Continuous monitoring of brain hemodynamics is important so that changes in cerebral blood perfusion are quickly detected, speeding up the physician's decision-making regarding patient's treatment. Since the traditional techniques for monitoring, such as magnetic resonance angiography (MRA) or single-photon emission computed tomography (SPECT) can not be done at the patient's bedside and in a continuous way, Electrical Impedance Tomography (EIT) appears as a promising candidate for this task.This work will develop dynamic statistical models of cerebral hemodynamics as EIT prior information for monitoring and classification of patients' cerebral vascular accidents.Magnetic resonance angiography images of patients from a databank will be used to construct a three-dimensional model of the cerebral arterial tree. A 1D numerical model containing the main arteries, terminal resistances and vessel compliance will be used to run blood flow transient analysis. The resulting pressure and flow waves from the 1D model will finally be used to assemble the 3D dynamic model through the mapping between 1D model and 3D geometry. The dynamic model will be added to EIT algorithms based on the minimization of a functional, such as Newton-Raphson and Kalman filters.The resulting EIT images will be evaluated and compared with the literature using figures of merit already established in the literature. Synthetic data coming from numerical phantoms and data collected in humans, available in an open database will be used.
News published in Agência FAPESP Newsletter about the scholarship: