Abstract
High-resolution medical images are desirable for practical application of high-resolution displays, however, it takes a long scanning time to improve image resolution in MRI. In this study, we applied and evaluated the sparse coding super-resolution (ScSR), which is one of the image processing techniques to obtain high-resolution images, for enhancing image resolution in MRI. For evaluation, T1-weighted images (T1), T2-weighted images (T2), fluid attenuated IR images (FLAIR), and time of flight images (TOF) were used as the test datasets. We up-sampled all images up to twice and compared the quality of the ScSR scheme and bilinear, bicubic, and lanczos interpolations, which are the traditional interpolation schemes. The image quality was evaluated by measuring peak signal-to-noise ratio (PSNR) and structural similarity (SSIM). As a result, PSNR and SSIM of the ScSR were significantly higher (p<0.05) than those of other three interpolations in T1 (original and contrast-enhanced), T2 and FLAIR. In TOF, PSNR and SSIM of the ScSR were higher than those of other three interpolations for all images. These results suggest that the ScSR schemes markedly improve PSNR and SSIM in T1, T2, FLAIR and TOF, in comparison with the traditional interpolation schemes.
