Multi-branch network for brain image denoising based on U-shaped dynamic convolution and multi-scale feature extraction

Huimin Qu; Haiyan Xie; Qianying Wang; Huimin Qu; Haiyan Xie; Qianying Wang

doi:10.3934/era.2025258

Electronic Research Archive

2025, Volume 33, Issue 9: 5794-5828. doi: 10.3934/era.2025258

Previous Article Next Article

Research article Special Issues

Multi-branch network for brain image denoising based on U-shaped dynamic convolution and multi-scale feature extraction

School of Science, Dalian Maritime University, Dalian 116026, China

Received: 03 August 2025 Revised: 14 September 2025 Accepted: 19 September 2025 Published: 26 September 2025

Brain images are often disturbed by noise during acquisition, compromising subsequent medical analysis and reducing diagnostic accuracy. Effective denoising that preserves the structural details is therefore essential. This essay proposes a multi-branch convolutional neural network for brain image denoising, integrating a U-shaped network, dynamic convolution, and multi-scale feature extraction. The model includes an attention-enhanced encoder–decoder to improve feature representation, dynamic convolution with a sparse mechanism for adaptive global modeling, multi-scale dense residual blocks with depth-separable convolution to capture local details efficiently, and a multi-branch fusion strategy to process features at different scales in parallel and refine them. Experiments on brain image datasets demonstrate that the proposed method achieves superior denoising performance, effectively suppressing noise while retaining fine structural details. In conclusion, the network significantly enhances the quality of brain images and shows potential for improving the accuracy of subsequent medical image analysis and diagnosis.
- brain image denoising,
- U-Net,
- dynamic convolution kernel,
- multi-scale feature extraction,
- multi-branch network
Citation: Huimin Qu, Haiyan Xie, Qianying Wang. Multi-branch network for brain image denoising based on U-shaped dynamic convolution and multi-scale feature extraction[J]. Electronic Research Archive, 2025, 33(9): 5794-5828. doi: 10.3934/era.2025258

Related Papers:

Abstract

Brain images are often disturbed by noise during acquisition, compromising subsequent medical analysis and reducing diagnostic accuracy. Effective denoising that preserves the structural details is therefore essential. This essay proposes a multi-branch convolutional neural network for brain image denoising, integrating a U-shaped network, dynamic convolution, and multi-scale feature extraction. The model includes an attention-enhanced encoder–decoder to improve feature representation, dynamic convolution with a sparse mechanism for adaptive global modeling, multi-scale dense residual blocks with depth-separable convolution to capture local details efficiently, and a multi-branch fusion strategy to process features at different scales in parallel and refine them. Experiments on brain image datasets demonstrate that the proposed method achieves superior denoising performance, effectively suppressing noise while retaining fine structural details. In conclusion, the network significantly enhances the quality of brain images and shows potential for improving the accuracy of subsequent medical image analysis and diagnosis.

