Autonomous Agents for Scientific Tasks - Sina Shahandeh, RADiCAIT
AI Engineer World's Fair 2026 · 19:23
AI medical imaging
RADiCAIT develops Insilico PET, an AI approach that produces synthetic PET-like images from standard CT scans. Its deep learning technology connects anatomical information with physiological information, aiming to reveal disease activity from routine imaging. The intended users are oncologists and radiologists, with hospital procurement officers as buyers. The goal is to make functional imaging available to clinics with CT scanners without requiring traditional PET equipment or radioactive injections for the generated images.
The company was founded by Seán Walsh, Regent Lee and Sina Shahandeh. Walsh, a medical physicist with a background in statistical modeling and medical imaging, serves as CEO; Shahandeh, a computational physicist, is CTO. Lee, an Oxford professor and surgeon-scientist, invented the underlying technology and serves as Chief Medical Innovation Officer. Their technical approach seeks to extract information from CT scans that is not visible to the human eye.
RADiCAIT remains in clinical development. In April 2026, the company reported four health systems signed up for clinical pilots: Mass General Brigham, UCSF, Emory and Hamilton Health. It was raising funds for a formal clinical trial and targeting its first reimbursed patient processing for mid-2027.
AI Engineer World's Fair 2026 · 19:23
Affiliations reflect their AIE appearances, not necessarily current employment.
The talk connects problem decomposition represented in Obsidian with Codex-driven experiments, qualitative image inspection, and hierarchical knowledge to organize iterative scientific research.
Affiliations reflect each recorded session, not necessarily current employment.