MedDeblur: Medical Image Deblurring with Residual Dense Spatial-Asymmetric Attention

Medical image acquisition devices are susceptible to producing blurry images due respiratory and patient movement. Despite having a notable impact on such blind-motion deblurring, medical deblurring is still underexposed. This study proposes an end-to-end scale-recurrent deep network learn the from multi-modal images. The proposed comprises novel residual dense block with spatial-asymmetric attention recover salient information while learning deblurring. performance of methods has been densely evaluated compared existing methods. experimental results demonstrate that method can remove blur without illustrating visually disturbing artifacts. Furthermore, it outperforms in qualitative quantitative evaluation by noticeable margin. applicability also verified incorporating into various analysis tasks as segmentation detection. helps accelerate removing inputs.