Benchmark RadarLethe
Lethe is a benchmark for federated unlearning in medical imaging. It evaluates twelve methods across eight task families, including classification, segmentation, denoising, cross-modality synthesis, and vision-language question answering. Three forgetting granularities (hospital, class, patient) are assessed against a retrained gold standard on utility, privacy, and cost.
- Released
- 2026-08-02
- Readiness
- Paper only
- Primary field
- Health & Life Sciences
Why it matters
Existing unlearning evaluations primarily use natural images, leaving unclear whether methods transfer to clinical data. Lethe provides a shared protocol for medical imaging, enabling comparison of unlearning methods across diverse tasks and forgetting difficulties.
Motivation
Federated learning enables medical-imaging models to be trained across hospitals, and privacy law, most explicitly the GDPR ``right to be forgotten'', turns removing a hospital's, a class's, or a patient's influence from such a model into a federated unlearning problem.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.