Optimized Solution for Multi-Disease Intelligent Analysis of Gastrointestinal Endoscopic Images Based on Transformer Algorithm
Vol. 22 (2026): 2026 3rd International Conference on the Frontiers of Social Sciences, Education, and the Development of Humanities Arts (EDHA 2026)
Received: 2026-06-13
Accepted: 2026-06-13
Published: 2026-06-13
Abstract
Gastrointestinal (GI) endoscopy plays a critical role in the early diagnosis of multiple diseases. However, traditional image analysis techniques often struggle with the complexity, heterogeneity, and subtle visual patterns present in endoscopic imagery. This paper proposes an optimized solution for intelligent multi-disease analysis of GI endoscopic images based on an enhanced Transformer architecture. We design a novel feature encoding pipeline tailored to the hierarchical and local-global nature of GI pathology images, integrating multi-scale token extraction and deformable attention mechanisms to capture both macro- and micro-lesion features. Furthermore, a multi-task decoupling strategy is introduced to handle the co-occurrence of heterogeneous disease labels within a single image. We implement a hybrid loss function and edge-calibrated optimization to enhance the model's discriminative power. Extensive experiments conducted on annotated real-world datasets demonstrate that the proposed method outperforms conventional CNN and baseline Vision Transformer (ViT) models in terms of accuracy, recall, and inference efficiency. Our findings suggest that Transformer-based optimization presents a promising path for real-time, high-precision GI disease screening in clinical applications.
Keywords
References
[1] S. Chaussade, E. Abou Ali, R. Hallit, A. Belle, M. Barret, and R. Coriat, "Air Circulation In Gastrointestinal Light Source Box And Endoscope In The Era Of Sars-Cov-2 And Microorganisms Airborne Transmission," Endoscopy, vol. 53, no. S 01, p. eP506, 2021. doi: 10.1055/s-0041-1724995
[2] S. Modi, K. Kannan, A. Modi, A. S. Nair, and N. Kamath, "Detecting Diseases in Gastrointestinal Biopsy Images Using CNN," Journal of Advanced Zoology, vol. 44, 2023.
[3] M. Afraz, A. M. Fayyaz, and A. Haseeb, "A Unified Paradigm of Classifying GI Tract Diseases in Endoscopy Images Using Multiple Features Fusion," Azerbaijan Journal of High Performance Computing, vol. 6, no. 1, pp. 49-76, 2023.
[4] R. M. Patil, and S. Giripunje, "Deep learning-based detection and classification of gastrointestinal tract diseases in endoscopy images," In 2024 2nd DMIHER International Conference on Artificial Intelligence in Healthcare, Education and Industry (IDICAIEI), November, 2024, pp. 1-6. doi: 10.1109/idicaiei61867.2024.10842695
[5] H. Tong, and H. Zhou, "Study on the classification method of gastrointestinal endoscope image based on CBAM and AlexNet," In Third International Conference on Computer Vision and Pattern Analysis (ICCPA 2023), August, 2023, pp. 133-142.
[6] Y. Okabe, Y. Ishida, K. Kuraoka, T. Ushijima, and O. Tsuruta, "Endoscopic bile duct and/or pancreatic duct cannulation technique for patients with surgically altered gastrointestinal anatomy," Digestive Endoscopy, vol. 26, pp. 122-126, 2014. doi: 10.1111/den.12274
[7] Y. Qi, J. Ding, L. Li, M. Ai, Y. Zhang, X. Chen, and K. Rin, "Application of endoscopic ultrasound image analysis in the treatment of digestive tract diseases and nursing," Journal of Medical Imaging and Health Informatics, vol. 10, no. 9, pp. 2211-2216, 2020. doi: 10.1166/jmihi.2020.3159
[8] S. Wu, R. Zhang, J. Yan, C. Li, Q. Liu, L. Wang, and H. Wang, "High-speed and accurate diagnosis of gastrointestinal disease: learning on endoscopy images using lightweight transformer with local feature attention," Bioengineering, vol. 10, no. 12, p. 1416, 2023. doi: 10.3390/bioengineering10121416
[9] Y. Song, B. Yang, W. Ren, and D. Hu, "Elevation of D-dimer in eosinophilic gastrointestinal diseases in the absence of venous thrombosis: A case series and literature review," Open Medicine, vol. 19, no. 1, p. 20240960, 2024. doi: 10.1515/med-2024-0960
[10] E. Özbay, "Gastrointestinal tract disease classification using residual-inception transformer with wireless capsule endoscopy image segmentation," IEEE Access, vol. 12, pp. 197988-197998, 2024. doi: 10.1109/access.2024.3522009
Copyright and License
Published in2026-06-13 17:20:58
DOI doi.org/10.70088/kxdwmh93
Creative Commons
Copyright: © 2026 by the authors.
Submitted for possible open access
publication under the terms and
conditions of the Creative Commons
Attribution (CC BY) license
(https://creativecommons.org/license
s/by/4.0/).
Copyright
Copyright © The Author(s), 2026. Published by EDHA 2026
Journal Information
- Vol. 22 (2026): 2026 3rd International Conference on the Frontiers of Social Sciences, Education, and the Development of Humanities Arts (EDHA 2026)
- 2026-06-13
- ISSN: (Print) 3078-770X/ (Online) 3078-7718
- Journal Homepage