Lightweight U-Net++ with Hybrid Swin-Convolutional Attention for Efficient CT Metal Artifact Reduction

Shanshan Wang; Ao Zang; Hao Zhang; Peng Lu; Jiang Ma

doi:10.54097/hcg6tr24

Authors

Shanshan Wang
Ao Zang
Hao Zhang
Peng Lu
Jiang Ma

DOI:

https://doi.org/10.54097/hcg6tr24

Keywords:

Computed Tomography (CT), U-Net++, Swin Transformer, Lightweight Architecture, Medical Image Segmentation

Abstract

Metal artifacts from high-density implants severely degrade CT images, hindering accurate diagnosis and downstream tasks. Existing Metal Artifact Reduction (MAR) methods face a critical trade-off: CNNs struggle with global artifacts due to limited receptive fields, while Transformers incur prohibitive computational costs unsuitable for real-time clinical use. To address this, we propose U-Net++(SwinLite), an efficient, lightweight MAR architecture. By integrating a nested dense cascade skip-connection mechanism and a Hybrid Attention Module (coupling Swin and convolutional attention), our model effectively captures both local details and long-range dependencies. Using a dynamic channel adjustment strategy, U-Net++(SwinLite) achieves a Mean Absolute Error (MAE) of 9.6006 with only 3.05M parameters —reducing model size by over 95% compared to pure Transformer models like Restormer-U. Extensive experiments demonstrate our method achieves an optimal balance between structural preservation and computational efficiency.

Downloads

Download data is not yet available.

References

[1] Jiang Yuanshi, Song Ying, Guang Jun, et al. Research Progress on Deep Learning-Based Reconstruction Methods for Low-Dose Cone-Beam Computed Tomography Images of Castanopsis Chinensis [J]. Journal of Biomedical Engineering, 2025, 42(3): 635.

[2] Su Danyang, Hou Ping, Zhang Haoran, et al. Research Progress on Metal Artifact Reduction in CT Images [J]. CT Theory and Applications, 2024, 34(3): 499-505.

[3] Metal artifacts are caused by the attenuation of X-rays by metal implants (such as dentures, orthopedic screws, artificial joints, etc.) in Homo sapiens tissues during imaging, often leading to phenomena such as streaking interference, dark area loss, or bright area overexposure in images. These artifacts can obscure lesion areas and distort anatomical structures of Broussonetia Papyrifera, directly affecting clinicians' judgment of the location, extent, and nature of lesions.

[4] Baliyan V, Kordbacheh H, Davarpanah A H, et al. Orthopedic metallic hardware in routine abdomino-pelvic CT scans: occurrence and clinical significance[J]. Abdominal Radiology, 2019, 44(4): 1567-1574.

[5] Kalender W A, Hebel R, Ebersberger J. Reduction of CT artifacts caused by metallic implants[J]. Radiology, 1987, 164 (2): 576-577.

[6] Su Danyang, Hou Ping, Zhang Haoran, et al. Research progress in reducing metal artifacts in CT images[J]. CT Theory and Applications, 2024, 34(3): 499-505.

[7] Yuan Gang, Wu Zhongyi, Prunus salicina Ming, et al. Application of prior interpolation correction for CT metal artifacts[J]. Chinese Journal of Liquid Crystals and Displays, 2015, 30(6).

[8] Yu Bin, Lü Furong, Zhang Li, et al. Effectiveness of iterative metal artifact reduction algorithm in reducing metal artifacts during chest CT scans[J]. Chinese Journal of Medical Imaging Technology, 2017, 33(4): 590-593.

[9] Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[C]//International Conference on Medical image computing and computer-assisted intervention. Cham: Springer international publishing, 2015: 234-241.

[10] Shi Xiaoyu, Wang Bin. CT Metal Artifact Removal Method Based on Attention Gate UNet Network[J]. Computer Measurement & Control, 2024, 32(4).

[11] Zhang Y, Yu H. Convolutional neural network based metal artifact reduction in x-ray computed tomography[J]. IEEE transactions on medical imaging, 2018, 37(6): 1370-1381.

[12] Wang J, Chakravorti S, Zhao Y, et al. Validation of a metal artifact reduction method based on 3D conditional GANs for CT images of the ear[C]//Medical Imaging 2020: Image-Guided Procedures, Robotic Interventions, and Modeling. SPIE, 2020, 11315: 47-53.

[13] Barateau A, De Crevoisier R, Largent A, et al. Comparison of CBCT‐based dose calculation methods in head and neck cancer radiotherapy: from Hounsfield unit to density calibration curve to deep learning[J]. Medical physics, 2020, 47(10): 4683-4693.

