Domain Transfer Learning for MCI Conversion Prediction.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Imaging Analysis

Background:

• Machine learning models are used to predict Alzheimer's disease (AD) progression from mild cognitive impairment (MCI).
• Current methods often overlook valuable data from related neurological domains like AD and normal controls (NC).
• This limitation hinders optimal performance in classifying MCI converters (MCI-C) from MCI nonconverters (MCI-NC).

Purpose of the Study:

• To develop a novel domain transfer learning method for enhanced MCI conversion prediction.
• To integrate data from MCI, AD, and NC domains to improve classification accuracy.
• To effectively utilize multi-modal data including MRI, FDG-PET, and CSF.

Main Methods:

• A domain transfer learning framework incorporating feature and sample selection.
• Selection of informative features and samples across target (MCI) and auxiliary (AD, NC) domains.
• A domain transfer support vector machine (SVM) classifier to fuse selected data for MCI-C vs. MCI-NC classification.

Main Results:

• The proposed method achieved a 79.4% accuracy in classifying MCI-C from MCI-NC patients.
• Demonstrated the benefit of incorporating auxiliary domain knowledge (AD and NC data).
• Successfully integrated multi-modal data (MRI, FDG-PET, CSF) for improved prediction.

Conclusions:

• Domain transfer learning offers a promising approach to enhance MCI conversion prediction.
• Leveraging data from related neurological conditions significantly improves predictive performance.
• The developed method provides a robust tool for early detection of Alzheimer's disease progression.