International Clinical Neuroscience Journal

Vol. 12 No. Continuous (2025)

Lightweight Vision Transformer Architecture for Brain Tumor Segmentation

International Clinical Neuroscience Journal, Vol. 12 No. Continuous (2025), 25 Bahman 2026 , Page e4
https://doi.org/10.22037/icnj.v12i1.51162

Abstract

Background: Accurate and timely segmentation of brain tumors in MRI images is essential for optimal treatment planning. While convolutional neural networks (CNNs) have achieved extensive success in medical image segmentation, they have limited ability to capture long-range spatial dependencies and often require high computational resources to achieve reasonable accuracy. Vision Transformers (ViTs), which utilize global self-attention, offer a promising alternative but are computationally expensive for high-resolution 3D medical images. In this study, we propose SegViTBT, a lightweight hybrid architecture combining a vision transformer encoder with a convolutional decoder for efficient brain tumor segmentation. The model integrates sparse attention to reduce computational load and learnable 2D positional embeddings to enhance spatial representation, delivering high accuracy with reduced resource demands.

Methods: The model is trained on MRI images from the BraTS benchmark dataset. Key performance metrics, including dice coefficient, accuracy, and loss, are evaluated over 25 epochs during training and validation. A comparison is made against conventional CNN and ViT models.

Results: The proposed SegViTBT model demonstrates a stable learning curve with rapid convergence. It achieves a dice score of 78.06% on the BraTS dataset, outperforming baseline CNNs and standard ViT implementations while using less than 60% of the computational resources. Visual results confirm the model’s ability to delineate tumor boundaries with high precision, even for irregularly shaped lesions.

Conclusion: SegViTBT successfully closes the performance gap between CNNs and ViTs in medical imaging by introducing a computationally efficient, pixel-accurate architecture. The model is suitable for deployment in low-resource clinical settings, enabling real-time, practical diagnostic support for brain tumor assessment.

Keywords:
  • Brain tumor segmentation; MRI; Deep learning; Medical Image analysis

How to Cite

1.
Taghavi Bayat Z, Kordnoori S, Sabeti M, Moradi E. Lightweight Vision Transformer Architecture for Brain Tumor Segmentation. Int Clin Neurosci J [Internet]. 2026 Feb. 18 [cited 2026 Jul. 5];12(Continuous):e4. Available from: https://journals.sbmu.ac.ir/neuroscience/article/view/51162

References

1. Karayegen G, Aksahin MF. Brain tumor prediction on MR images with semantic segmentation by using deep learning network and 3D imaging of tumor region. Biomed Signal Process Control. 2021;66:102458. doi: 10.1016/j.bspc.2021.102458

2. Li X, Chen H, Qi X, Dou Q, Fu CW, Heng PA. DenseUNet: Hybrid densely connected UNet for liver and tumor segmentation from CT volumes. IEEE Trans Med Imaging. 2018;37(12):2663–74. doi: 10.1109/TMI.2018.2845918

3. Wang W, Chen C, Ding M, Yu H, Zha S, Li J. TransBTS: Multimodal brain tumor segmentation using transformer. Lect Notes Comput Sci. 2021;12901:109–19. doi: 10.1007/978-3-030-87193-2_11

4. Krithika MA, Suganthi K. Review of semantic segmentation of medical images using modified architectures of UNET. Diagnostics (Basel). 2022;12(12):3064. doi: 10.3390/diagnostics12123064

5. Hatamizadeh A, Tang Y, Vishwesh N, Yang D, Myronenko A, Landman B. UNETR: Transformers for 3D medical image segmentation. Proc IEEE WACV. 2022:574–84. doi: 10.1109/WACV51458.2022.00063

6. Hossain S, Chakrabarty A, Gadekallu TR, Alazab M, Piran J. Vision transformers, ensemble model, and transfer learning leveraging explainable AI for brain tumor detection. IEEE J Biomed Health Inform. 2024;28(3):1261–72. doi: 10.1109/JBHI.2023.3324213

7. Krishnan PT, Krishnadoss P, Khandelwal M, Gupta D, Nihaal A, Kumar TS. Enhancing brain tumor detection in MRI with a rotation invariant vision transformer. Front Neuroinform. 2024;18:1414925. doi: 10.3389/fninf.2024.1414925

8. Asiri AA, Shaf A, Ali T, Shakeel U, Irfan M, Mehdar KM, et al. Exploring the power of deep learning: Fine-tuned vision transformer for accurate and efficient brain tumor detection in MRI scans. Diagnostics (Basel). 2023;13(12):2094. doi: 10.3390/diagnostics13122094

9. Tiu E, Talius E, Patel P, Langlotz CP, Ng AY, Rajpurkar P. Expert-level detection of pathologies from unannotated chest X-ray images via self-supervised learning. Nat Biomed Eng. 2022;6:1399–406. doi: 10.1038/s41551-022-00936-9

10. Zhang J, Lv R, Chen W, Du G, Fu Q, Jiang H. A novel residual network based on multidimensional attention and pinwheel convolution for brain tumor classification. Sci Rep. 2025;15(1):31066. doi: 10.1038/s41598-025-31066-2

11. Saha A, Zhang YD, Satapathy SC. Brain tumour segmentation with a multi-pathway ResNet-based UNet. J Grid Comput. 2021;19(4):43. doi: 10.1007/s10723-021-09568-1

12. Fang L, Wang X. Multi-input UNet model based on the integrated block and the aggregation connection for MRI brain tumor segmentation. Biomed Signal Process Control. 2023;79:104027. doi: 10.1016/j.bspc.2022.104027

