Advanced Hybrid CNN-ViT Ensemble with Attention and FPN Mechanism for Retinal OCT Disease Classification

DOI

10.2991/978-94-6239-664-7_41How to use a DOI?

Keywords

Retinal OCT; Deep Learning; Convolutional Neural Networks; Vision Transformer; Feature Pyramid Network; Ensemble Learning

Abstract

Retinal diseases can be considered one of the leading causes of vision loss on the global scene, and OCT imaging is crucial in the timely diagnosis of the disease. In this paper, a hybrid model of deep learning, which combines Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), Feature Pyramid Networks (FPN) and cross-modal attention, is proposed to simultaneously exploit local retinal texture and global structural features. An ensemble mechanism is used to combine three fusion strategies into attention-based, concatenation strategy, and weighted fusion to enhance robustness and accuracy. Experiments on a publicly available OCTDL dataset of seven disease classes get 96.3% accuracy and a 95.1% macro F1-score and outperform the traditional CNN and ViT baselines. These results prove the high promise of the hybrid ensemble models towards the effective retinal OCT multi-disease classification.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Abdullah Al Noman
AU  - Eamin Hasan Shanto
AU  - Mahir Faysal
AU  - Jamil Hasan
AU  - Samidul Islam Imran Kayes
AU  - Mohammad Jahangir Alam
PY  - 2026
DA  - 2026/06/08
TI  - Advanced Hybrid CNN-ViT Ensemble with Attention and FPN Mechanism for Retinal OCT Disease Classification
BT  - Proceedings of the International Conference on Intelligent Data Analysis and Applications (IDAA 2025)
PB  - Atlantis Press
SP  - 593
EP  - 607
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-664-7_41
DO  - 10.2991/978-94-6239-664-7_41
ID  - AlNoman2026
ER  -