Multimodal Hippocampal Subfield Grading For Alzheimer's Disease Classification

Area of Science:

• Neuroimaging
• Biomarker Discovery
• Alzheimer's Disease Research

Background:

• Magnetic resonance imaging (MRI) is crucial for detecting Alzheimer's disease (AD) progression.
• Hippocampal alterations are early indicators of AD, but subfield-specific changes are complex.
• Current patch-based grading methods show promise but can be improved.

Purpose of the Study:

• To develop and evaluate a novel multimodal patch-based framework combining structural and diffusion MRI for improved AD classification.
• To assess the framework's efficacy across different hippocampal subfields.
• To compare the new framework against existing volumetric and mean diffusivity measures.

Main Methods:

• A multimodal patch-based framework integrating structural and diffusion MRI data was developed.
• The framework was applied to the whole hippocampus and its subfields.
• Classification accuracy was compared using volume, mean diffusivity, and the novel multimodal approach.

Main Results:

• The multimodal patch-based method applied to the whole hippocampus yielded the most discriminating biomarker for advanced AD detection.
• Applying the novel framework to the subiculum achieved the best results for AD prediction.
• The subiculum-focused approach improved prediction accuracy by two percentage points compared to using the whole hippocampus.

Conclusions:

• A novel multimodal MRI framework significantly enhances AD detection and prediction accuracy.
• The subiculum is a key region for early AD prediction using this advanced neuroimaging technique.
• This approach offers a more refined biomarker for understanding and diagnosing Alzheimer's disease.