Abstract
In order to suppress noise and artifacts in low-dose computed tomography (LDCT), various deep learning techniques, especially encoder-decoder networks, have been introduced to improve the quality of LDCT images. However, in the encoder-decoder convolutional neural network, fixed-size convolution kernel, continuous down-sampling operation, and the mean square error (MSE) objective function are used, which cause problems such as low utilization of image information, image information loss, and over-smoothing of denoised image. To improve the quality of reconstructed CT images, in this paper, a LDCT image denoising network based on residual multi-scale feature extraction and hybrid loss function is proposed. On the one hand, the multi-scale feature extraction module is designed and introduced into the residual connection to improve the utilization of image feature information; on the other hand, zero padding is used to solve the information loss problem caused by continuous down-sampling operations, and batch normalization (BN) layer is used to alleviate the over-fitting problem caused by network deepening. In addition, a hybrid loss function consisting of MSE loss, structural similarity (SSIM) loss, and perceptual loss is introduced to generate denoised images with high relevance to human perception. Experimental results show that the proposed algorithm can not only improve the quality of denoised images, but also greatly improve the computational speed compared with the state of the art algorithms.
Similar content being viewed by others
Data availability
The data and material can be made available on request.
Code availability
The code can be made available on request.
References
Luo, L., Hu, Y., Chen, Y.: Research status and prospect for low-dose ct imaging. J. Data Acquisition Process. 30(1), 224–234 (2015)
Aliasharzadeh, A., et al.: A survey of computed tomography dose index and dose length product level in usual computed tomography protocol. J. Cancer Res. Ther. 14(3), 549 (2018)
Brenner, D.J., Hall, E.J.: Computed tomography—an increasing source of radiation exposure. N. Engl. J. Med. 357, 2277–2284 (2007)
Zhang, Y., Salehjahromi, Z., Yu, H.: Tensor decomposition and non-local means based spectral CT image denoising. J. J. X-ray Sci. Technol. 27(3), 397–416 (2012)
Luo, Z., Yin, Y., Bi, S.: Denoising algorithm for CT oral image based on Bayesian threshold and Non-local mean. In: C. Computers and Software Engineering (AEMCSE), pp. 288–292. IEEE (2020)
Yahya, A., Tan, J., Su, B., et al.: BM3D image denoising algorithm based on an adaptive filtering. J. Multimedia Tools Appl. 79, 20391–20427 (2020)
Chen, H., Zhang, Y., Kalra, M.K., et al.: Low-dose ct with a residual encoder-decoder convolutional neural network. IEEE Trans. Med. Imaging 36(12), 2524–2535 (2017)
Dong, J., Roth, S., Schiele, B.: Deep wiener deconvolution: Wiener meets deep learning for image deblurring. Adv. Neural. Inf. Process. Syst. 33(12), 1048–1059 (2022)
Shin, S.-Y., Kim, D.-M., Suh, J.-W.: Image denoiser using convolutional neural network with deconvolution and modified residual network. IEICE Trans. Inf. Syst. 102(8), 1598–1601 (2019)
You, H., Yu, L., Tian, S., et al.: Mc-net: Multiple max-pooling integration module and cross multi-scale deconvolution network. Knowl.-Based Syst. 231, 107456 (2021)
Liu, H., Jin, X., Liu, L.: Low-dose CT image denoising based on improved DD-net and local filtered mechanism. Comput. Intell. Neurosci. (2022)
Yang, L., Shangguan, H., Zhang, X., et al.: High-frequency sensitive generative adversarial network for low-dose ct image denoising. IEEE Access. 8, 930–943 (2019)
Rassil, A., Chougrad, H., Zouaki, H.: Augmented graph neural network with hierarchical global-based residual connections. Neural Netw. 150, 149–166 (2022)
Yang, Q., Yan, P., Zhang, Y., et al.: Low-dose ct image denoising a generative adversarial network with wasserstein distance and perceptual loss. IEEE Trans. Med. Imaging 37(6), 1348–1357 (2018)
Li, M., Du, Q., Duan, L., et al.: Incorporation of residual attention modules into two neural networks for low-dose ct denoising. Med. Phys. 48(6), 2973–2990 (2021)
Liang, T., Jin, Y., Li, Y., et al.: EDCNN: edge enhancement-based densely connected network with compound loss for low-dose CT denoising. In: IEEE International Conference on Signal Processing (ICSP), pp. 193–198 (2020)
Wang,D., Fan, F., Wu, Z., et al.: CTformer: convolution-free Token2Token dilated vision transformer for low-dose CT denoising. arXiv e-prints (2022)
Ma, Y., Wei, B., Feng, P., et al.: Low-dose ct image denoising using a generative adversarial network with a hybrid loss function for noise learning. IEEE Access. 8, 65719–65729 (2020)
Kyung, S., Won, J., Pak, S., et al.: Mtd-Gan: Multi-task discriminator based generative adversarial networks for low-dose CT denoising. In: International Workshop on Medicine Learning for Medical Image Reconstruction, pp 133–144. Springer (2022)
Huang, Z., Zhang, J., Zhang, Y., et al.: Du-gan: generative adversarial networks with dual-domain u-net-based discriminators for low-dose ct denoising. IEEE Trans. Instrum. Meas. 71, 1–12 (2021)
Naseem, R., Alaya Cheikh, F., Beghdadi, A., et al.: Cross-modal guidance assisted hierarchical learning based siamese network for mr image denoising. Electronics 10(22), 2855 (2021)
Wu, H., Qu, Y., Lin, S., et al.: Contrastive learning for compact single image dehazing. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 10551–10560 (2021)
Tang, Y., Gong, W., Chen, X., et al.: Deep inception-residual laplacian pyramid networks for accurate single-image super-resolution. IEEE Trans. Neural Netw. Learn. Syst. 31(5), 1514–1528 (2019)
Han, M., Shim, H., Baek, J.: Perceptual CT loss: implementing CT image specific perceptual loss for CNN-based low-dose CT denoiser. IEEE Access. 10, 62412–62422 (2022)
Moen, T.R., Chen, B., Holmes III, D.R., et al.: Low-dose ct image and projection dataset. Med. Phys. 48(2), 902–911 (2021)
Zhang, Z.: Improved adam optimizer for deep neural networks. In: 2018 IEEE/ACM 26th International Symposium on Quality of Service (IWQoS), IEEE, pp. 1–2 (2018)
Shan, H., Zhang, Y., Yang, Q., et al.: 3-d convolutional enccoder-decoder network for low-dose ct via transfer learning from a 2-d trained network. IEEE Trans. Med. Imaging 37(6), 1522–1534 (2022)
Funding
This work was supported in part by the Natural Science Foundation of Shanxi Province under Grant 202203021222015, in part by the State Council and the central government guide local funds of China under Grant YDZX20201400001547, in part by the postgraduate practice and innovation program of Shanxi Province under Grant 2023SJ011.
Author information
Authors and Affiliations
Contributions
AMH develop the idea and accomplished the manuscript writing and performed the experiments. All authors accomplished the manuscript revising. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Conflict of interest
We declare that we have no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Jia, L., Huang, A., He, X. et al. A residual multi-scale feature extraction network with hybrid loss for low-dose computed tomography image denoising. SIViP 18, 1215–1226 (2024). https://doi.org/10.1007/s11760-023-02809-3
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11760-023-02809-3