Predictive classification of Alzheimer's disease using brain imaging and genetic data

Area of Science:

• Neuroscience
• Genetics
• Medical Imaging

Background:

• Alzheimer's disease (AD) is currently incurable, necessitating early diagnosis for effective treatment.
• Existing diagnostic methods often rely on single or multi-modal imaging, with limited integration of genetic features.
• Accurate differentiation between healthy controls (HC), early mild cognitive impairment (EMCI), late mild cognitive impairment (LMCI), and AD is challenging.

Purpose of the Study:

• To develop an integrated feature selection method combining brain imaging and genetic data for accurate diagnosis of AD and MCI stages.
• To improve classification accuracy compared to methods using only imaging features.
• To identify key interrelated brain imaging phenotypes and genetic factors associated with cognitive decline.

Main Methods:

• Proposed an integrated Fisher score and multi-modal multi-task feature selection approach.
• Applied Fisher score for dimensionality reduction of genetic features to address scale differences with imaging data.
• Selected five imaging and five genetic features through the developed selection program.

Main Results:

• Achieved high classification accuracies: 98% for HC vs. AD, 82% for HC vs. EMCI, 86% for HC vs. LMCI, 80% for EMCI vs. LMCI, 88% for EMCI vs. AD, and 72% for LMCI vs. AD.
• Demonstrated improved classification performance compared to using imaging features alone.
• Identified a set of interrelated imaging and genetic features crucial for disease diagnosis.

Conclusions:

• The integrated approach combining brain imaging and genetic features significantly enhances the accuracy of diagnosing Alzheimer's disease and its early stages.
• The selected subset of imaging and genetic features provides valuable biomarkers for cognitive impairment detection.
• This method offers a promising avenue for early and accurate diagnosis, potentially aiding in disease management.