Third-party Demo of RADAR: Generalist AI for Abdominal CT Diagnosis

Upload a contrast-enhanced abdominal CT (.nii.gz) and RADAR will return per-finding positive probabilities across 146 findings in 18 anatomical structures.

Findings whose organ is not detected in the uploaded scan (or is cropped at the volume boundary) are marked N/A.

Sliding-window inference with ROI 96×256×384.

Research / demonstration use only. Do not use for clinical decisions. Model weights are released under CC BY-NC-SA 4.0.

Weights © 2026 DAMO Academy · Paper: Science 2026, eaec6129 · Code: alibaba-damo-academy/damo-radar.
More public cases for testing: PancancerCTSeg.

Demo built by Jun Ma.

Top 20 detected findings (highest probability)

Example CT volumes (click to run inference)