A Combined Partial Volume Reduction and Super-resolution Reconstruction for Magnetic Resonance Images

Magnetic resonance imaging provides a superior soft tissue contrast and a noninvasive means for automatic diagnosis of tissue pathogenesis. However, like most imaging modalities, it suffers from a compromise between the achievable spatial resolution, scan time efficiency, and signal-to-noise-ratio. To address this difficulty, super-resolution techniques have been proposed to enhance the spatial resolution of images in the post-acquisition steps. Most of those methods are proposed for nonmedical images. Thus, they do not consider the specific requirements of medical imaging in respect of data fidelity. In the present work, we propose a novel approach for super-resolution estimation that simultaneously reduces partial volume effects in order to enhance the edges without introducing artefactual effects to medical images. In this method, instead of using an edge-preserving preconditioner an interpolation based on the reverse diffusion process of material has been incorporated into the iterative estimation of images of higher spatial resolution. The proposed scheme outperforms the edge-preserving preconditioner in terms of image fidelity and speed of estimation.