Diagnosis of Alzheimer's Disease Using Brain Network

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Biology

Background:

• Brain functional connectivity impairment is an early indicator of Alzheimer's disease (AD) and mild cognitive impairment (MCI).
• Modeling the brain as a graph-based network aids in studying these impairments.

Purpose of the Study:

• To develop and evaluate a novel diagnostic approach for discriminating between AD, MCI, and healthy control (HC) subjects.
• To leverage graph theory-based features from functional magnetic resonance (fMRI) images for enhanced diagnostic accuracy.

Main Methods:

• Constructed brain networks using pairwise Pearson's correlation-based functional connectivity.
• Employed Node2vec graph embedding to convert graph features into vectors.
• Utilized classification techniques including Linear Support Vector Machine (LSVM) and Regularized Extreme Learning Machine (RELM).
• Compared classification performance on Alzheimer's Disease Neuroimaging Initiative (ADNI) datasets.

Main Results:

• The Support Vector Machine (SVM) combined with LASSO feature selection demonstrated superior classification accuracy compared to other tested methods.
• Graph theory-based features effectively discriminated between AD, MCI, and HC groups.

Conclusions:

• The proposed graph theory-based approach using fMRI data shows significant potential for early diagnosis of AD and MCI.
• SVM with LASSO feature selection is a highly effective classification strategy for this neuroimaging-based diagnostic task.