Transactions on Machine Intelligence

Transactions on Machine Intelligence

Improving Medical Image Segmentation Using a Hybrid ResUNet-Transformer Architecture for Liver Tumor Detection

Document Type : Original Article

Authors
N. Shakeri 1
O. Rahmani Seryasat 2
1 Department of Biomedical Engineering, South Tehran Branch, Islamic Azad University, Tehran, Iran
2 Assistant Professor, Department of Electrical Engineering, Shams Higher Education Institute, Gorgan, Iran
10.47176/TMI.2025.57
Abstract
In this research, a new method for segmentation of medical images is presented using a combination of ResUNet architecture and transformer layers. The main objective of this study is to improve the accuracy and efficiency of segmentation models in identifying liver tumors from medical images. In this method, the ResUNet50 architecture is used as the encoder for extracting deep features from images, and transformer layers have been added to the model to enhance the model's ability to understand spatial and channel relationships between features. Then, a decoder section with U-Net structure has been designed to reconstruct the predicted maps. To evaluate the proposed method, a dataset of medical images related to liver tumors was used. The experimental results show that the proposed method performs better compared to baseline models according to metrics such as accuracy, Jaccard coefficient (IoU), and Dice coefficient, and has achieved an average accuracy of 92.5%, Jaccard coefficient of 75.4%, and Dice coefficient of 83.1%. These results indicate that the combination of ResUNet and transformer architecture can provide an effective and powerful tool for segmentation of medical images and more accurate identification of liver tumors. In the future, using more diverse data and applying further optimization techniques can improve the efficiency of this model.
Keywords
Medical image segmentation
Liver tumors
ResUNet architecture
Transformer
Accuracy
Jaccard coefficient (IoU)
Dice coefficient
  • Jha, D., Smedsrud, P. H., Riegler, M. A., Johansen, D., De Lange, T., Halvorsen, P., & Johansen, H. D. (2019, December). ResUNet++: An advanced architecture for medical image segmentation. In 2019 IEEE International Symposium on Multimedia (ISM) (pp. 225–226). IEEE. https://doi.org/10.1109/ISM46123.2019.00049
  • Wang, Z., Zou, Y., & Liu, P. X. (2021). Hybrid dilation and attention residual U‑Net for medical image segmentation. Computers in Biology and Medicine, 134, 104469. https://doi.org/10.1016/j.compbiomed.2021.104449
  • Salpea, N., Tzouveli, P., & Kollias, D. (2022, October). Medical image segmentation: A review of modern architectures. In European Conference on Computer Vision (pp. 708–761). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-25082-8_47
  • Huang, L., Miron, A., Hone, K., & Li, Y. (2024, June). Segmenting medical images: From U‑Net to Res‑UNet and nnU‑Net. In 2024 IEEE 37th International Symposium on Computer‑Based Medical Systems (CBMS) (pp. 459–463). IEEE. https://doi.org/10.1109/CBMS61543.2024.00086
  • Pan, H., Gao, B., Bai, W., Li, B., Li, Y., Zhang, M., … Kong, W. (2023). WA‑ResUNet: A focused tail‑class MRI medical image segmentation algorithm. Bioengineering, 10(8), 945. https://doi.org/10.3390/bioengineering10080945
  • Wu, X., Tang, P., & Zhang, B. (2024, September). Research on image segmentation based on ResUNet. In 2024 IEEE 7th International Conference on Information Systems and Computer Aided Education (ICISCAE) (pp. 342–347). IEEE. https://doi.org/10.1109/ICISCAE62304.2024.10761354
