Glioblastoma is the most common primary malignant brain tumor in the adult population with a very poor prognostic. The combination of surgery, chemotherapy and radiotherapy are used as treatment. Magnetic Resonance Imaging (MRI) is used to delineate the disease before surgery and to monitor patients after surgery. Two of the challenges in interpreting these images are the delineation of infiltrative tumor before surgery, and the detection of recurrence after the surgery. While diffusion MRI (dMRI) can help with these two tasks, the lack of a robust biophysical MRI model limits the optimization and the validation of this technique. We propose to develop a new model based on histological brain tumor slices which are the most realistic representation of the tumor geometry. This will result in a more accurate and robust dMRI model, allowing to optimize the pulsed-sequence on the machine and the post-processing of the images. This will result in a better delineation of the infiltrative tumor, which will guide neurosurgeons to remove this tumor component during the surgery – a technique called supramaginal resection that has been shown to improve survival. This will also decrease morbidity in glioblastoma patients from prompt detection of recurrence and will reduce post-surgical functional deficit from prevention of unnecessary removal of healthy brain tissue.