References

[1]	M. Rahimpour, J. Bertels, A. Radwan, H. Vandermeulen, S. Sunaert, D. Vandermeulen, et al., Cross-modal distillation to improve MRI-based brain tumor segmentation with missing MRI sequences, IEEE Trans. Biomed. Eng., 69 (2022), 2153–2164. https://doi.org/10.1109/TBME.2021.3137561 doi: 10.1109/TBME.2021.3137561
[2]	V. Jain, A. Datar, Y. K. Jain, Medical image denoising using optimal attention block-based pyramid denoising network, Biomed. Signal Process. Control, 108 (2025), 107794. https://doi.org/10.1016/j.bspc.2025.107794 doi: 10.1016/j.bspc.2025.107794
[3]	S. Jiang, L. Guo, G. Cheng, X. Chen, C. Zhang, Z. Chen, Brain extraction from brain MRI images based on Wasserstein GAN and O-Net, IEEE Access, 9 (2021), 136762–136774. https://doi.org/10.1109/ACCESS.2021.3113309 doi: 10.1109/ACCESS.2021.3113309
[4]	A. Sanaat, H. Arabi, I. Mainta, V. Garibotto, H. Zaidi, Projection space implementation of deep learning–guided low-dose brain PET imaging improves performance over implementation in image space, J. Nucl. Med., 61 (2020), 1388–1396. https://doi.org/10.2967/jnumed.119.239327 doi: 10.2967/jnumed.119.239327
[5]	F. Hashimoto, Y. Onishi, K. Ote, H. Tashima, A. J. Reader, T. Yamaya, Deep learning-based PET image denoising and reconstruction: a review, Radiol. Phys. Technol., 17 (2024), 24–46. https://doi.org/10.1007/s12194-024-00780-3 doi: 10.1007/s12194-024-00780-3
[6]	A. Kaur, G. Dong, A complete review on image denoising techniques for medical images, Neural Process. Lett., 55 (2023), 7807–7850. https://doi.org/10.1007/s11063-023-11286-1 doi: 10.1007/s11063-023-11286-1
[7]	A. Bal, M. Banerjee, P. Sharma, M. Maitra, An efficient wavelet and curvelet-based PET image denoising technique, Med. Biol. Eng. Comput., 57 (2019), 2567–2598. https://doi.org/10.1007/s11517-019-02014-w doi: 10.1007/s11517-019-02014-w
[8]	G. Elaiyaraja, N. Kumaratharan, T. C. S. Rao, Fast and efficient filter using wavelet threshold for removal of Gaussian noise from MRI/CT scanned medical images/color video sequence, IETE J. Res., 68 (2022), 10–22. https://doi.org/10.1080/03772063.2019.1579679 doi: 10.1080/03772063.2019.1579679
[9]	H. V. Bhujle, B. H. Vadavadagi, NLM based magnetic resonance image denoising–A review, Biomed. Signal Process. Control, 47 (2019), 252–261. https://doi.org/10.1016/j.bspc.2018.08.031 doi: 10.1016/j.bspc.2018.08.031
[10]	A. K. Singh, Major development under Gaussian filtering since unscented Kalman filter, IEEE/CAA J. Autom. Sin., 7 (2020), 1308–1325. https://doi.org/10.1109/JAS.2020.1003303 doi: 10.1109/JAS.2020.1003303
[11]	W. Liang, J. Long, K. C. Li, J. Xu, N. Ma, X. Lei, A fast defogging image recognition algorithm based on bilateral hybrid filtering, ACM Trans. Multimedia Comput. Commun. Appl., 17 (2021), 42. https://doi.org/10.1145/3391297 doi: 10.1145/3391297
[12]	R. Liu, Y. Li, H. Wang, J. Liu, A noisy multi-objective optimization algorithm based on mean and Wiener filters, Knowl.-Based Syst., 228 (2021), 107215. https://doi.org/10.1016/j.knosys.2021.107215 doi: 10.1016/j.knosys.2021.107215
[13]	B. Shi, M. Li, Y. Lou, Adaptively weighted difference model of anisotropic and isotropic total variation for image denoising, J. Nonlinear Var. Anal., 7 (2023), 563–580. https://doi.org/10.23952/jnva.7.2023.4.07 doi: 10.23952/jnva.7.2023.4.07
