Automated Liver and Tumor Segmentation in CT Images Using Improved and Attention-Gated U-Net (AG-U-Net) Architectures

DOI

10.2991/978-94-6239-616-6_72How to use a DOI?

Keywords

Liver segmentation; Tumor segmentation; CT imaging; Deep learning; U-Net; Improved U-Net; Attention-gated U-Net; Dice coefficient; Intersection over Union (IoU); Medical image analysis

Abstract

Accurate segmentation of liver and tumor regions in Computed Tomography (CT) scans is crucial for precise diagnosis, treatment planning, and surgical navigation. Manual segmentation is not only time-consuming but also susceptible to inter-observer variability, emphasizing the necessity for robust automated methods. This study investigates and compares U-Net-based deep learning architectures for medical image segmentation, including the baseline U-Net, an improved U-Net, and an attention-gated U-Net model. The improved U-Net incorporates dropout regularization, batch normalization, and a cosine annealing learning rate schedule to enhance convergence stability and mitigate overfitting. The attention-gated U-Net introduces attention mechanisms to selectively focus on relevant anatomical regions while suppressing irrelevant background information. Experimental evaluation on CT datasets demonstrates substantial improvements in segmentation accuracy, achieving validation Dice coefficients of 0.6279 and 0.7103, and IoU scores of 0.5937 and 0.6011 for the improved and attention-gated models, respectively. The findings underscore the effectiveness of attention mechanisms in refining tumor delineation and enhancing segmentation precision. Future work will explore volumetric CNN-based U-Net extensions with residual block and multi-scale feature fusion strategies to improve generalization and clinical applicability across diverse imaging systems.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - B. Margaretmary
AU  - M. Mary Shanthi Rani
PY  - 2026
DA  - 2026/03/31
TI  - Automated Liver and Tumor Segmentation in CT Images Using Improved and Attention-Gated U-Net (AG-U-Net) Architectures
BT  - Proceedings of the International Conference on Artificial Intelligence and Secure Data Analytics (ICAISDA 2025)
PB  - Atlantis Press
SP  - 978
EP  - 993
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-616-6_72
DO  - 10.2991/978-94-6239-616-6_72
ID  - Margaretmary2026
ER  -