SepU-Net MRI Segmentation Algorithm Using Depthwise Separable Convolution and Pointwise Convolution Integrated U-Net
DOI:
https://doi.org/10.54644/jte.2026.1890Tóm tắt
Accurate segmentation of brain tumors in MRI remains challenging due to the computational demands of conventional deep learning models. We present SepU-Net, a lightweight convolutional neural network that employs Depthwise Separable Convolutions and efficiently channel attention to reduce model complexity by 69.3% compared to standard U-Net (2.39M vs. 7.76M parameters).Evaluated on both BraTS2020 and BraTS2021, SepU-Net achieves high accuracy (0.9938 and 0.994), mean IoU (0.842 and 0.8318), and Dice coefficients (0.846 and 0.8325), with only minor declines on the more heterogeneous BraTS2021 dataset. Notably, SepU-Net delivers a 32.34% improvement in Tumor Core segmentation and a 10.14% gain in Enhancing Tumor segmentation over U-Net, while maintaining strong precision (0.994/0.9942) and sensitivity (0.9921/0.9915) across datasets. SepU-Net requires only 8.3 GFLOPs per inference, 65% fewer than U-Net and operates efficiently on embedded devices with a memory footprint of 1.4GB. These results validate its ability to balance accuracy and efficiency, enabling real-time segmentation in clinical settings. Future work will integrate attention mechanisms and extend the architecture to 3D for enhanced spatial context.
Lượt tải
Tài liệu tham khảo
[1] M. Havaei et al., “Brain tumor segmentation with deep neural networks,” Med. Image Anal., vol. 35, pp. 18–31, 2017,
doi: 10.1016/j.media.2016.05.004.
[2] L. Zhao et al., “MM-UNet: A multimodality brain tumor segmentation network in MRI images,” Front. Oncol., vol. 12, Art. no. 950706, 2022, doi: 10.3389/fonc.2022.950706.
[3] J. K. Ruffle et al., “Brain tumour segmentation with incomplete imaging data,” Brain Commun., vol. 5, no. 2, 2023,
doi: 10.1093/braincomms/fcad118.
[4] P. Li et al., “mResU-Net: Multi-scale residual U-Net-based brain tumor segmentation from multimodal MRI,” Med. Biol. Eng. Comput., 2024, doi: 10.1007/s11517-023-02965-1.
[5] L. Alzubaidi, J. Zhang, and A. J. Humaidi, “Review of deep learning: Concepts, CNN architectures, challenges, applications, future directions,” J. Big Data, vol. 8, Art. no. 53, 2021, doi: 10.1186/s40537-021-00444-8.
[6] B. Hou and S. Guan, “Brain tumor segmentation using deep learning: A review,” J. Comput. Electron. Inf. Manag., vol. 16, 2025,
doi: 10.54097/31ag9n29.
[7] Y. Zhao and L. Lin, “A lightweight U-Net for medical image segmentation,” in Proc. PIERS, 2024, pp. 1–5,
doi: 10.1109/PIERS62282.2024.10618503.
[8] L. Shen et al., “MBDRes-U-Net: Multi-scale lightweight brain tumor segmentation network,” arXiv preprint, arXiv:2411.01896, 2024, doi: 10.48550/arXiv.2411.01896.
[9] G. E. S. Shahid et al., “LIU-NET: Lightweight inception U-Net for efficient brain tumor segmentation,” PeerJ Comput. Sci., 2025,
doi: 10.1109/CW58918.2023.00012.
[10] A. G. Howard et al., “MobileNets: Efficient convolutional neural networks for mobile vision applications,” arXiv preprint, arXiv:1704.04861, 2017.
[11] M. Sandler et al., “MobileNetV2: Inverted residuals and linear bottlenecks,” in Proc. IEEE CVPR, 2018, pp. 4510–4520,
doi: 10.1109/CVPR.2018.00474.
[12] K. Avazov et al., “Dynamic focus on tumor boundaries: A lightweight U-Net for MRI brain tumor segmentation,” Bioengineering, vol. 11, Art. no. 1302, 2024, doi: 10.3390/bioengineering11121302.
[13] D. Liu et al., “SGEResU-Net for brain tumor segmentation,” Math. Biosci. Eng., 2022, doi: 10.3934/mbe.2022073.
[14] Awsaf, “Brain tumor segmentation 2020 dataset,” Kaggle, 2020. [Online]. Available: https://www.kaggle.com/datasets/awsaf49/brats20-dataset-training-validation
[15] S. Zabihi et al., “SepUNet: Depthwise separable convolution integrated U-Net,” in Proc. IEEE ICIP, 2021, pp. 2503–2507,
doi: 10.1109/ICIP42928.2021.9506283.
