Bilinear Model Predictive Control Framework of the OncoReach, a Tendon-Driven Steerable Stylet for Brachytherapy
A new AI control system guides steerable needles around organs with sub-2mm accuracy for safer radiation therapy.
A research team has published a breakthrough in robotic surgery with the OncoReach, a tendon-driven steerable stylet designed for brachytherapy—a cancer treatment involving the implantation of radioactive seeds. The key innovation is a bilinear Model Predictive Control (MPC) framework, an AI-driven control system that calculates optimal movements in real-time. Unlike previous systems built for custom needles, this framework is designed for a stylet that fits inside standard, commercially available brachytherapy needles, making clinical adoption more feasible. The AI model translates high-level commands (insertion speed and bending rates) into precise physical actions by controlling tendon tensions.
In validation experiments using tissue-mimicking phantom material, the system demonstrated its potential. For fixed-target positioning, the closed-loop AI control achieved an error as low as 1.45mm, which is just 1.7% of the inserted needle length. This high precision is critical for avoiding damage to sensitive anatomical structures like blood vessels or nerves during procedures. However, the paper also notes challenges, with errors in certain bending directions reaching 8.3mm, highlighting areas for improvement in sensor calibration. Overall, the research proves the feasibility of using AI for both fixed-target positioning and moving-target trajectory tracking in a clinically compatible system, marking a significant step toward more adaptable and safer minimally invasive cancer treatments.
- The OncoReach system uses a bilinear Model Predictive Control (MPC) AI framework to steer a tendon-driven stylet inside standard brachytherapy needles.
- In physical tests, the AI-controlled needle achieved a precision of 1.45mm (1.7% error) for fixed-target tracking in tissue-mimicking material.
- The technology enables curved needle paths to navigate around organs, addressing a major limitation in current brachytherapy procedures for prostate and other cancers.
Why It Matters
This AI-driven precision steering could make minimally invasive cancer radiation therapy safer by allowing doctors to avoid critical organs with sub-millimeter accuracy.