Laibel is a platform for data holders and data labellers who want to create cost-effective and accurate labels to enable medical artificial intelligence.
How might we make medical image
labelling cheaper, non-biased and faster?
During our interviews with multinational technology companies and AI startups, we found that medical image labeling has 3 pain points: medical image labeling requires an expensive, biased, and slow process.
and USER TESTING
After making paper prototypes and high-fidelity prototypes, we conducted 3 rounds of user testing with medical students and radiologists.
Several changes were needed: mostly concerning the overall navigation and organization of the platform, and the different drawing tools of the labeling feature.
We conducted an experiment where we asked 7 medical students to label lung scans. We ended up identifying 3 main insights: get better with time, improve with sample images, and want to use Laibel to study.