Radicait: AI-Powered Affordable Diagnostic Imaging | TechCrunch Disrupt 2025

The Challenges of Traditional PET Scans

Undergoing a PET (Positron Emission Tomography) scan can be a significant undertaking for patients. While crucial for cancer detection and monitoring disease progression, the procedure presents considerable logistical difficulties.

Preparation typically involves a four to six-hour fast prior to arriving at the hospital. Access can be particularly challenging for individuals in rural areas lacking local PET scanner availability. Following an injection of radioactive material, a waiting period of one hour is required to allow distribution throughout the body.

Accessibility and Limitations of Current Technology

The scanning process itself demands approximately 30 minutes of stillness from the patient while images are acquired by radiologists. Post-scan precautions necessitate avoiding close contact with vulnerable populations – the elderly, children, and pregnant women – for up to 12 hours due to residual radioactivity.

A key constraint lies in the geographical concentration of PET scanners. These machines rely on radioactive tracers produced in cyclotrons, requiring their use within a limited timeframe, thus restricting access for hospitals in rural and regional locations.

RADiCAIT: An AI-Powered Alternative

What if CT (Computed Tomography) scans, which are more readily available and cost-effective, could be transformed into PET scans using AI? This is the core concept behind RADiCAIT, an Oxford spinout that recently emerged from stealth mode with $1.7 million in pre-seed funding.

The Boston-based startup is a Top 20 finalist in the Startup Battlefield at TechCrunch Disrupt 2025 and is currently seeking $5 million to further its clinical trials. According to Sean Walsh, RADiCAIT’s CEO, the company’s goal is to replace a complex and expensive imaging solution with a simpler, more accessible alternative.

The Technology Behind the Innovation

RADiCAIT’s technology centers around a foundational model – a generative deep neural network developed in 2021 at the University of Oxford by a team led by Regent Lee, the startup’s co-founder and chief medical information officer.

This model functions by comparing CT and PET scans, identifying correlations, and recognizing patterns in their relationship. Sina Shahandeh, RADiCAIT’s chief technologist, describes this as translating anatomical structure into physiological function, connecting “distinct physical phenomena.”

The model is then trained to prioritize specific scan features, such as tissue types or abnormalities, through repeated learning with diverse examples. This process allows it to discern clinically significant patterns.

Mimicking PET Scan Accuracy

The final image presented to physicians is generated by the collaborative effort of multiple models. Shahandeh draws a parallel to Google DeepMind’s AlphaFold, noting that both systems translate one form of biological data into another.

Walsh asserts that RADiCAIT can mathematically demonstrate the statistical similarity between its synthetic PET images and those produced by conventional chemical PET scans. Clinical trials support this claim, showing equivalent decision-making quality for doctors using either imaging method.

Potential Impact on Diagnostic Imaging

While not intended to replace PET scans in all applications, such as radioligand therapy, RADiCAIT’s technology has the potential to render PET scans unnecessary for diagnostic, staging, and monitoring purposes.

Walsh explains that the limited supply of PET scanners struggles to meet the demand for both diagnostics and theragnostics – a combined approach of imaging and targeted therapy. RADiCAIT aims to address the diagnostic demand, freeing up PET scanners for theragnostic applications.

Clinical Trials and Future Expansion

RADiCAIT has initiated clinical pilots for lung cancer testing with prominent health systems, including Mass General Brigham and UCSF Health. The company is now pursuing an FDA clinical trial, which will be funded by the $5 million seed round.

Successful FDA approval will pave the way for commercial pilots to demonstrate the product’s commercial viability. RADiCAIT also plans to extend its approach to colorectal and lymphoma use cases.

Shahandeh believes RADiCAIT’s AI-driven approach to generating valid insights without the burdens of complex tests is widely applicable. He anticipates further innovations linking diverse fields like materials science, biology, chemistry, and physics.

Learn more about RADiCAIT at Disrupt, October 27 to 29 in San Francisco. Learn more here.