A Sigmoid-Based Multi-Label U-Net with Weighted Tversky Loss for Cardiac Fat Segmentation in CT Imaging
DOI:
https://doi.org/10.11113/ijic.v16n1.674Keywords:
Cardiac Fat Segmentation, Computed tomography (CT), Multi-Label U-NetAbstract
Cardiovascular diseases remain among the leading causes of mortality worldwide. Cardiac fat, particularly epicardial adipose tissue and mediastinal adipose tissue, has increasingly been recognized as a key indicator in the early detection of cardiovascular diseases. However, this detection remains challenging due to the thin and often indistinct pericardial membrane that separates epicardial and mediastinal in non-contrast computed tomography (CT) images. Accurate segmentation of these fat regions and the pericardial membrane in CT imaging is essential to support clinical risk assessment and timely medical interventions. Traditional manual segmentation techniques have been employed in the past, but they are time-consuming and prone to inter-observer variability. Meanwhile, conventional machine learning algorithms often struggle to capture the complex spatial information inherent in tomographic CT images. To address these challenges, this study proposes an improved U-Net model designed to enhance the accuracy and adaptability of coronary fat segmentation in non-contrast CT scans. The model specifically addresses issues such as class imbalance, ambiguous labeling, and limited generalizability by incorporating a sigmoid activation function, which enables binary predictions for each anatomical structure and allows the model to effectively handle overlapping regions in multi-class segmentation tasks. The proposed model is evaluated on an annotated CT dataset using standard segmentation performance metrics. The results demonstrate that the advanced U-Net architecture developed in this study achieves improved accuracy in cardiac fat segmentation, thereby offering potential to enhance clinical diagnosis and support better patient treatment in medical imaging.
References
Gaidai, O., Cao, Y., Loginov, S. (2023). Global cardiovascular diseases death rate prediction. Current Problems in Cardiology, 48(5), 101622. https://doi.org/10.1016/j.cpcardiol.2023.101622.
Benčević, M., Galić, I., Habijan, M., & Pižurica, A. (2022). Recent progress in epicardial and pericardial adipose tissue segmentation and quantification based on deep learning: a systematic review. Applied Sciences, 12(10), 5217. https://doi.org/10.3390/app12105217.
Kaushal, A., Gupta, A.K., Sehgal, V.K. (2024). A semantic segmentation framework with UNet-pyramid for landslide prediction using remote sensing data. Scientific Reports. 14(1), 30071. https://doi.org/10.1038/s41598-024-79266-6.
Ronneberger, O., Fischer, P., Brox, T. (2015). U-Net: Convolutional Networks for biomedical image segmentation. International Conference on Medical image computing and computer-assisted intervention. Cham: Springer international publishing. 234241. https://doi.org/10.1007/978-3-319-24574-4_28.
Litjens, G., Kooi, T., Bejnordi, B.E., Setio, A.A.A., Ciompi, F., Ghafoorian, M., van der Laak, J.A.W.M., van Ginneken, B., Sánchez, C.I. (2017). A survey on deep learning in medical image analysis. Medical Image Analysis. 42, 60–88. https://doi.org/10.1016/j.media.2017.07.005.
Abraham, N., Khan, N. (2023). A novel Focal Tversky Loss Function with improved attention U-Net for lesion segmentation. In 2019 IEEE 16th international symposium on biomedical imaging (ISBI 2019) IEEE. 683687. https://doi.org/10.32920/22734398.v1.
Rebelo, A.F., Ferreira, A.M., Fonseca, J.M. (2022). Automatic epicardial fat segmentation and volume quantification on non-contrast cardiac computed tomography. Computer Methods and Programs in Biomedicine Update. 2, 100079. https://doi.org/10.1016/j.cmpbup.2022.100079.
Rodrigues, É., Cordeiro de Morais, F., Conci, A. (2015). On the automated segmentation of epicardial and mediastinal cardiac adipose tissues using classification algorithms. MEDINFO 2015: eHealth-enabled Health. IOS Press. https://doi.org/10.3233/978-1-61499-564-7-726.
