TR
EN
Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings
Abstract
Abstract
Background: Automated stratification of pigmented skin lesions remains a formidable challenge in computational pathology due to high inter-class visual similarity and severe class imbalance. While Vision Transformers (ViTs) have established new benchmarks by capturing global context, they typically process image patches as independent tokens, often failing to explicitly model the topological and spatial relationships critical for differentiating complex lesion morphologies.
Methods: We propose a hybrid framework that synergizes self-supervised feature extraction with geometric deep learning to model the latent structural topology of dermoscopic images. The architecture leverages a pre-trained foundation model as a backbone, integrated with low-rank adaptation layers for domain-specific refinement. Unlike conventional grid-based methods, our approach dynamically constructs a k-Nearest Neighbor (k-NN) graph from high-dimensional patch embeddings. This latent graph is then processed through a Graph Attention Network (GAT), enabling the model to explicitly reason over spatial relationships and multi-scale structural dependencies.
Results: Evaluated on the HAM10000 benchmark, Under identical experimental conditions, our method achieves an accuracy of 97.80% and an AUC of 99.65%, outperforming both CNN- and Transformer-based baselines. Notably, this performance is achieved with high computational efficiency, utilizing only 3.19% trainable parameters (10M) and demonstrating rapid convergence within the first epoch.
Conclusion: This study demonstrates that explicit modeling of spatial interactions via GNNs significantly enhances diagnostic precision in dermoscopy without the need for massive retraining. Although our experiments focus on HAM10000, the design of the framework is expected to generalize to broader dermoscopic datasets due to its reliance on SSL features and topology-aware reasoning. The proposed architecture offers a scalable, interpretable, and clinically viable solution for computer-aided diagnosis systems.
Keywords
- Computational Dermatology
- Graph Neural Networks
- Spatial Reasoning
- Skin Lesion Classification
- Deep Learning
- Computer Vision
- Image Processing
Supporting Institution
Yıldız Technical University
Project Number
FKD-2024-6446
Ethical Statement
This work was supported by Research Fund of the Yıldız Technical University. Project Number: FKD-2024-6446
Thanks
This work was supported by Research Fund of the Yıldız Technical University. Project Number: FKD-2024-6446
References
- [1] A. Esteva, B. Kuprel, R.A. Novoa, J. Ko, S.M. Swetter, H.M. Blau, and S. Thrun, Nature, 542, 115–118 (2017).
- [2] P. Tschandl, C. Rosendahl, and H. Kittler, Sci Data, 5, (2018).
- [3] K. He, X. Zhang, S. Ren, and J. Sun, Deep residual learning for image recognition, in: Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, IEEE Computer Society, (2016), pp. 770–778.
- [4] A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, (2021).
- [5] M. Oquab, T. Darcet, T. Moutakanni, H. Vo, M. Szafraniec, V. Khalidov, P. Fernandez, D. Haziza, F. Massa, A. El-Nouby, M. Assran, N. Ballas, W. Galuba, R. Howes, P.-Y. Huang, S.-W. Li, I. Misra, M. Rabbat, V. Sharma, et al., (2024).
- [6] M. Moor, O. Banerjee, Z.S.H. Abad, H.M. Krumholz, J. Leskovec, E.J. Topol, and P. Rajpurkar, Nature, 616, 259–265 (2023).
- [7] J.P. Huix, A.R. Ganeshan, J.F. Haslum, M. Söderberg, C. Matsoukas, and K. Smith, (2023).
- [8] D. Raval and J.N. Undavia, International Journal of Engineering Trends and Technology, 72, 147–156 (2024).
Details
Primary Language
English
Subjects
Image Processing, Computer Vision and Multimedia Computation (Other), Deep Learning, Bioinformatics, Machine Learning (Other), Artificial Intelligence (Other)
Journal Section
Research Article
Authors
Publication Date
September 27, 2026
Submission Date
January 15, 2026
Acceptance Date
July 25, 2026
Published in Issue
Year 2026 Volume: 38 Number: 3
APA
Başçetin, T. S. (2026). Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings. International Journal of Advances in Engineering and Pure Sciences, 38(3), 449-467. https://doi.org/10.7240/jeps.1862463
AMA
1.Başçetin TS. Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings. JEPS. 2026;38(3):449-467. doi:10.7240/jeps.1862463
Chicago
Başçetin, Tolga Saim. 2026. “Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings”. International Journal of Advances in Engineering and Pure Sciences 38 (3): 449-67. https://doi.org/10.7240/jeps.1862463.
EndNote
Başçetin TS (September 1, 2026) Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings. International Journal of Advances in Engineering and Pure Sciences 38 3 449–467.
IEEE
[1]T. S. Başçetin, “Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings”, JEPS, vol. 38, no. 3, pp. 449–467, Sept. 2026, doi: 10.7240/jeps.1862463.
ISNAD
Başçetin, Tolga Saim. “Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings”. International Journal of Advances in Engineering and Pure Sciences 38/3 (September 1, 2026): 449-467. https://doi.org/10.7240/jeps.1862463.
JAMA
1.Başçetin TS. Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings. JEPS. 2026;38:449–467.
MLA
Başçetin, Tolga Saim. “Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings”. International Journal of Advances in Engineering and Pure Sciences, vol. 38, no. 3, Sept. 2026, pp. 449-67, doi:10.7240/jeps.1862463.
Vancouver
1.Tolga Saim Başçetin. Hybrid Graph Attention Framework for Dermoscopic Classification via Topology-Aware Latent Embeddings. JEPS. 2026 Sep. 1;38(3):449-67. doi:10.7240/jeps.1862463