Deep learning based noise level classification of medical images

Abstract

In recent years, medical image technology is widely used in medical field. However, the process of imaging, storage and transmission often make the image quality reduced and affect the visual and post-processing effect, and the degradation of medical image often leads to the interference of noise. Therefore, obtaining the level of noise in medical images is an important part in image quality improvement. In order to obtain the noise level in medical image, a novel image noise level classification network based on deep learning is designed, which incorporates inception structure and dense blocks to make full use of their advantages to extract the features of noise. The inception structure is used to extract the features of noise under different resolutions to get the features of different receptive fields. Meanwhile, the dense block structure is used to take advantage of feature reuse to ensure noise features transfer across the network. Experiments on lung CT images show that the classification accuracy of the proposed method is 99.5%. The method proposed has a good effect in the application of noise level classification and provides a reliable noise prior for the image enhancement using SRMD (super-resolution of MAP and dimensionality stretching) method.