Overview
What this challenge is about.
You receive a curated public head-CT dataset (about 2,800 scans, slice-level labels for hemorrhagic stroke) and a held-out 600-scan hospital cohort. Train a 3D CNN or 2.5D slice-aggregating CNN for binary stroke detection. Add Monte-Carlo dropout for an uncertainty estimate. Evaluate AUROC, sensitivity at 95% specificity, and calibration (ECE + reliability diagrams) on the held-out cohort. Show how an 'uncertainty-aware' triage ordering compares with risk-only ordering on time-to-radiologist-read for the highest-priority 20% of cases. Deliver a 4-page memo.
The Brief
What you'll do, and what you'll demonstrate.
Train a stroke-detection model with calibrated uncertainty and show how an uncertainty-aware triage order changes time-to-read for top-priority cases.
Earning criteria — what you'll demonstrate
- Apply a 3D / 2.5D CNN to a real medical-imaging classification task
- Estimate model uncertainty via Monte-Carlo dropout
- Calibrate model probabilities and report ECE on held-out hospital data
- Translate uncertainty into operational triage-ordering claims
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
ML Researcher
Calibrated uncertainty + held-out-hospital evaluation are the rigorous portfolio piece radiology-AI labs hire ML researchers on.
This challenge sharpens
- medical-imaging
- uncertainty-quantification
- model-calibration
Computer Vision Engineer
3D / 2.5D CNN training and triage-ordering analysis are core CV-engineer work at any radiology-AI startup.
This challenge sharpens
- convolutional-neural-networks
- classification
- pytorch
Applied AI Scientist
Quantifying the operational benefit of uncertainty-aware triage in time-to-read terms is the applied-AI-scientist's daily work at clinical-AI companies.
This challenge sharpens
- uncertainty-quantification
- model-calibration
- medical-imaging