Fast Magnetic Resonance Image Reconstruction With Deep Learning Using An Efficientnet Encoder

This thesis aims to develop an efficient, deep network based method for Magnetic Resonance Imaging (MRI) acceleration through undersampled MR image reconstruction. Deep Neural Networks, particularly Deep Convolutional Networks, have been demonstrated to be highly effective in a wide variety of computer vision tasks, including MRI reconstruction. However, modern highly efficient encoder structures, such as the EfficientNet can potentially reduce reconstruction times further while improving reconstruction quality. To that end, we have developed a multi-channel U-Net MRI reconstruction network which uses an EfficientNet encoder and a custom asymmetric. The network was trained and tested using 5x undersampled multi-channel brain MR image data from the Calgary Campinas dataset and was found to outperform comparable traditional U-Net structures in terms of image quality metric analysis and basic visual comparison while achieving a four-fold reduction in inference time