Reducing Motion Artifacts in Microwave-Based Hemorrhagic Stroke Detection

Paper in proceeding, 2019

The use of microwave technology for brain stroke detection opens up the opportunity for developing low-cost transportable devices which can be employed for rapid and prehospital detection of bleeding in patients with traumatic brain injury. Since the detection relies on changes in the scattering parameters of the tissues during pathological conditions, the movements in head or the antennas can deteriorate the detection accuracy. As a countermeasure to this problem, we introduce a technique for reducing the sensitivity of the detection system to motion artifacts. Via analyzing the scattering from an anatomically accurate model of human brain, we evidence that the channels between different pairs of antennas exhibit significantly different levels of sensitivity to the movement. Using this as a ground truth, we develop a generalized singular value decomposition (GSVD)-based approach for assigning appropriate weights to the channels between different antennas such that the probability of correct bleeding detection is maximized.