How ramp-kits/autism works
This repository is an IMaging-PsychiAtry Challenge starting kit for classifying autism spectrum disorder from participant, anatomical MRI, and functional MRI data. The intended workflow is to assemble train/test features and labels, extract functional connectivity and/or anatomical features, fit a classifier, and produce predictions. It also includes utilities for downloading precomputed fMRI time series and preparing fCON data. The submission workflow’s runtime wiring is not shown in the excerpts, so links from the challenge definition to the example submission components are marked unverified.
Main entry points: problem.py Problem definition · data Challenge metadata · download_data.py Time-series downloader · feature_extractor.py Functional features
Components 13
Challenge workflow
- Problem definition
problem.py
MRI preprocessing
- Time-series extraction
preprocessing/extract_time_series.py - fCON preparation
preprocessing/prepare_fcon.py
Submission models
- Functional features
submissions/starting_kit_functional/feature_extractor.py - Functional classifier
submissions/starting_kit_functional/classifier.py - Anatomical features
submissions/starting_kit_anatomy/feature_extractor.py - Anatomical classifier
submissions/starting_kit_anatomy/classifier.py - Combined features
submissions/combine_anatomy_functional/feature_extractor.py - Combined classifier
submissions/combine_anatomy_functional/classifier.py
Data assets
- Challenge metadata
data - Time-series downloader
download_data.py
Other components
- Challenge user
- Zenodo atlas archives
Connections 11
- Challenge user → Problem definition: runs challenge
- Problem definition → Challenge metadata: loads metadata
problem.py - Problem definition → Functional features: configures workflow
- Problem definition → Anatomical features: configures workflow
- Problem definition → Combined features: configures workflow
- Functional features → Functional classifier: provides features
- Anatomical features → Anatomical classifier: provides features
- Combined features → Combined classifier: provides features
- Time-series extraction → Challenge metadata: writes time series
preprocessing/extract_time_series.py - Time-series downloader → Zenodo atlas archives: downloads archives
download_data.py - Time-series downloader → Challenge metadata: stores atlas data
download_data.py