Detection of Brain Strokes Using Microwave Tomography

Brain stroke is a major cause of disability and death worldwide. There are two types of stroke, ischemic or cerebral infarction (85% of cases) and hemorrhagic (15%). The diagnosis must be made quickly (within 3 to 4 hours after the onset of symptoms) to determine the nature of the stroke and proceed to treatment. Recent works have shown the modification of the complex permittivity according to the nature of stroke [1] in the microwave domain. We are interested here in the detection of brain strokes using microwave tomography. We present results obtained by electromagnetic simulations coupled to a realistic noise model of measurements. The forward problem is based on a massively parallel computing using domain decomposition method, and an inverse problem based on L-BFGS algorithm with a regularization based on total variation (TV).