Guidance

Project participants: This project is jointly carried out by the TIMC-IMAG laboratory with Yohan Payan and Matthieu Chabanas, and the ICube laboratory by my implication. It has helped finance, half with the region, the thesis of Fanny Morin. This project has made it possible to develop a close collaboration with SINTEF in Trondheim in Norway, with whom we have set up an exchange of data on scientific methods. Finally, the Neurochirugie Department of the University Hospital of Grenoble provides us with clinical assistance through the involvement of Professor Olivier Palombi.

Project summary : Accurate localization is essential to reduce the risk of complications during surgical excision of a brain tumor. However, the soft tissues of the brain deform during the intervention under the influence of a phenomenon called “brain-shift” related to the loss of cerebrospinal fluid. Due to this deformation, the imaging data acquired before the operation no longer correspond to the reality of the operating theater. The location of the tumor becomes very difficult, which greatly complicates the procedure and therefore requires surgeons to acquire a great experience to practice it.

The objective of this thesis is to propose a biomechanical model of soft brain tissues to simulate their deformations and to adapt in real time the preoperative images to the patient’s intraoperative morphology (see Figure 4). Two key challenges are to work in a hyperelastic modeling framework, because of the very high nonlinearities of the tissues, and to interactively simulate the tissue loss corresponding to the removal of the tumor.

The brain model will be completed integrating the cerebrovascular tree, reconstructed from a preoperative magnetic resonance angiography. At different times of the procedure, the surgeon can then perform a Doppler ultrasound scan of the region of interest, to measure the deformations of the vascular tree. These deformations will be used as boundary conditions for the biomechanical model, which will then provide an estimate of the position and deformation of all anatomical structures. This procedure will be used first to evaluate the accuracy of the simulations and validate the model, then to assist the surgeon throughout the procedure. To be compatible with use in the operating room, the biomechanical model must meet the strong computation time constraints needed to update the imaging data interactively. Thus, simulation techniques on highly parallel GPU type architecture will also be studied in this thesis.

Publications: This project resulted in 3 publications [MCCP16, MRCP16, MCCP15]