A Dual-Stage Deep Learning Framework for Obesity Assessment with Ultrasound Image Segmentation and Severity Classification
Received: 26 October 2025 | Revised: 11 November 2025 and 18 November 2025 | Accepted: 19 November 2025 | Online: 27 December 2025
Corresponding author: Rohini Pawar
Abstract
Obesity is a major global health challenge associated with metabolic dysfunction and multi-organ impairment. Early detection of fat accumulation is critical but depended on the resolution and operator dependency of conventional ultrasound techniques. This study proposes a two-stage deep learning framework for automated segmentation and severity classification of obesity using abdominal ultrasound images. The pipeline integrates four encoder–decoder architectures (ENet, SegNet, U-Net, and U-Net++) for precise fat-layer segmentation, followed by convolutional neural networks (LeNet-5, VGG19, ResNet50, and EfficientNetB0) for four-level obesity severity classification. A curated dataset of 550 radiologist-annotated ultrasound images was used for training and evaluation. The U-Net++ model achieved the highest segmentation performance with a Dice coefficient of 0.986 and an Intersection over Union of 0.973, while EfficientNetB0 recorded a superior classification accuracy of approximately 97% with stable validation performance. The results demonstrate that ultrasound-based deep learning pipelines can deliver reliable, non-invasive, and cost-effective solutions for early obesity detection, offering strong potential for integration into routine diagnostic practice.
Keywords:
ultrasound imaging, obesity assessment, deep learning, U-Net , EfficientNetB0, image segmentation, classificationDownloads
References
"Obesity and overweight," WHO, Dec. 08, 2025. https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight.
J. A. M. J. L. Janssen, "The Causal Role of Ectopic Fat Deposition in the Pathogenesis of Metabolic Syndrome," International Journal of Molecular Sciences, vol. 25, no. 24, Dec. 2024. DOI: https://doi.org/10.3390/ijms252413238
H. Kiyoyama et al., "Association of visceral fat obesity with structural change in abdominal organs: fully automated three-dimensional volumetric computed tomography measurement using deep learning," Abdominal Radiology, vol. 50, no. 9, pp. 4395–4402, Sep. 2025. DOI: https://doi.org/10.1007/s00261-025-04834-x
T.-H. Chou et al., "Deep learning for abdominal ultrasound: A computer-aided diagnostic system for the severity of fatty liver," Journal of the Chinese Medical Association, vol. 84, no. 9, Sep. 2021, Art. no. 842. DOI: https://doi.org/10.1097/JCMA.0000000000000585
S. M. Shaaban, M. Nawaz, Y. Said, and M. Barr, "An Efficient Breast Cancer Segmentation System based on Deep Learning Techniques," Engineering, Technology & Applied Science Research, vol. 13, no. 6, pp. 12415–12422, Dec. 2023. DOI: https://doi.org/10.48084/etasr.6518
Y. K. Desai, "Diagnosis of Medical Images Using Convolutional Neural Networks," Journal of Electrical Systems, vol. 20, no. 6s, pp. 2371–2376, May 2024. DOI: https://doi.org/10.52783/jes.3220
P. Dong et al., "An ultrasound image segmentation method for thyroid nodules based on dual-path attention mechanism-enhanced UNet++," BMC Medical Imaging, vol. 24, no. 1, Dec. 2024, Art. no. 341. DOI: https://doi.org/10.1186/s12880-024-01521-z
C. Zhu et al., "Swin-Net: A Swin-Transformer-Based Network Combing with Multi-Scale Features for Segmentation of Breast Tumor Ultrasound Images," Diagnostics, vol. 14, no. 3, Jan. 2024, Art. no. 269. DOI: https://doi.org/10.3390/diagnostics14030269
A. Pati, S. R. Addula, A. Panigrahi, B. Sahu, D. S. K. Nayak, and M. Dash, "Artificial intelligence in improving disease diagnosis: A case study of cardiovascular disease prediction," in Artificial Intelligence in Medicine and Healthcare, A. Kumar, S. Rani, S. Rathee, N. Hemrajani, and M. Dahiya, Eds. CRC Press, 2025. DOI: https://doi.org/10.1201/9781003508595-2
T. Kim, D. H. Lee, E.-K. Park, and S. Choi, "Deep Learning Techniques for Fatty Liver Using Multi-View Ultrasound Images Scanned by Different Scanners: Development and Validation Study," JMIR Medical Informatics, vol. 9, no. 11, Nov. 2021, Art. no. e30066. DOI: https://doi.org/10.2196/30066
F. M. Alshagathrh, M. S. Househ, F. M. Alshagathrh, and M. S. Househ, "Artificial Intelligence for Detecting and Quantifying Fatty Liver in Ultrasound Images: A Systematic Review," Bioengineering, vol. 9, no. 12, Dec. 2022, Art. no. 748. DOI: https://doi.org/10.3390/bioengineering9120748
F. Alshagathrh et al., "Large annotated ultrasound dataset of non-alcoholic fatty liver from Saudi hospitals for analysis and applications," Data in Brief, vol. 58, Feb. 2025, Art. no. 111266. DOI: https://doi.org/10.1016/j.dib.2024.111266
H. M. Danish, Z. Suhail, and F. Farooq, "Deep learning-based automation for segmentation and biometric measurement of the gestational sac in ultrasound images," Frontiers in Pediatrics, vol. 12, Dec. 2024. DOI: https://doi.org/10.3389/fped.2024.1453302
M. S. White, A. Horikawa-Strakovsky, K. P. Mayer, B. W. Noehren, and Y. Wen, "Open-Source AI for Vastus Lateralis and Adipose Tissue Segmentation to Assess Muscle Size and Quality," Ultrasound in Medicine and Biology, vol. 51, no. 12, pp. 2276–2280, Dec. 2025. DOI: https://doi.org/10.1016/j.ultrasmedbio.2025.08.008
M. Hayat, S. Aramvith, S. Bhattacharjee, and N. Ahmad, "Attention GhostUNet++: Enhanced Segmentation of Adipose Tissue and Liver in CT Images." arXiv, Apr. 14, 2025. DOI: https://doi.org/10.1109/EMBC58623.2025.11254588
T.-J. Yen, C.-T. Yang, Y.-J. Lee, C. Chen, and H.-C. Yang, "Fatty liver classification via risk controlled neural networks trained on grouped ultrasound image data," Scientific Reports, vol. 14, no. 1, Mar. 2024, Art. no. 7345. DOI: https://doi.org/10.1038/s41598-024-57386-3
Downloads
How to Cite
License
Copyright (c) 2025 Rohini Pawar, Rohini Jadhav, Rohit Jadhav, Gagandeep Kaur, Rutuja Rajendra Patil
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain the copyright and grant the journal the right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) after its publication in ETASR with an acknowledgement of its initial publication in this journal.
