Deep MR Image Super-Resolution Using Structural Priors

TL;DR

This paper introduces a deep learning approach for MR image super-resolution that incorporates structural priors like low-rank and sharpness to improve image quality, especially with limited training data.

Contribution

It proposes a novel regularized CNN that integrates differentiable approximations of low-rank and sharpness priors for enhanced MR image super-resolution.

Findings

Improved super-resolution results on MR brain images.

Effective with limited training data.

Incorporation of priors enhances image sharpness and structure.

Abstract

High resolution magnetic resonance (MR) images are desired for accurate diagnostics. In practice, image resolution is restricted by factors like hardware, cost and processing constraints. Recently, deep learning methods have been shown to produce compelling state of the art results for image super-resolution. Paying particular attention to desired hi-resolution MR image structure, we propose a new regularized network that exploits image priors, namely a low-rank structure and a sharpness prior to enhance deep MR image superresolution. Our contributions are then incorporating these priors in an analytically tractable fashion in the learning of a convolutional neural network (CNN) that accomplishes the super-resolution task. This is particularly challenging for the low rank prior, since the rank is not a differentiable function of the image matrix (and hence the network parameters), an issue we address by pursuing differentiable approximations of the rank. Sharpness is emphasized by the variance of the Laplacian which we show can be implemented by a fixed {\em feedback} layer at the output of the network. Experiments performed on two publicly available MR brain image databases exhibit promising results particularly when training imagery is limited.