IMPROVING SEGMENTATION IN MULTIPHASE CT IMAGES USING TRAINABLE PHASE SUPERIMAGING
Abstract
Joint Automated analysis of multiphase CT scans, acquired at different time points after contrast agent administration, is highly important for accurate pathology detection. However, it faces a fundamental problem of spatial misalignment between phases due to patient breathing and movement, which leads to a significant reduction in the accuracy of automatic segmentation. Existing pre-registration (phase alignment) methods require manual parameter tuning and are not integrated into trainable pipelines, hindering the full automation of the segmentation process. The aim of this study was to develop a differentiable method for aligning multiphase CT images based on trainable linear affine transformations, fully embedded into the segmentation model training pipeline. Unlike traditional approaches, the proposed method implements a differentiable affine registration module (including translation, rotation, and scaling), whose parameters are optimized via gradient descent without manual tuning and are directly integrated into the computational graph of the nnU-Net model. The study compared the proposed method with a baseline multiphase segmentation approach (based on simple phase concatenation without registration) and an affine registration method implemented using the ITK.Elastix library on the open abdominal CT dataset WAW-TACE. Compared to segmentation without registration, a substantial improvement in quality metrics was achieved: an increase in the Dice coefficient by 43.19% and in the ROC-AUC metric by 13.55%. Compared to Elastix, the improvements were 13.43% in Dice, 6.86% in ROC-AUC, and 19.6% in the accuracy of pathology count detection per CT scan. The practical significance of the research lies in the development of a ready-to-use PyTorch module for integration into CT image analysis pipelines. It enables fully automated registration without the need for manual hyperparameter tuning and offers high computational efficiency
References
1. Isensee, F., Jaeger, P.F., Kohl, S. A., Petersen, J., & Maier-Hein, K.H. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation, Nature methods, 2021, 18 (2), pp. 203-211.
2. Palazzo S., Zambetta G., Calbi R. An overview of segmentation techniques for CT and MRI images: Clinical implications and future directions in medical diagnostics, Medical Imaging Process & Technol-ogy, 2024.
3. Kulikov A.A., Kashirina I.L., Savkina E.F. Segmentatsiya ob"emnykh obrazovaniy pecheni na mul'ti-faznykh KT-izobrazheniyakh s ispol'zovaniem freymvorka nnU-Net [Segmentation of volumetric liver lesions on multiphase CT images using the nnU-Net framework], Modelirovanie, optimizatsiya i infor-matsionnye tekhnologii [Modeling, Optimization and Information Technology], 2025, Vol. 13, No. 1 (48).
4. Zang L. et al. Research on liver cancer segmentation method based on PCNN image processing and SE-ResUnet, Scientific Reports, 2023, Vol. 13, No. 1, pp. 12779.
5. Sun Q. et al. UCR-Net: U-shaped context residual network for medical image segmentation, Computers in Biology and Medicine, 2022, Vol. 151, pp. 106203.
6. Liu S. et al. Segmenting nailfold capillaries using an improved U-net network, Microvascular Research, 2020, Vol. 130, pp. 104011.
7. Lv P. et al. An improved residual U-Net with morphological-based loss function for automatic liver segmentation in computed tomography, Math. Biosci. Eng., 2022, Vol. 19, No. 2, pp. 1426-1447.
8. Cheremiskin A.V., Kashirina I.L. Segmentatsiya mul'tifaznykh KT-izobrazheniy s ispol'zovaniem an-samblya modeley RESUNET [Segmentation of multiphase CT images using the RESUNET model en-semble], Vestnik Voronezhskogo gosudarstvennogo universiteta. Seriya: Sistemnyy analiz i infor-matsionnye tekhnologii [Bulletin of the Voronezh State University. Series: Systems Analysis and Information Technology], 2025, No. 3, pp. 140-152.
9. Ermolenko S.V., Kashirina I.L., Starichkova Yu.V. Postobrabotka rezul'tatov segmentatsii meditsinskikh izobrazheniy [Post-processing of medical image segmentation results], Computational Nanotechnology, 2025, Vol. 12, No. 2, pp. 109-118.
10. Islam K.T., Wijewickrema S. & O’Leary S. A deep learning based framework for the registration of three dimensional multi-modal medical images of the head, Sci Rep., 2021, 11, 1860. Available at: https://doi.org/10.1038/s41598-021-81044-7.
11. Uhm K.H. et al. Deep learning for end-to-end kidney cancer diagnosis on multi-phase abdominal com-puted tomography, NPJ precision oncology, 2021, Vol. 5, No. 1, pp. 54.
12. Pérez de Frutos J. et al. Learning deep abdominal CT registration through adaptive loss weighting and synthetic data generation, Plos one, 2023, Vol. 18, No. 2, pp. e0282110.
13. Balakrishnan G. et al. Voxelmorph: a learning framework for deformable medical image registration, IEEE transactions on medical imaging, 2019, Vol. 38, No. 8, pp. 1788-1800.
14. Wu L. et al. Beyond radiologist-level liver lesion detection on multi-phase contrast-enhanced CT images by deep learning, Iscience, 2023, Vol. 26, No. 11.
15. Nayantara P.V. et al. Automatic liver segmentation from multiphase CT using modified SegNet and ASPP module, SN Computer Science, 2024, Vol. 5, No. 4, pp. 377.
16. Bartnik K. et al. WAW-TACE: a hepatocellular carcinoma multiphase CT dataset with segmentations, radiomics features, and clinical data, Radiology: Artificial Intelligence, 2024, Vol. 6, No. 6,
pp. e240296.
17. Wasserthal J. et al. TotalSegmentator: robust segmentation of 104 anatomic structures in CT images, Radiology: Artificial Intelligence, 2023, Vol. 5, No. 5, pp. e230024.
18. Isensee F. et al. nnu-net revisited: A call for rigorous validation in 3d medical image segmentation, In-ternational Conference on Medical Image Computing and Computer-Assisted Intervention. Cham: Springer Nature Switzerland, 2024, pp. 488-498.
19. Klein S. et al. Elastix: a toolbox for intensity-based medical image registration, IEEE transactions on medical imaging, 2009, Vol. 29, No. 1, pp. 196-205.
20. Boussot V. et al. IMPACT: A Generic Semantic Loss for Multimodal Medical Image Registration, arXiv preprint arXiv:2503.24121, 2025.
21. Paszke A. et al. Pytorch: An imperative style, high-performance deep learning library, Advances in neu-ral information processing systems, 2019, Vol. 32.
22. Li Y., Zhang X. Adaptive moment estimation optimization algorithm using projection gradient for deep learning, Fifth International Conference on Applied Mathematics, Modelling, and Intelligent Computing (CAMMIC 2025). SPIE, 2025, Vol. 13644, pp. 501-511.