[14]	X. Xu, Q. Wang, L. Guo, J. Zhang, S. Ding, FEMRNet: Feature-enhanced multi-scale residual network for image denoising, Appl. Intell., 53 (2023), 26027–26049. https://doi.org/10.1007/s10489-023-04895-9 doi: 10.1007/s10489-023-04895-9
[15]	L. Xiang, Y. Qiao, D. Nie, L. An, W. Lin, Q. Wang, et al., Deep auto-context convolutional neural networks for standard-dose PET image estimation from low-dose PET/MRI, Neurocomputing, 267 (2017), 406–416. https://doi.org/10.1016/j.neucom.2017.06.048 doi: 10.1016/j.neucom.2017.06.048
[16]	K. Zhang, W. Zuo, Y. Chen, D. Meng, L. Zhang, Beyond a Gaussian denoiser: Residual learning of deep CNN for image denoising, IEEE Trans. Image Process., 26 (2017), 3142–3155. https://doi.org/10.1109/TIP.2017.2662206 doi: 10.1109/TIP.2017.2662206
[17]	C. Tian, Y. Xu, L. Fei, J. Wang, J. Wen, N. Luo, Enhanced CNN for image denoising, CAAI Trans. Intell. Technol., 4 (2019), 17–23. https://doi.org/10.1049/trit.2018.1054 doi: 10.1049/trit.2018.1054
[18]	B. Pan, D. Pu, L. Fu, F. Zhang, Z. Ping, H. Liu, et al., Deep learning enhanced FDG PET acquisition: a true low dose evaluation, J. Nucl. Med., 64 (2023), 810.
[19]	K. Spuhler, M. Serrano-Sosa, R. Cattell, C. DeLorenzo, C. Huang, Full-count PET recovery from low-count image using a dilated convolutional neural network, Med. Phys., 47 (2020), 4928–4938. https://doi.org/10.1002/mp.14402 doi: 10.1002/mp.14402
[20]	X. Hong, Y. Zan, F. Weng, W. Tao, Q. Peng, Q. Huang, Enhancing the image quality via transferred deep residual learning of coarse PET sinograms, IEEE Trans. Med. Imaging, 37 (2018), 2322–2332. https://doi.org/10.1109/TMI.2018.2830381 doi: 10.1109/TMI.2018.2830381
[21]	I. Häggström, C. R. Schmidtlein, G. Campanella, T. J. Fuchs, DeepPET: A deep encoder–decoder network for directly solving the PET image reconstruction inverse problem, Med. Image Anal., 54 (2019), 253–262. https://doi.org/10.1016/j.media.2019.03.013 doi: 10.1016/j.media.2019.03.013
[22]	A. Sano, T. Nishio, T. Masuda, K. Karasawa, Denoising PET images for proton therapy using a residual U-net, Biomed. Phys. Eng. Express, 7 (2021), 025014. https://doi.org/10.1088/2057-1976/abe33c doi: 10.1088/2057-1976/abe33c
[23]	A. Mehranian, S. D. Wollenweber, M. D. Walker, K. M. Bradley, P. A. Fielding, K. H. Su, et al., Image enhancement of whole-body oncology [¹⁸F]-FDG PET scans using deep neural networks to reduce noise, Eur. J. Nucl. Med. Mol. Imaging, 49 (2022), 539–549. https://doi.org/10.1007/s00259-021-05478-x doi: 10.1007/s00259-021-05478-x
[24]	C. Zheng, Y. Li, J. Li, N. Li, P. Fan, J. Sun, et al., Dynamic convolution neural networks with both global and local attention for image classification, Mathematics, 12 (2024), 1856. https://doi.org/10.3390/math12121856 doi: 10.3390/math12121856
[25]	T. Lei, D. Zhang, X. Du, X. Wang, Y. Wan, A. K. Nandi, Semi-supervised medical image segmentation using adversarial consistency learning and dynamic convolution network, IEEE Trans. Med. Imaging, 42 (2023), 1265–1277. https://doi.org/10.1109/TMI.2022.3225687 doi: 10.1109/TMI.2022.3225687
[26]	I. Soloviev, A. Kovalchuk, V. Klinshov, Dynamic convolution for image matching, Eur. Phys. J. Spec. Top., 2024. https://doi.org/10.1140/epjs/s11734-024-01373-2
[27]	S. Yun, Y. Ro, Strengthening dynamic convolution with attention and residual connection in kernel space, IEEE Access, 12 (2024), 13626–13633. https://doi.org/10.1109/ACCESS.2024.3356064 doi: 10.1109/ACCESS.2024.3356064
