Deep Learning-Based Feature Extraction with MRI Data in Neuroimaging Genetics for Alzheimer's Disease

Area of Science:

• Neuroscience
• Genetics
• Artificial Intelligence

Background:

• Alzheimer's disease (AD) poses significant challenges in prognosis and treatment.
• Identifying genetic variants associated with brain atrophy is crucial for understanding AD mechanisms.
• Current neuroimaging analyses using predefined brain atlases may lose critical functional information.

Purpose of the Study:

• To develop a data-driven method for automatic feature extraction from 3D MRI data for Alzheimer's disease research.
• To identify genetic variants associated with brain atrophy using advanced computational models.
• To improve the understanding of genetic underpinnings of neurodegenerative and mental disorders.

Main Methods:

• Application of convolutional neural network (CNN) models to whole-brain and regional 3D MRI data.
• Data-driven and automatic extraction of image-derived features (endophenotypes).
• Genome-wide association studies (GWASs) utilizing these extracted features to identify genetic variants.

Main Results:

• Successfully identified single nucleotide polymorphisms (SNPs) associated with brain atrophy using the proposed CNN-based method.
• The identified SNPs have known associations with Alzheimer's disease, depression, and schizophrenia.
• Demonstrated the efficacy of the data-driven approach in uncovering relevant genetic associations.

Conclusions:

• The proposed CNN-based method offers a powerful, data-driven approach for feature extraction in neuroimaging studies.
• This method can identify genetic variants linked to brain atrophy and various neurological disorders.
• The findings contribute to a better understanding of the genetic basis of Alzheimer's disease and related conditions.