[16] S. Gore, “Brain tumour segmentation and analysis using BraTS dataset with improvised 2D and 3D U-Net models,” Research Square, 2023, doi: 10.21203/rs.3.rs-2791706/v1.
[17] S. Bakas et al., “Advancing TCGA glioma MRI collections with expert segmentation labels and radiomic features,” Sci. Data, vol. 4, Art. no. 170117, 2017, doi: 10.1038/sdata.2017.117.
[18] S. Bakas et al., “Identifying the best machine learning algorithms for brain tumor segmentation,” arXiv preprint, arXiv:1811.02629, 2018.
[19] A. M. Winkler, “The NIfTI file format,” 2012. [Online]. Available: https://brainder.org/2012/09/23/the-nifti-file-format/
[20] M. Brett et al., “NiBabel: Access a cacophony of neuroimaging file formats,” 2012. [Online]. Available: https://nipy.org/nibabel/nifti_images.html
[21] A. M. Winkler, “Brainder,” 2012. [Online]. Available: https://brainder.org/2012/09/23/
[22] H. Dong et al., “Automatic brain tumor detection using U-Net,” in Med. Image Underst. Anal., Springer, 2017, pp. 506–517,
doi: 10.1007/978-3-319-60964-5_44.
[23] F. Chollet, “Xception: Deep learning with depthwise separable convolutions,” in Proc. IEEE CVPR, 2017, pp. 1251–1258,
doi: 10.1109/CVPR.2017.195.
[24] B. S. Hua, M. K. Tran, and S. K. Yeung, “Pointwise convolutional neural networks,” in Proc. IEEE/GVF, 2017.
[25] D. Haase and M. A. Daniel, “Rethinking depthwise separable convolutions,” arXiv preprint, arXiv:2003.13549, 2020,
doi: 10.48550/arXiv.2003.13549.
[26] J. Hu et al., “Squeeze-and-excitation networks,” in Proc. IEEE CVPR, 2018, pp. 7132–7141, doi: 10.1109/CVPR.2018.00745.
[27] N. Klingler, “Squeeze-and-excitation networks: A performance upgrade,” 2024. [Online]. Available: https://viso.ai/deep-learning/squeeze-and-excite-networks/
[28] S. Woo, J. Park, J. Y. Lee, and I. S. Kweon, “CBAM: Convolutional block attention module,” arXiv preprint, arXiv:1807.06521, 2018.
[29] Q. Wang et al., “ECA-Net: Efficient channel attention for deep convolutional neural networks,” in Proc. IEEE CVPR, 2020, pp. 11534–11542, doi: 10.1109/CVPR42600.2020.01155.
[30] H. Yang et al., “RS-YOLOX: A high-precision detector for object detection in satellite remote sensing images,” Appl. Sci., 2025,
doi: 10.3390/app12178707.
[31] S. R. Dubey, S. K. Singh, and B. B. Chaudhuri, “Activation functions in deep learning: A comprehensive survey and benchmark,” arXiv preprint, arXiv:2109.14545, 2022, doi: 10.48550/arXiv.2109.14545.
[32] N. Srivastava et al., “Dropout: A simple way to prevent neural networks from overfitting,” J. Mach. Learn. Res., vol. 15, pp. 1929–1958, 2014.
[33] K. Matoba, N. Dimitriadis, and F. Fleuret, “The theoretical expressiveness of max pooling,” arXiv preprint, arXiv:2203.01016, 2022, doi: 10.48550/arXiv.2203.01016.
[34] O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional networks for biomedical image segmentation,” in Proc. MICCAI, 2015, pp. 234–241, doi: 10.1007/978-3-319-24574-4_28.
[35] K. Machida, I. Nambu, and Y. Wada, “Transposed convolution as alternative preprocessor for brain–computer interface using EEG,” Appl. Sci., 2023, doi: 10.3390/app13063578.
[36] R. Rastislav and K. Sayed, “3D MRI brain tumor segmentation using U-Net,” Kaggle, 2024. [Online]. Available: https://www.kaggle.com/code/khaledsayedaaaaa/3d-mri-brain-tumor-segmentation-u-net-acc-99
[37] R. Preetha et al., “Brain tumor segmentation using multi-scale attention U-Net with EfficientNetB4 encoder for enhanced MRI analysis,” Sci. Rep., 2025, doi: 10.1038/s41598-025-94267-9.
