High Quality Brain Image Reconstruction Based on DBIM and U-Net

Diagnosing the type of stroke through brain image reconstruction is a widely employed method that initiates the treatment of stroke patients. Microwave imaging has emerged as a recent technique for image reconstruction, yet it suffers from the drawback of low resolution. In this study, we propose a novel approach to enhancing the quality of reconstructed brain images by leveraging the power of the U-Net network. Our methodology involves reconstructing brain 2D images using the distorted Born iterative method and treating these images as input to the U-Net network. We conduct the reconstruction process on brain images obtained at two different noise levels. In both noise levels, we observe a significant improvement in the quality of the reconstructed images.