[28]	J. Lee, J. Nam, Multi-level and multi-scale feature aggregation using pretrained convolutional neural networks for music auto-tagging, IEEE Signal Process. Lett., 24 (2017), 1208–1212. https://doi.org/10.1109/LSP.2017.2713830 doi: 10.1109/LSP.2017.2713830
[29]	J. Liu, X. Fan, J. Jiang, R. Liu, Z. Luo, Learning a deep multi-scale feature ensemble and an edge-attention guidance for image fusion, IEEE Trans. Circuits Syst. Video Technol., 32 (2021), 105–119. https://doi.org/10.1109/TCSVT.2021.3056725 doi: 10.1109/TCSVT.2021.3056725
[30]	X. Huo, G. Sun, S. Tian, Y. Wang, L. Yu, J. Long, et al., HiFuse: Hierarchical multi-scale feature fusion network for medical image classification, Biomed. Signal Process. Control, 87 (2024), 105534. https://doi.org/10.1016/j.bspc.2023.105534 doi: 10.1016/j.bspc.2023.105534
[31]	C. Tian, Y. Xu, W. Zuo, B. Du, C. W. Lin, D. Zhang, Designing and training of a dual CNN for image denoising, Knowl.-Based Syst., 226 (2021), 106949. https://doi.org/10.1016/j.knosys.2021.106949 doi: 10.1016/j.knosys.2021.106949
[32]	C. Tian, Y. Xu, W. Zuo, Image denoising using deep CNN with batch renormalization, Neural Networks, 121 (2020), 461–473. https://doi.org/10.1016/j.neunet.2019.08.022 doi: 10.1016/j.neunet.2019.08.022
[33]	J. J. Yang, H. Y. Xie, N. N. Xue, A. M. Zhang, Research on underwater image denoising based on dual-channels residual network, Comput. Eng., 49 (2023), 188–198. https://doi.org/10.19678/j.issn.1000-3428.0064662 doi: 10.19678/j.issn.1000-3428.0064662
[34]	H. Qu, H. Xie, Q. Wang, Brain image denoising using dual-channel attentional residual network, Digit. Signal Process., 165 (2025), 105309. https://doi.org/10.1016/j.dsp.2025.105309 doi: 10.1016/j.dsp.2025.105309
[35]	X. Wang, Y. Tang, C. Yao, Y. Gao, Y. Chen, DuINet: A dual-branch network with information exchange and perceptual loss for enhanced image denoising, Digit. Signal Process., 156 (2025), 104835. https://doi.org/10.1016/j.dsp.2024.104835 doi: 10.1016/j.dsp.2024.104835
[36]	H. Qu, H. Xie, Q. Wang, Multi-convolutional neural network brain image denoising study based on feature distillation learning and dense residual attention, Electron. Res. Arch., 33 (2025), 1231–1266. https://doi.org/10.3934/era.2025055 doi: 10.3934/era.2025055
[37]	Z. Cai, L. Xu, J. Zhang, Y. Feng, L. Zhu, F. Liu, ViT-DualAtt: An efficient pornographic image classification method based on Vision Transformer with dual attention, Electron. Res. Arch., 32 (2024), 6698–6716. https://doi.org/10.3934/era.2024313 doi: 10.3934/era.2024313
[38]	B. Chen, W. Wu, Z. Li, T. Han, Z. Chen, W. Zhang, Attention-guided cross-modal multiple feature aggregation network for RGB-D salient object detection, Electron. Res. Arch., 32 (2024), 643–669. https://doi.org/10.3934/era.2024031 doi: 10.3934/era.2024031
[39]	J. Gurrola-Ramos, T. Alarcon, O. Dalmau, J. V. Manjón, MRI Rician noise reduction using recurrent convolutional neural networks, IEEE Access, 12 (2024), 128272–128284. https://doi.org/10.1109/ACCESS.2024.3446791 doi: 10.1109/ACCESS.2024.3446791
[40]	Z. Zhou, M. M. R. Siddiquee, N. Tajbakhsh, J. Liang, UNet++: A nested U-Net architecture for medical image segmentation, Lect. Notes Comput. Sci., 11045 (2018), 3–11. https://doi.org/10.1007/978-3-030-00889-5_1 doi: 10.1007/978-3-030-00889-5_1
