2021-2024

Diagnosis is one of the most important acts performed by a practitioner during surgery. However, current in-situ diagnostic methods are not real-time and therefore time-consuming. In this thesis, we developed a new multimodal optical method based on the principle of full-field optical coherence tomography (FF-OCT), capable of performing real-time diagnostics with micrometric resolution, elastography, Doppler imaging, etc.

• PhD student: E. Martins Seromenho
• Co-supervisors: N. Bahlouli, S. Facca & A. Nahas, ICube & HUS

Autonomous Endovascular Navigation

Endovascular interventions are pivotal in treating cardiovascular diseases, but they involve acquiring fluoroscopic images that expose  both clinicians and patients to X-rays, raising tumor formation risks.  Current robotic systems mitigate radiation exposure only for clinicians.  This project aims to develop a Deep Reinforcement Learning-based  controller for these robots. The algorithm relies solely on optical  fibers for device shape estimation, a limited set of fluoroscopic images, and pre-operative anatomy reconstruction.

• PhD student: V. Scarponi
• Co-supervisors: S. Cotin, F. Nageotte & M. Duprez, INRIA & ICube