Medical Image Denoising in MRI Reconstruction Procedure

Dong Han, Ronny Velastegui

Abstract: With the rapid development of computer technology, deep learning can be used in nearly every field, and it always has the potential to achieve a high efficiency performance. Specifically, in the field of medical images, it makes doctors possible to distinguish and diagnose diseases in a more accurate way. Medical images like any other form of imaging techniques are affected by noise and artifacts. There are many types of noise, such as quantum, random, electric, and gaussian noise, etc. The presence of noise affects image clarity and may obstruct the recognition and analysis of diseases. The traditional image denoising method has much more limitation when came to medical images, and the results cannot meet some specific medical image standards. Hence, denoising of medical images in deep learning can be an important technique for further medical image processing. In this work, we conducted a deep learning method, which is a sparse dilated convolution neural network based on compressed sensing, for medical image denoising.

External IDs:dblp:conf/iccsa/HanV21