Pathogeny Detection for Mild Cognitive Impairment via Weighted Evolutionary Random Forest With Brain Imaging and...

Area of Science:

• Neuroscience
• Genomics
• Medical Imaging

Background:

• Mild cognitive impairment (MCI) diagnosis and pathogenesis analysis often rely on medical imaging and gene sequencing separately.
• Limited research fuses radiomics and genomics data to leverage complementary information for detecting MCI pathogenic factors.

Purpose of the Study:

• To develop a multimodal data analysis framework for identifying MCI patients and extracting pathogenic factors.
• To fuse functional magnetic resonance imaging (fMRI) and single nucleotide polymorphism (SNP) data for enhanced MCI analysis.

Main Methods:

• Constructed fusion features by correlating regions of interest (ROIs) from fMRI with digitalized gene sequences.
• Developed a novel weighted evolutionary random forest (WERF) model incorporating a weighted evolution strategy for feature selection.
• Established an overall multimodal data analysis framework for MCI detection and pathogenic factor extraction.

Main Results:

• The proposed framework effectively identifies MCI patients.
• The WERF model successfully eliminated inefficient features, improving analysis accuracy.
• Comparative analysis against existing methods demonstrated the framework's superior performance using ADNI database data.

Conclusions:

• The multimodal fusion framework offers a powerful approach for detecting MCI pathogenic factors.
• This study highlights the potential of integrating radiomics and genomics for a comprehensive understanding of MCI.
• The developed framework shows promise as an effective tool for clinical applications in MCI research.