Comparison of brain connectomes by MRI and genomics and its implication in Alzheimer's disease - BMC Medicine

• MRI observes brain connectome change → genetic studies observe molecular change
  • how one brain region impacts molecular pathways
• DATASET →

Data Aggregation:

• Gene expression/Transcriptome Data
• Cell Feature Data → Morphology (Histogram, Image Data), Electrophysiological
• MRI/CT Scans → Visual Scans and Queues

• multiylayer analysis of root-data and high-level data (each brain layer, brain anatomy, gene expression, signalling, etc)

Synthesis:

• CNN image segmentation model on brain MRI to find inflamed regions
• Analysis of gene regulation pathways to inflamed regions to find which genes are being under/over expressed to cause inflammation in pathways
  • Use this data to find relations between inflamed regions, gene, and transcript data
• Feature selection algorithms on reduced gene expression, transcriptome data to find most pertinent gene features before running
  • Dimensionality reduction (TSNE) for GE, transcriptome data
  • Dimensionality reduction (PCA) for cell feature data
• Analyze statistically significant gene features (differential transcript and gene expression using Sleuth) for gene targets for Alzheimer's

Type of ML Task:

• CNN segmentation
• Feature selection algorithms