Low-dose CT lung images denoising based on multiscale parallel convolution neural network
- PDF / 3,831,179 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 78 Downloads / 223 Views
ORIGINAL ARTICLE
Low-dose CT lung images denoising based on multiscale parallel convolution neural network Xiaoben Jiang1 · Yan Jin1
· Yu Yao1
Accepted: 12 October 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract The continuous development and wide application of CT in medical practice have raised public concern over the associated radiation dose to the patient. However, reducing the radiation dose may result in increasing the noise and artifacts, which may adversely interfere with the judgment and belief of radiologists. Therefore, we propose a low-dose CT denoising model based on multiscale parallel convolution neural network to improve the visual effect. Residual learning is utilized to reduce the difficulty of network learning, and batch normalization is adopted to solve the problem of performance degradation due to the increase in neural network layers. Specifically, we introduce the dilated convolution to expand the receptive field by inserting weights of zero in the standard convolution kernel, while not increasing the extra parameters. Furthermore, the multiscale parallel method is utilized to extract multiscale detail features from lung images. Compared to the traditional methods such as Wiener filter, NLM, and models based on CNN, e.g., SCNN, DnCNN, our extensive experimental results demonstrate that our proposed model (CT-ReCNN) can not only reduce the LDCT lung images noise level, but also retain more exact information as well. Keywords CT lung image denoising · Multiscale parallel · Convolution · Neural network · Dilated convolution · Residual learning
1 Introduction With the development of computed tomography (CT) scanning technology, CT diagnosis has played an important role in the early screening of lung diseases. High levels of radiation are generated during CT scanning, which are harmful to the human body [1]. Therefore, low-dose CT (LDCT) [2] is proposed to reduce radiation. However, it also leads to degraded lung image quality, which can interfere with the diagnosis of the disease by the radiologist. Therefore, denoising and restoring the LDCT lung image is of great significance. There are several traditional algorithms for improving the quality of LDCT images, such as image reconstruction algorithms [3], projection domain denoising algorithms [4], and image domain denoising algorithms [5]. Filtered back projection (FBP) [6] is one of the most common image reconstruction algorithms for image LDCT images denois-
B 1
Yan Jin [email protected] College of Information Engineering, Zhejiang University of Technology, Hangzhou 310023, Zhejiang Province, China
ing, which has a high resolution. However, it also increases the complexity of the algorithm and the computational time while improving the quality of LDCT image quality. The optimal variable of projection domain denoising algorithm is projection image, such as the bilateral filtering method [7], the adaptive balanced mean filtering method [8], and the adaptive convolutional filtering method [9]. The projection domain
Data Loading...
