Abstract

Automatic segmentation of cardiac magnetic resonance (CMR) images presents a significant challenge due to the dynamic shape and size variations of heart regions during the diastole and systole phases of the cardiac cycle. This study introduces an enhanced shift-window attention-based Swin transformer for precise segmentation of the left ventricle (LV), right ventricle (RV), and myocardium (MYO). The proposed model improves the self-attention mechanism by employing a cyclic shift-window approach, which facilitates repeated extraction of both low-level and high-level features across all local windows within the Swin transformer block. This shifting strategy enhances the model’s capacity to capture interdependencies between windows, thereby improving the understanding of global context and spatial relationships—critical for accurately delineating anatomical boundaries in CMR segmentation. Additionally, the decoder path is modified to incorporate remote dependencies through skip connections, enabling effective retrieval of global contextual information. This architectural enhancement streamlines the segmentation process, improving both accuracy and computational efficiency. The model was trained and evaluated using publicly available datasets, including the automated cardiac diagnosis challenge (ACDC) 2017 dataset, the multi-centre, multi-vendor & multi-disease cardiac image segmentation (M&Ms) dataset, and the multi-disease, multi-view, multi-centre right ventricular segmentation in  CMR imaging (M&Ms-2) dataset. Compared to prior state-of-the-art methods, the proposed model achieved superior segmentation performance across all three datasets, with the dice coefficient demonstrating statistically significant improvements.

Keywords

Cardiac MRI segmentation, Swin transformer, Shift window attention, Left and right ventricle segmentation, Myocardium segmentation, Deep learning in medical imaging.

References

[1] Update AS. Heart disease and stroke statistics-2019 update: a report from the American heart association. Circulation. 2020; 141(2).

[2] White HD, Norris RM, Brown MA, Brandt PW, Whitlock RM, Wild CJ. Left ventricular end-systolic volume as the major determinant of survival after recovery from myocardial infarction. Circulation. 1987; 76(1):44-51.

[3] Tan LK, Liew YM, Lim E, Mclaughlin RA. Convolutional neural network regression for short-axis left ventricle segmentation in cardiac cine MR sequences. Medical image analysis. 2017; 39:78-86.

[4] Petitjean C, Dacher JN. A review of segmentation methods in short axis cardiac MR images. Medical Image Analysis. 2011; 15(2):169-84.

[5] Lee HY, Codella NC, Cham MD, Weinsaft JW, Wang Y. Automatic left ventricle segmentation using iterative thresholding and an active contour model with adaptation on short-axis cardiac MRI. IEEE Transactions on Biomedical Engineering. 2009; 57(4):905-13.

[6] Lu Y, Radau P, Connelly K, Dick A, Wright GA. Segmentation of left ventricle in cardiac cine MRI: an automatic image-driven method. In functional imaging and modeling of the heart: 5th international conference, Nice, France. Proceedings 2009 (pp. 339-47). Springer Berlin Heidelberg.

[7] Nambakhsh CM, Yuan J, Punithakumar K, Goela A, Rajchl M, Peters TM, et al. Left ventricle segmentation in MRI via convex relaxed distribution matching. Medical Image Analysis. 2013; 17(8):1010-24.

[8] Hu H, Pan N, Wang J, Yin T, Ye R. Automatic segmentation of left ventricle from cardiac MRI via deep learning and region constrained dynamic programming. Neurocomputing. 2019; 347:139-48.

[9] Nasr-esfahani M, Mohrekesh M, Akbari M, Soroushmehr SR, Nasr-esfahani E, Karimi N, et al. Left ventricle segmentation in cardiac MR images using fully convolutional network. In 40th annual international conference of the IEEE engineering in medicine and biology society (EMBC) 2018 (pp. 1275-8). IEEE.

[10] Fu Y, Lei Y, Wang T, Curran WJ, Liu T, Yang X. A review of deep learning based methods for medical image multi-organ segmentation. Physica Medica. 2021; 85:107-22.

[11] Shelhamer E, Long J, Darrell T. Fully convolutional networks for semantic segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2016; 39(4):640-51.

[12] Al KY, Amirrajab S, Lorenz C, Weese J, Pluim J, Breeuwer M. Reducing segmentation failures in cardiac MRI via late feature fusion and GAN-based augmentation. Computers in Biology and Medicine. 2023; 161:106973.

[13] Wang Y, Zhang Y, Wen Z, Tian B, Kao E, Liu X, et al. Deep learning based fully automatic segmentation of the left ventricular endocardium and epicardium from cardiac cine MRI. Quantitative Imaging in Medicine and Surgery. 2021; 11(4):1600-12.

[14] Xu S, Lu H, Cheng S, Pei C. Left ventricle segmentation in cardiac MR images via an improved ResUnet. International Journal of Biomedical Imaging. 2022; 2022(1):1-10.

[15] Tong Q, Li C, Si W, Liao X, Tong Y, Yuan Z, et al. RIANet: recurrent interleaved attention network for cardiac MRI segmentation. Computers in Biology and Medicine. 2019; 109:290-302.

[16] Irshad M, Yasmin M, Sharif MI, Rashid M, Sharif MI, Kadry S. A novel light u-net model for left ventricle segmentation using MRI. Mathematics. 2023; 11(14):1-21.

