A multi-compartment fingerprinting model for non-invasive tumor cell characterization via diffusion MRI

Dausort, Manon [UCL] Delinte, Nicolas [UCL] Dessain, Quentin [UCL] Vanden Bulcke, Colin [UCL] Macq, Benoît [UCL]

Brain tumor tissue characteristics are important for treatment planning but are nowadays often recovered via invasive biopsy analysis. This work attempts to characterize the microstructural properties of tumor cells using diffusion MRI and Monte Carlo simulations to build a dictionary composed of several fingerprints, combining both a representation of axonal fibers and tumor cells. We demonstrate the use of our method on in-vivo brain data as a reliable estimation of tumor cell properties.