Anne Micken

Radboudumc  Research software engineer

Project title: Grand-challenge.org: making AI application benchmarking for medical diagnostics accessible to the public 

The goal: This project aims at increasing the accessibility and usability of grand-challenge.org (GC), an open-source platform for developing machine learning solutions in biomedical imaging.  

The why: With the surge of artificial intelligence in biomedical imaging, there is an increased need for the open, fair and reproducible benchmarking of AI solutions. Grand-challenge.org is a platform that facilitates this benchmarking process through its challenge feature. It enables researchers to objectively compare the performance of AI solutions on a common, hidden data set. However, setting up a challenge is a complex task and can be overwhelming, especially for first time organizers. To make the challenge feature more accessible, we will organize regular workshops to train organizers in setting up their challenge. Participants will benefit from practical sessions during which they get to experience the challenge feature firsthand. A secondary goal is to create short hands-on tutorials for our documentation. By facilitating regular workshops and by enhancing our documentation, we aim to expedite the challenge launch process and lower the entry barrier for potential challenge organizers; improving overall user experience. 

Links

Github

LinkedIn