SFMN Parkinson's

Challenge

Today, visual interpretation of DaT scans is reaching its limits - nearly one in five exams remains indeterminate, especially in early or atypical presentations. This challenge asks: what if AI could reduce that uncertainty and bring expert-level interpretation to every case?

The French College of Nuclear Medicine (SFMN) and DrivenData are launching a new medical imaging competition with a €25,000 prize pool.

Sign up below to get notified as soon as the challenge launches!
This email will only be used to let you know when the challenge is live. That's it.

HOW IT WORKS

Three steps to making an impact

Register & access the data

Sign up when the competition
goes live on DrivenData

Build your model & climb the leaderboard

Develop an AI classifier to distinguish normal from abnormal DaT scans.

Win prizes & advance the field

Top solutions earn a share of the €25,000 prize pool and will be open-sourced for the medical community to build on.

Participants will use computer vision techniques to classify DaT scans such as those above showing a normal scan (left) and a pathological scan (right).

About the challenge

Classify DaT scans for Parkinson's disease

DaT imaging can reveal Parkinson's disease with remarkable clarity, but reliable interpretation depends on specialist readers that aren't available everywhere.

This challenge invites solvers of all backgrounds, including data scientists, ML engineers, medical imaging researchers, and nuclear medicine specialists, to advance open-source models for automatic DaT scan classification for Parkinson's diagnosis.