AutismCNN

Developing a CNN-based classfier for identifying connectomes from subjects with ASD.

Dataset

• CHARM dataset provided by collaborators at Penn Medicine
  • Connectomes from 450 subjects in total, 313 of which passed Data QA check
  • 163 ASD (Autism-Spectrum Disorder) & 150 TDC (Typically Developing Controls)
  • 2 Sets of connectomes:
    • Desikan atlas with 86 parcellations
    • Schaefer atlas with 220 parcellations

Replicating Results

# Clone the repository
git clone ...
cd ...

# Set up conda environment
conda install environment.yml

# Clean up demographics table
python