Medical image denoising using convolutional dual-decoder autoencoder†
Eon Seung Seong1 · Ji Hye Heo2 · Seong Hyun Han3 · Dong Hoon Lim4
1234Department of Information Statistics, Gyeongsang National University
Correspondence to: † This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (No. NRF-2022R1A2C1011140).
1 Undergraduate, Dept. of Information Statistics, Gyeongsang National University, Jinju 52828, Korea.
2 Graduate student, Dept. of Information Statistics, Gyeongsang National University, Jinju 52828, Korea.
3 Undergraduate, Dept. of Information Statistics, Gyeongsang National University, Jinju 52828, Korea.
4 Professor, Dept. of Information Statistics and RINS, Gyeongsang National University, Jinju 52828, Korea. E-mail: dhlim@gnu.ac.kr
1 Undergraduate, Dept. of Information Statistics, Gyeongsang National University, Jinju 52828, Korea.
2 Graduate student, Dept. of Information Statistics, Gyeongsang National University, Jinju 52828, Korea.
3 Undergraduate, Dept. of Information Statistics, Gyeongsang National University, Jinju 52828, Korea.
4 Professor, Dept. of Information Statistics and RINS, Gyeongsang National University, Jinju 52828, Korea. E-mail: dhlim@gnu.ac.kr
Received September 5, 2022; Revised September 27, 2022; Accepted October 10, 2022.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
- Abstract
- Noise existing in medical images such as CT images and MRI images is a cause of obstructing accurate diagnosis of disease names of patients. Therefore, effective noise removal is very important to increase the utilization value of medical images. In this paper, noise is removed using the CDDAE (Convolutional Dual-Decoder Autoencoder) model, which is a variant of the deep learning CAE (Convolutional Autoencoder) model. The existing CAE model is an autoencoder (AE) structure composed of a single encoder and a single decoder, whereas the CDDAE model is an AE structure composed of a single encoder and dual decoders. It is designed to restore as close to the original image as possible. To evaluate the performance of the CDDAE model proposed in this paper, CT images and MRI images damaged by various noises such as Gaussian noise, impulse noise, and speckle noise were tested. As a result of the performance experiment, the CDDAE model showed good results in terms of MAE (Mean Absolute Error), PSNR (Peak Signal-to-Noise Ratio), and SSIM (Structural Similarity Index Measure), which are existing deep learning models.
- Keywords : Convolutional dual-decoder autoencoder (CDDAE), deep learning, medical image, noise reduction.
Full Text(PDF) Free
