Explainability First. We design modular networks (e.g., TMO-Net) that map predictions directly to gene pathways, decoding the “Black Box” for trusted mechanism discovery.

All-in-One Perception. Leveraging Self-Supervised Learning to unify genomic, pathological, and radiological data (CRCFound), overcoming annotation bottlenecks.

Perception to Action. Advancing from AR navigation to robot cognitive control, empowering surgeons with “transparent vision” and autonomous assistance.

Infinite Context. We decouple reasoning from knowledge, using Agents to orchestrate guidelines and evidence for transparent, traceable clinical decisions.