A Quantitative Microwave Imaging Approach for Brain Stroke Classification Based on the Generalized Tikhonov Regularization

Early diagnosis of stroke type by imaging is one of the most important tasks for stroke patients. In this article, we propose a new approach to reconstruct the brain image with high accuracy and quality. In this approach, first we use the Born iterative method to reconstruct the brain image. Then by comparing this image with a set of MRI-based brain images, using structural similarity index measure criterion, we choose the best one as reference image. Finally, we reconstruct the brain image by distorted Born iterative method or Born iterative method along with generalized Tikhonov regularization using the reference image. The reconstructed images are compared with those that obtained based on Tikhonov regularization. These comparisons demonstrate that the accuracy and quality of images in the proposed approach are significantly increased.