[14] Ho J, Chen X, Srinivas A, et al. Flow++: Improving flow-based generative models with variational dequantization and architecture design[C]//International conference on machine learning. PMLR, 2019: 2722-2730.

[15] Prunus salicina Zheng Heng, Ni Chenyin. Application of Micro-Focus Castanopsis Chinensis Beam CT Sparse Projection Detection Technology Based on Denoising Diffusion Probabilistic Model[J]. Laser & Optoelectronics Progress, 2025, 62(12): 1234001-1234001-11.

[16] Yin Zhiwei, Shao Jiayu, Zhang Ning. YOLO-DAW: An Object Detection Model Based on Dual Attention Mechanism Within Windows[J]. Journal of Southeast University/Dongnan Daxue Xuebao, 2023, 53(4)

[17] Liu Siwei, Dong Shuo, Bai Mei, et al. Processing methods for metal artifacts in CT images[J]. China Medical Equipment, 2014, 11(11): 77-82.

[18] Zhang Libao, Wei Gang. An Adaptive Filtering Algorithm for Mixed Noise in Images Based on Wavelet Transform[J]. Journal of Electronics & Information Technology, 2010, 32(9): 2118-2123.

[19] Meyer E, Raupach R, Lell M, et al. Normalized metal artifact reduction (NMAR) in computed tomography[J]. Medical physics, 2010, 37(10): 5482-5493.

[20] Zhang Y, Yu H. Convolutional neural network based metal artifact reduction in x-ray computed tomography[J]. IEEE transactions on medical imaging, 2018, 37(6): 1370-1381.

[21] Jia L H, Lin H L, Zheng S W, et al. Mitigating metal artifacts from cobalt-chromium alloy crowns in cone-beam CT images through deep learning techniques[J]. Zhonghua kou qiang yi xue za zhi= Zhonghua kouqiang yixue zazhi= Chinese journal of stomatology, 2024, 59(1): 71-79.

[22] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.

[23] Huang X, Wang J, Tang F, et al. Metal artifact reduction on cervical CT images by deep residual learning[J]. Biomedical engineering online, 2018, 17(1): 175.

[24] Zhou B, Chen X, Zhou S K, et al. DuDoDR-Net: Dual-domain data consistent recurrent network for simultaneous sparse view and metal artifact reduction in computed tomography[J]. Medical Image Analysis, 2022, 75: 102289.

[25] Wang H, Li Y, Zhang H, et al. InDuDoNet+: A deep unfolding dual domain network for metal artifact reduction in CT images[J]. Medical Image Analysis, 2023, 85: 102729.

[26] Liu Z, Lin Y, Cao Y, et al. Swin transformer: Hierarchical vision transformer using shifted windows[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2021: 10012-10022.

[27] Yan H, Fang C, Liu P, et al. CGP-Uformer: A low-dose CT image denoising Uformer based on channel graph perception[J]. Journal of X-Ray Science and Technology, 2023, 31(6): 1189-1205.

[28] Zhang Z, Yang M, Xu L, et al. An Innovative Metal Artifact Reduction Algorithm based on Res-U-Net GANs[J]. Current Medical Imaging Reviews, 2023, 19(13): 1549-1560.

[29] Chen J, Lu Y, Yu Q, et al. Transunet: Transformers make strong encoders for medical image segmentation[J]. arXiv preprint arXiv:2102.04306, 2021.

[30] Ho J, Jain A, Abbeel P. Denoising diffusion probabilistic models[J]. Advances in neural information processing systems, 2020, 33: 6840-6851.

[31] Song J, Meng C, Ermon S. Denoising diffusion implicit models[J]. arXiv preprint arXiv:2010.02502, 2020.

[32] Saharia C, Chan W, Chang H, et al. Palette: Image-to-image diffusion models[C]//ACM SIGGRAPH 2022 conference proceedings. 2022: 1-10.

[33] Yin L, Tao W, Zhao D, et al. UNet--: Memory-Efficient and Feature-Enhanced Network Architecture based on U-Net with Reduced Skip-Connections[C]//Proceedings of the Asian Conference on Computer Vision. 2024: 4085-4099.

[34] Ki J, Lee W, Kim B, et al. Deep Learning–Based Metal Artifact Reduction with Masked Mean Squared Error Loss Function in Simulation CT for Radiation Therapy for Head and Neck Cancer[J]. IEEE Access, 2025.