Super-Resolution of Brain MRI via U-Net Architecture

This paper proposes a U-Net-based deep learning architecture for the task of super-resolution lower resolution brain magnetic resonance images (MRI). The proposed system, called MRI-Net, is designed to learn mapping between low-resolution and high-resolution MRI images. system trained using 50-800 2D scans, depending on architecture, evaluated peak signal-to-noise ratio (PSNR) metrics 10 randomly selected U-Net outperforms current state-of-the-art networks in terms PSNR when with 3 x downsampling index. system's ability super-resolve scans has potential enable physicians detect pathologies better perform wider range applications. symmetrical pipeline used this study allows generically representing highlight proof concept approach. implemented PyTorch 1.9.0 NVIDIA GPU processing speed up training time. promising tool medical applications MRI, which can provide accurate high-quality diagnoses treatment plans. approach reduce costs associated by providing solution enhancing image quality scans.