13. Cao Y, et al. Automatic detection and segmentation of multiple brain metastases on magnetic resonance images using asymmetric UNet architecture. Phys Med Biol. 2021;66(1):015003. doi: 10.1088/1361-6560/abc5c3

14. Lakshmi K, Amaran S, Subbulakshmi G, Padmini S, Joshi GP, Cho W. Explainable artificial intelligence with UNet-based segmentation and Bayesian machine learning for classification of brain tumors using MRI images. Sci Rep. 2025;15(1):690. doi: 10.1038/s41598-025-00690-4

15. Zhang X, Liu Y, Guo S, Song Z. EG-Unet: Edge-guided cascaded networks for automated frontal brain segmentation in MR images. Comput Biol Med. 2023;158:106891. doi: 10.1016/j.compbiomed.2023.106891

16. Aghalari M, Aghagolzadeh A, Ezoji M. Brain tumor image segmentation via asymmetric/symmetric UNet based on two-pathway-residual blocks. Biomed Signal Process Control. 2021;69:102841. doi: 10.1016/j.bspc.2021.102841

17. Hu HX, Mao WJ, Lin ZZ, Hu Q, Zhang Y. Multimodal brain tumor segmentation based on an intelligent UNET-LSTM algorithm in smart hospitals. ACM Trans Internet Technol. 2021;21(3):1–14. doi: 10.1145/3452143

18. Maji D, Sigedar P, Singh M. Attention Res-UNet with guided decoder for semantic segmentation of brain tumors. Biomed Signal Process Control. 2022;71:103077. doi: 10.1016/j.bspc.2021.103077

19. Lan YL, Zou S, Qin B, Zhu X. Potential roles of transformers in brain tumor diagnosis and treatment. Brain-X. 2023;1:e23. doi: 10.1002/brx2.23

20. Zhang W, Chen S, Ma Y, Liu Y, Cao X. ETUNet: Exploring efficient transformer-enhanced UNet for 3D brain tumor segmentation. Comput Biol Med. 2024;171:108005. doi: 10.1016/j.compbiomed.2024.108005

21. Rasool N, Bhat JI, Wani NA, Ahmad N, Alshara M. TransResUNet: Revolutionizing glioma brain tumor segmentation through transformer-enhanced residual UNet. IEEE Access. 2024;12:72105–16. doi: 10.1109/ACCESS.2024.3385123

22. Dosovitskiy A, et al. An image is worth 16×16 words: Transformers for image recognition at scale. arXiv. 2020;arXiv:2010.11929. doi: 10.48550/arXiv.2010.11929

23. Menze BH, Jakab A, Bauer S, Kalpathy-Cramer J, Farahani K, Kirby J, et al. The multimodal brain tumor image segmentation benchmark (BRATS). IEEE Trans Med Imaging. 2014;34(10):1993–2024. doi: 10.1109/TMI.2014.2377694

24. Yang Q, Wang C, Pan K, Xia B, Xie R, Shi J. An improved 3D-UNet-based brain hippocampus segmentation model based on MR images. BMC Med Imaging. 2024;24(1):166. doi: 10.1186/s12880-024-01166-7

25. Chahbar F, Merati M, Mahmoudi S. MPB-UNet: Multi-parallel blocks UNet for MRI automated brain tumor segmentation. Electronics. 2024;14(1):40. doi: 10.3390/electronics14010040

26. Liang J, Yang C, Zeng L. 3D PSwinBTS: An efficient transformer-based UNet using 3D parallel shifted windows for brain tumor segmentation. Digit Signal Process. 2022;131:103784. doi: 10.1016/j.dsp.2022.103784

27. Soh WK, Yuen HY, Rajapakse JC. HUT: Hybrid UNet transformer for brain lesion and tumour segmentation. Heliyon. 2023;9(12):e22412. doi: 10.1016/j.heliyon.2023.e22412

28. Huang Z, Zhao Y, Liu Y, Song G. GCAUNet: A group cross-channel attention residual UNet for slice-based brain tumor segmentation. Biomed Signal Process Control. 2021;70:102958. doi: 10.1016/j.bspc.2021.102958

29. Agrawal P, Katal N, Hooda N. Segmentation and classification of brain tumor using 3D-UNet deep neural networks. Int J Cogn Comput Eng. 2022;3:199–210. doi: 10.1016/j.ijcce.2022.03.004

30. Cinar N, Ozcan A, Kaya M. A hybrid DenseNet121-UNet model for brain tumor segmentation from MR images. Biomed Signal Process Control. 2022;76:103647. doi: 10.1016/j.bspc.2022.103647

31. Tiwary PK, Johri P, Katiyar A, Chhipa MK. Deep learning-based MRI brain tumor segmentation with EfficientNet-enhanced UNet. IEEE Access. 2025;13:54920–37. doi: 10.1109/ACCESS.2025.3454920

32. Mallampati B, Ishaq A, Rustam F, Kuthala V, Alfarhood S, Ashraf I. Brain tumor detection using 3D-UNet segmentation features and hybrid machine learning model. IEEE Access. 2023;11:135020–34. doi: 10.1109/ACCESS.2023.3332894

33. Zhang L, Lan C, Fu L, Mao X, Zhang M. Segmentation of brain tumor MRI image based on improved attention module UNet network. Signal Image Video Process. 2023;17(5):2277–85. doi: 10.1007/s11760-023-02516-5

  • Abstract Viewed: 97 times
  • PDF Downloaded: 51 times

Download Statastics