[17] You R, Zhu Q, Wang Z. Boundary attentive spatial multi-scale network for cardiac MRI image segmentation. In international conference on artificial neural networks 2023 (pp. 75-86). Cham: Springer Nature Switzerland.

[18] Chen D, Bhopalwala H, Dewaswala N, Arunachalam SP, Enayati M, Farahani NZ, et al. Deep neural network for cardiac magnetic resonance image segmentation. Journal of Imaging. 2022; 8(5):1-11.

[19] Shaaf ZF, Jamil MM, Ambar R, Alattab AA, Yahya AA, Asiri Y. Automatic left ventricle segmentation from short-axis cardiac MRI images based on fully convolutional neural network. Diagnostics. 2022; 12(2):1-15.

[20] Silva IF, Silva AC, Paiva AC, Gattass M, Cunha AM. A multi-stage automatic method based on a combination of fully convolutional networks for cardiac segmentation in short-axis MRI. Applied Sciences. 2024; 14(16):1-25.

[21] Ribeiro MA, Nunes FL. Left ventricle segmentation combining deep learning and deformable models with anatomical constraints. Journal of Biomedical Informatics. 2023; 142:1-15.

[22] Khan A, Asad M, Benning M, Roney C, Slabaugh G. Convolution and attention-free mamba-based cardiac image segmentation. In winter conference on applications of computer vision (WACV) 2025 (pp. 1893-903). IEEE.

[23] Singh KR, Sharma A, Singh GK. Attention-guided residual W-Net for supervised cardiac magnetic resonance imaging segmentation. Biomedical Signal Processing and Control. 2023; 86:105177.

[24] Singh KR, Sharma A, Singh GK. W-Net: novel deep supervision for deep learning-based cardiac magnetic resonance imaging segmentation. IETE Journal of Research. 2023; 69(12):8960-76.

[25] Cao H, Wang Y, Chen J, Jiang D, Zhang X, Tian Q, et al. Swin-unet: unet-like pure transformer for medical image segmentation. In European conference on computer vision 2022 (pp. 205-18). Cham: Springer Nature Switzerland.

[26] Maurício J, Domingues I, Bernardino J. Comparing vision transformers and convolutional neural networks for image classification: a literature review. Applied Sciences. 2023; 13(9):1-17.

[27] Liu Z, Lin Y, Cao Y, Hu H, Wei Y, Zhang Z, et al. Swin transformer: hierarchical vision transformer using shifted windows. In proceedings of the international conference on computer vision 2021 (pp. 10012-22). IEEE.

[28] Fu Z, Zhang J, Luo R, Sun Y, Deng D, Xia L. TF-Unet: an automatic cardiac MRI image segmentation method. Mathematical Biosciences and Engineering. 2022; 19(5):5207-22.

[29] Wang L, Huang J, Xing X, Yang G. Hybrid swin deformable attention U-net for medical image segmentation. In 19th international symposium on medical information processing and analysis 2023 (pp. 1-5). IEEE.

[30] Yang R, Liu K, Liang Y. A fusion‐attention swin transformer for cardiac MRI image segmentation. IET Image Processing. 2024; 18(1):105-15.

[31] Hatamizadeh A, Nath V, Tang Y, Yang D, Roth HR, Xu D. Swin unetr: swin transformers for semantic segmentation of brain tumors in MRI images. In international MICCAI brainlesion workshop 2021 (pp. 272-84). Cham: Springer International Publishing.

[32] Grzeszczyk MK, Płotka S, Sitek A. Multi-task Swin transformer for motion artifacts classification and cardiac magnetic resonance image segmentation. In international workshop on statistical atlases and computational models of the heart 2022 (pp. 409-17). Cham: Springer Nature Switzerland.

[33] Bernard O, Lalande A, Zotti C, Cervenansky F, Yang X, Heng PA, et al. Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: is the problem solved? IEEE Transactions on Medical Imaging. 2018; 37(11):2514-25.

[34] Campello VM, Gkontra P, Izquierdo C, Martin-isla C, Sojoudi A, Full PM, et al. Multi-centre, multi-vendor and multi-disease cardiac segmentation: the M&Ms challenge. IEEE Transactions on Medical Imaging. 2021; 40(12):3543-54.

[35] Martín-isla C, Campello VM, Izquierdo C, Kushibar K, Sendra-balcells C, Gkontra P, et al. Deep learning segmentation of the right ventricle in cardiac MRI: the M&Ms challenge. IEEE Journal of Biomedical and Health Informatics. 2023; 27(7):3302-13.

[36] Li D, Peng Y, Guo Y, Sun J. MFAUNet: multiscale feature attentive U‐Net for cardiac MRI structural segmentation. IET Image Processing. 2022; 16(4):1227-42.

[37] Full PM, Isensee F, Jäger PF, Maier-hein K. Studying robustness of semantic segmentation under domain shift in cardiac MRI. In statistical atlases and computational models of the heart. M&Ms and EMIDEC challenges. 11th international workshop, STACOM, held in conjunction with MICCAI, Lima, Peru, 2020 (pp. 238-49). Springer International Publishing.

[38] Fulton MJ, Heckman CR, Rentschler ME. Deformable bayesian convolutional networks for disease-robust cardiac MRI segmentation. In international workshop on statistical atlases and computational models of the heart 2021 (pp. 296-305). Cham: Springer International Publishing.

Cite this article