I am honoured to be featured in an article published by Inria on the use of digital sciences — AI, simulation and robotics — to improve the diagnosis of prostate cancer.

The Challenge: Precision in Prostate Biopsy

Prostate cancer is one of the most common cancers in men. Confirming the diagnosis often requires a biopsy: a needle is inserted into the prostate to collect tissue samples from suspicious regions identified by MRI. The procedure sounds simple, but performing it with sufficient precision — avoiding the urethra, hitting a millimetre-scale target, and accounting for the soft deformations of prostate tissue — is a significant engineering challenge.

Our Contribution: Deformable Robotics meets Medical Imaging

As part of the ROBOCOP project (2017–2021), our team at the DEFROST group (Inria / CRIStAL UMR 9189 / University of Lille) developed a robotic system that carries a flexible ultrasound probe and biopsy needle. Because the robot’s end-effector is itself soft and deformable, controlling it requires mathematical models that go beyond classical rigid-body robotics.

We used Cosserat rod theory and finite element methods (FEM) — implemented in the SOFA framework and our Cosserat plugin — to model the deformations in real time. These models allow the robot to:

  • Account for the bending and twisting of the flexible probe
  • Compensate for prostate tissue deformation during insertion
  • Align the needle with targets identified in ultrasound or MRI imagery

Video

Ouvrir la vidéo sur la Mediatheque Inria

Read the Full Article

The Inria article gives an accessible overview of the multiple facets of this problem: machine learning for lesion detection, digital modelling for probe control, and the broader clinical context.

Read the Inria Article (FR)   Read in English

This work connects to ongoing projects in the DEFROST team:

  • 🔬 COSSEROOTS — extending Cosserat-based control to growing soft robots
  • 🤖 CONECT — autonomous needle insertion guidance
  • 🧲 Active Prostate Phantom — hardware phantom for validating biopsy robots

The Cosserat plugin and the SoftRobots.Inverse plugin are both open-source and freely available for the research community.