Task 3: Self-Supervised Multi-Abnormality Localization

Participants synthesize anatomically plausible 3‑D chest CT volumes directly from radiology text prompts, aiming for high visual fidelity, realistic Hounsfield distributions, and tight semantic alignment; success is measured with CT‑adapted generative metrics (FVDI3D, FVDCT‑Net, CT‑CLIP, FID) and ranked via the same permutation‑based point system.