Feature Fusion and Detection in Alzheimer's Disease Using a Novel Genetic Multi-Kernel SVM Based on MRI Imaging and...

Area of Science:

• Neuroimaging
• Genetics
• Computational Biology

Background:

• Voxel-based morphometry (VBM) offers subtle insights into Alzheimer's disease (AD) neurobiology.
• Identifying critical brain voxels for classifying AD, early mild cognitive impairment (EMCI), and healthy controls (HC) is crucial for understanding AD mechanisms.

Purpose of the Study:

• To develop a novel method combining MRI and genetic data for identifying important features in AD detection.
• To explore the neurobiological underpinnings of AD by analyzing significant brain regions and genes.

Main Methods:

• Proposed a novel feature construction method using eigenvalues of top Single Nucleotide Polymorphisms (SNPs) from AD-associated genes.
• Developed a genetic multi-kernel Support Vector Machine (SVM) to identify optimal kernel weights and significant features.
• Analyzed feature significance to pinpoint affected brain regions and associated genes.

Main Results:

• Identified key brain regions affected in AD, including the right superior frontal gyrus, right inferior temporal gyrus, and right superior temporal gyrus.
• Discovered significant AD susceptibility genes such as CSMD1, RBFOX1, PTPRD, CDH13, and WWOX.
• Highlighted significant pathways including calcium signaling and cell adhesion molecules.

Conclusions:

• The proposed model demonstrates strong performance and generalization for AD detection.
• The study provides new candidate abnormal brain features and insights into their contribution to AD.
• Findings contribute to a deeper understanding of the neurobiological mechanisms underlying AD.