  • Baldeon‑Calisto, M., & Lai‑Yuen, S. (2018). ResU‑Net: Residual convolutional neural network for prostate MRI segmentation. In IIE Annual Conference Proceedings (pp. 736–741). Institute of Industrial and Systems Engineers.
  • Elghazy, H. L., & Fakhr, M. W. (2022). Dual‑ and triple‑stream RESUNET/UNET architectures for multi‑modal liver segmentation. IET Image Processing, 17(5), 1232–1245. https://doi.org/10.1049/ipr2.12708
  • Huang, S., Han, X., Fan, J., Chen, J., Du, L., Gao, W., … Yang, J. (2021). Anterior mediastinal lesion segmentation based on two‑stage 3D ResUNet with attention gates and lung segmentation. Frontiers in Oncology, 10, 513837. https://doi.org/10.3389/fonc.2020.618357
  • Weng, W., & Zhu, X. (2021). INet: Convolutional networks for biomedical image segmentation. IEEE Access, 9, 16630–16591. https://doi.org/10.1109/ACCESS.2021.3053408
  • Zhang, X., Zheng, Y., Bai, X., Cai, L., Wang, L., Wu, S., … Huang, J. (2022). Femoral image segmentation based on two‑stage convolutional network using 3D‑DMFNet and 3D‑ResUNet. Computer Methods and Programs in Biomedicine, 226, 107112. https://doi.org/10.1016/j.cmpb.2022.107110
  • Chincholkar, V., Srivastava, S., Pawankar, A., & Chaudhari, S. (2024, January). Deep learning techniques in liver segmentation: Evaluating U‑Net, Attention U‑Net, ResNet50, and ResUNet models. In 2024 15th International Conference on Cloud Computing, Data Science & Engineering (Confluence) (pp. 779–775). IEEE. https://doi.org/10.1109/Confluence60223.2024.10463507
  • Niu, K., Guo, Z., Peng, X., & Pei, S. (2022). P‑ResUNet: Segmentation of brain tissue with purified residual U‑Net. Computers in Biology and Medicine, 151, 106254. https://doi.org/10.1016/j.compbiomed.2022.106294
  • Appati, J. K., Azuponga, N. A., Boante, L. M., & Mensah, J. A. (2024). A multi‑attention ResUNet and modified U‑Net architecture for liver tumor segmentation. Applied Computational Intelligence and Soft Computing, 2024, Article 8263549. https://doi.org/10.1155/2024/8365349
  • Wodzinski, M., & Müller, H. (2023). Automatic aorta segmentation with heavily augmented, high‑resolution 3‑D ResUNet: Contribution to the SEG_A challenge. In MICCAI Challenge on Segmentation of the Aorta (pp. 54–62). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-53241-2_4
  • Sheng, N., Liu, D., Zhang, J., Che, C., & Zhang, J. (2021). Second‑order ResU‑Net for automatic MRI brain tumor segmentation. Mathematical Biosciences and Engineering, 18(5), 4967–4983. https://doi.org/10.3934/mbe.2021251
  • Zang, L., Liang, W., Ke, H., Chen, F., & Shen, C. (2023). Research on liver cancer segmentation method based on PCNN image processing and SE‑ResUNet. Scientific Reports, 13(1), 13779. https://doi.org/10.1038/s41598-023-39240-0
  • Sun, M., Li, X., & Sun, W. (2024). Image generation and lesion segmentation of brain tumors and stroke based on GAN and 3‑D ResU‑Net. IEEE Access. https://doi.org/10.1109/ACCESS.2024.3383880
  • Han, Z., Jian, M., & Wang, G. G. (2022). ConvUNeXt: An efficient convolution neural network for medical image segmentation. Knowledge-Based Systems, 253, 109512. https://doi.org/10.1016/j.knosys.2022.109512
  • Chincholkar, V., Srivastava, S., Pawanarkar, A., & Chaudhari, S. (2024, January). Deep learning techniques in liver segmentation: evaluating U-Net, Attention U-Net, ResNet50, and ResUNet models. In 2024 14th International Conference on Cloud Computing, Data Science & Engineering (Confluence) (pp. 775-779). IEEE.https://doi.org/10.1109/Confluence60223.2024.10463507
Transactions on Machine Intelligence
Volume 8, Issue 1
Winter 2025
Pages 57-68

  • Receive Date 22 January 2025
  • Revise Date 02 March 2025
  • Accept Date 26 March 2025