A Domain-Adaptive Deep Learning Framework for Robust Liver Steatosis Classification Across Ultrasound Imaging Devices

Author

A. Sahaya Mercy , Dr. G. Arockia Sahaya Sheela


Abstract

 Objectives: This study aims to develop a domain-adaptive deep learning framework for robust liver steatosis classification across heterogeneous ultrasound imaging devices. It addresses domain shift caused by variations in scanners, transducers, acquisition protocols, and clinical environments to improve cross-device generalization. Methodology: The proposed framework integrates device-aware image normalization, speckle-robust preprocessing, CNN-based feature extraction, domain alignment, and confidence-guided adaptation. A lightweight convolutional neural network extracts liver-texture features, while domain-adversarial learning reduces scanner-specific dependencies. Statistical feature alignment and entropy-based adaptation facilitate knowledge transfer from labeled source images to unlabeled target images. Performance is evaluated using accuracy, sensitivity, specificity, F1-score, area under the receiver operating characteristic curve (AUC), calibration, and ablation studies. Findings: The experimental evaluation will determine whether the proposed framework improves cross-device classification performance and reduces domain-related degradation compared with a conventional CNN baseline. Numerical results will be reported after experimental validation. Novelty: The novelty lies in combining device-aware normalization, speckle-robust feature extraction, domain-adversarial learning, statistical feature alignment, and confidence-guided adaptation within a unified framework. Unlike conventional single-device approaches, this method emphasizes cross-device generalization using unlabeled target-domain images, supporting more reliable and transferable ultrasound-based liver steatosis classification across diverse clinical settings.


Keywords

Liver steatosis, Ultrasound imaging, Deep learning, Domain adaptation, Cross-device generalization, MobileNetV3.



Full Text:

Download Paper PDF


References


[1]   Sahaya Mercy, A., & DG, A. S. S. (2025). Enhancing Prediction Accuracy Using Machine Learning Techniques in Identifying Fatty Liver Disease. International Multidisciplinary Research Journal Reviews,2(9). http://DOI.10.17148/IMRJR.2025.020905 

[2]      Zhang, H., Liu, J., Su, D., Bai, Z., Wu, Y., Ma, Y., ... & Yang, X. (2025). Diagnostic of fatty liver using radiomics and deep learning models on non-contrast abdominal CT. PloS one, 20(2), e0310938.

 [3]   Sahaya Mercy, A., & DG, A. S. S. (2025). Region- Focused Patch-Based Mixup for Improving Classification in Low- Resource Medical Imaging. Indian Journal of Science and Technology, 18(34),2811-2820. https://doi.org/10.17485/IJST/v18i34.1313

 [4]     Sahaya Mercy A, Dr. G. Arockia Sahaya Sheela (2026). Approaches for Analysing Ultrasound Images Using Image Processing and Machine Learning Techniques, 5(1), 01-10. ISSN: 2583-5343 https://www.ijitra.com/index.php/ijitra/issue/archive 

[5]   Zhang, H., Wu, X., Ni, W., Wu, J., Zhou, S., Jia, L., ... & Li, J. (2025). Application of machine learning and deep learning in metabolic dysfunction-associated Steatotic Liver disease: a systematic review and meta-analysis. Journal of Advanced Research.

[6]   Mercy, A. S., & Sheela, G. A. S. (2025). Speckle-Robust Local Phase and Ternary Texture Encoding (SLaP TEX) based Feature Extraction for Liver Steatosis Classification in Ultrasound Imaging. THE SCIENTIFIC TEMPER, 5206. https://doi.org/10.58414/SCIENTIFICTEMPER.2025.16.12.08 

[7]   Mercy, A. S., & Sheela, G. A. S. (2025). Deep Learning Framework for Accurate Detection of Liver Steatosis. Journal of Engineering Research and Reports, 27(10), 255-265. https://doi.org/10.9734/jerr/2025/v27i101671 

[8]   Mercy, A. S., & Sheela, G. A. S. (2026). Enhancing Diagnostic Precision of Liver Steatosis through Deep Learning Models. International Journal of Current Science (IJCSPUB), Volume 16, Issue 2, ISSN: 2250-1770, https://doi doi.org/101555/ijarp.4328

  [9]    Mercy, A. S., & Sheela, G. A. S. (2026). Integrating Traditional Diagnostic Principles with Multimodal Ai for Clinical Assessment of Liver Steatosis. Journal of Emerging Trends and Novel Research (JETNR), Volume 4, Issue 4, ISSN: 2984-9276, http://www.jetnr.org

 [10]   Mercy, A. S., & Sheela, G. A. S. (2026). Optimized Deep Learning Model for Accurate Detection of Liver Diseases Using Ultrasound Imaging: A Case Study. International Multidisciplinary Research Journal Reviews, Volume 3, Issue 1, http://DOI.10.17148/IMRJR.2026.030108

 [11]   Sahaya Mercy, A., & DG, A. S. S. (2025). Optimized Deep Learning Model for Accurate Detection of Liver Steatosis Using Ultrasound Images. International Journal Advanced Research Publications Volume: 01 Issue: 03, ISSN 2456-9992. https://doi-doi.org/101555/ijarp.4328 

[12]   Ginter-Matuszewska, B., Adamek, A., Majchrzak, M., Rozplochowski, B., Zientarska, A., Kowala-Piaskowska, A., & Lukasiak, P. (2025). The machine learning approach to identify the early stage of liver fibrosis and steatosis. International Journal of Medical Informatics, 197, 105837.

[13]   Sahaya Mercy, A., & DG, A. S. S. (2026). Deep Learning and Image Processing Techniques for Identifying Liver Steatosis. International Journal Advanced Research Publications. Volume: 02 Issue: 03, ISSN 2456-9992. https://doi doi.org/101555/ijarp.7950

 [14]   Sahaya Mercy, A., & DG, A. S. S. (2025). Deep Learning-Based Automated Identification of Fatty Liver Disease from Ultrasound Images: A Case Study. International Journal Research Publication Analysis Volume: 01 Issue: 07 ISSN 2456-9995. https://doi-doi.org/101555/ijrpa.8597

[15] Sahaya Mercy, A., & DG, A. S. S. (2025). Integrating Convolutional and Transformer Models for Liver Steatosis Detection Using Ultrasound Imaging. International Journal of Research Publication and Reviews, Vol (6), Issue (10), October (2025), Page – 3395-3402, ISSN 2582-7421. www.ijrpr.com.

[16] Sahaya Mercy, A., & DG, A. S. S. (2026). A Multimodal Deep Learning Framework for Improving the Clinical Reliability of Ultrasound-Based Liver Steatosis Assessment. Volume: 5, Issue 3, ISSN: 2583-5343 http://doi:10.59461/ijitra.v5i3.238

 

 

 

 

 

Share your valuable work from Social Media Buttons