[41]	N. Siddique, S. Paheding, C. P. Elkin, V. Devabhaktuni, U-Net and its variants for medical image segmentation: A review of theory and applications, IEEE Access, 9 (2021), 82031–82057. https://doi.org/10.1109/ACCESS.2021.3086020 doi: 10.1109/ACCESS.2021.3086020
[42]	J. Yang, D. S. Marcus, A. Sotiras, DMC-Net: Lightweight dynamic multi-scale and multi-resolution convolution network for pancreas segmentation in CT images, Biomed. Signal Process. Control, 109 (2025), 107896. https://doi.org/10.1016/j.bspc.2025.107896 doi: 10.1016/j.bspc.2025.107896
[43]	C. Tian, M. Zheng, W. Zuo, B. Zhang, Y. Zhang, D. Zhang, Multi-stage image denoising with the wavelet transform, Pattern Recognit., 134 (2023), 109050. https://doi.org/10.1016/j.patcog.2022.109050 doi: 10.1016/j.patcog.2022.109050
[44]	Y. Cheng, W. Zhu, D. Li, L. Wang, Multi-label classification of arrhythmia using dynamic graph convolutional network based on encoder-decoder framework, Biomed. Signal Process. Control, 95 (2024), 106348. https://doi.org/10.1016/j.bspc.2024.106348 doi: 10.1016/j.bspc.2024.106348
[45]	T. Wu, P. Li, J. Sun, B. P. Nguyen, Adaptive edge prior-based deep attention residual network for low-dose CT image denoising, Biomed. Signal Process. Control, 98 (2024), 106773. https://doi.org/10.1016/j.bspc.2024.106773 doi: 10.1016/j.bspc.2024.106773
[46]	D. P. Kingma, J. Ba, Adam: A method for stochastic optimization, preprint, arXiv: 1412.6980. https://doi.org/10.48550/arXiv.1412.6980
[47]	K. A. Johnson, J. A. Becker, The Whole Brain Atlas, Harvard Medical School, 2005. Available from: https://www.med.harvard.edu/aanlib/home.html.
[48]	S. Roth, M. J. Black, Fields of experts: A framework for learning image priors, in 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05), San Diego, CA, USA, 2 (2005), 860–867. https://doi.org/10.1109/CVPR.2005.160
[49]	M. Bevilacqua, A. Roumy, C. Guillemot, M. A. Morel, Low-complexity single-image super-resolution based on nonnegative neighbor embedding, in Proceedings British Machine Vision Conference, (2012), 135. https://doi.org/10.5244/C.26.135
[50]	A. Abdelhamed, S. Lin, M. S. Brown, A high-quality denoising dataset for smartphone cameras, in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, (2018), 1692–1700.
[51]	M. Lebrun, M. Colom, J. M. Morel, The noise clinic: a blind image denoising algorithm, Image Process. OnLine, 5 (2015), 1–54. https://doi.org/10.5201/ipol.2015.125 doi: 10.5201/ipol.2015.125
[52]	Z. Wang, A. C. Bovik, H. R. Sheikh, E. P. Simoncelli, Image quality assessment: from error visibility to structural similarity, IEEE Trans. Image Process., 13 (2004), 600–612. https://doi.org/10.1109/TIP.2003.819861 doi: 10.1109/TIP.2003.819861
[53]	P. Jakub, M. Grzegorz, Based on spectral analysis of voice controllable surveillance system using normalized mean square error, Adv. Mater. Res., 340 (2012), 156–160. https://doi.org/10.4028/www.scientific.net/AMR.340.156 doi: 10.4028/www.scientific.net/AMR.340.156
[54]	T. Li, H. Feng, L. Wang, L. Zhu, Z. Xiong, H. Huang, Stimulating diffusion model for image denoising via adaptive embedding and ensembling, IEEE Trans. Pattern Anal. Mach. Intell., 46 (2024), 8240–8257. https://doi.org/10.1109/TPAMI.2024.3432812 doi: 10.1109/TPAMI.2024.3432812

Reader Comments

Your name:*

Email:*
© 2025 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)