dispersion and density imaging (NODDI) model.
• To develop and evaluate a machine-learning algorithm for classifying FTD subtypes based on diffusion MRI metrics and neuropsychological data.
• To compare the efficacy of NODDI with standard diffusion tensor (DT) imaging in detecting disease-relevant microstructural changes.

Main Methods:

• Acquired multi-shell diffusion MRI and conducted neuropsychological assessments in controls and participants with behavioural variant FTD (bvFTD), semantic-variant PPA (svPPA), nonfluent-variant PPA (nfvPPA), and semantic-bvFTD (sbvFTD).
• Analyzed diffusion MRI metrics including fractional anisotropy (FA), mean diffusivity (MD), intracellular-volume fraction (ICVF), and orientation-dispersion index (ODI) using tract-based and GM-based spatial statistics.
• Trained Support Vector Machine (SVM) models using combinations of diffusion MRI features and cognitive scores to classify FTD subtypes.

Main Results:

• Widespread WM alterations (reduced FA and increased MD) were observed across all FTD variants.
• Reductions in GM and WM ICVF and ODI showed distinct patterns correlating with specific FTD subtypes.
• The SVM algorithm, integrating NODDI metrics and cognitive data, achieved 98.6% accuracy in classifying FTD subtypes, outperforming standard DT imaging models.

Conclusions:

• NODDI effectively captures subtle microstructural alterations in both GM and WM in FTD, offering advantages over standard DT imaging.
• Integrating NODDI-derived diffusion MRI metrics with cognitive data in machine-learning models facilitates accurate differentiation of FTD subtypes.
• This approach holds promise for improving diagnostic accuracy and understanding the pathophysiology of diverse FTD syndromes.