Rodrigues, É.O., Morais, F.F.C., Morais, N.A.O.S., Conci, L.S., Neto, L.V., Conci, A. (2016). A novel approach for the automated segmentation and volume quantification of cardiac fats on computed tomography. Computer Methods and Programs in Biomedicine, 123, 109128. https://doi.org/10.1016/j.cmpb.2015.09.017.
Liu, L., Ma, R., van Ooijen, P.M.A., Oudkerk, M., Vliegenthart, R., Veldhuis, R.N.J., Brune, C. (2023). The U-Net family for epicardial adipose tissue segmentation and quantification in low-dose CT. Technologies, 11(4), 104. https://doi.org/10.3390/technologies11040104.
Su, R., Zhang, D., Liu, J., Cheng, C. (2021). MSU-net: Multi-Scale U-net for 2d medical image segmentation. Frontiers in Genetics, 12, 639930.
Hoori, A., Hu, T., Lee, J., Al-Kindi, S., Rajagopalan, S., Wilson, D.L. (2022). Deep learning segmentation and quantification method for assessing epicardial adipose tissue in CT calcium score scans. Scientific Reports, 12(1), 2276. https://doi.org/10.1038/s41598-022-06351-z.
Firouznia, M., Ylipää, E., Henningsson, M., Carlhäll, C.-J. (2025). Poincare guided geometric UNet for left atrial epicardial adipose tissue segmentation in Dixon MRI images. Scientific Reports, 15(1), 25549. https://doi.org/10.1038/s41598-025-10110-1.
Liu, L., Wu, K., Wang, K., Han, Z., Qiu, J., Zhan, Q., Wu, T., Xu, J., Zeng, Z. (2024). SEU 2-Net: multi-scale U 2 -Net with SE attention mechanism for liver occupying lesion CT image segmentation. PeerJ Computer Science, 10, e1751. https://doi.org/10.7717/peerj-cs.1751.
Kazemi, A., Keshtkar, A., Rashidi, S., Aslanabadi, N., Khodadad, B., Esmaeili, M. (2019). Segmentation of cardiac epicardial and pericardial fats by using Gabor Filter Bank Based GLCM. In: 2019 26th National and 4th International Iranian Conference on Biomedical Engineering (ICBME). IEEE. 177–182. https://doi.org/10.1109/ICBME49163.2019.9030380.
Santos da Silva, G., Casanova, D., Oliva, J.T., Rodrigues, E.O. (2024). Cardiac fat segmentation using computed tomography and an image-to-image conditional generative adversarial neural network. Medical Engineering & Physics, 124(1), 104104. https://doi.org/10.1016/j.medengphy.2024.104104.
Zheng, Y., Tian, B., Yu, S., Yang, X., Yu, Q., Zhou, J., Jiang, G., Zheng, Q., Pu, J., Wang, L. (2025). Adaptive boundary-enhanced Dice loss for image segmentation. Biomedical Signal Processing and Control, 106, 107741. https://doi.org/10.1016/j.bspc.2025.107741.
Taha, A.A., Hanbury, A. (2015). Metrics for evaluating 3D medical image segmentation: analysis, selection, and tool. BMC Medical Imaging, 15(1), 29. https://doi.org/10.1186/s12880-015-0068-x.
Supriyadi, M.R., Samah, A.B.A., Muliadi, J., Awang, R.A.R., Ismail, N.H., Majid, H.A., Othman, M.S. Bin, Hashim, S.Z.B.M. (2025). A systematic literature review: exploring the challenges of ensemble model for medical imaging. BMC Med Imaging, 25, 128. https://doi.org/10.1186/s12880-025-01667-4.
Salehi, A.W., Khan, S., Gupta, G., Alabduallah, B.I., Almjally, A., Alsolai, H., Siddiqui, T., Mellit, A. (2023). A study of CNN and transfer learning in medical imaging: advantages, challenges, future scope. Sustainability, 15, 5930. https://doi.org/10.3390/su15075930.
Goodfellow, I.J., Shlens, J., Szegedy, C. (2014). Explaining and harnessing adversarial examples. 3rd International Conference on Learning Representations.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 International Journal of Innovative Computing
This work is licensed under a Creative Commons Attribution 4.0 International License.
