EUVIMEDOpdag EUVIMED
Open access · CC BY 4.0

FocusNet: A deep convolutional neural network for automated retinal disease classification using OCT images

Md. Alamin, Md. Safayat Hossain, Mohammad Azam Khan

Array · 2026-12 · 32 · 101411

DOI: 10.1016/j.array.2026.101411 ↗

At a glance

Journal article in Array, 2026-12. Written by Md. Alamin and 2 co-authors. 31 references. The full text is freely available (open access, CC BY 4.0).

AbstractBibliografiske oplysningerCiteRelated studies

Abstract (original language)

Retinal diseases are among the leading causes of irreversible vision loss worldwide, making early and accurate diagnosis essential for effective clinical intervention. Optical Coherence Tomography (OCT) has become a cornerstone of retinal imaging by providing high-resolution cross-sectional visualization of retinal microstructures, enabling the detection of subtle pathological changes. However, automated multi-class classification of OCT images remains challenging due to low image contrast, speckle noise, and high inter-class morphological similarity, while the limited interpretability of deep learning models presents an additional challenge for clinical adoption. To address these issues, this study proposes FocusNet, a task-specific deep convolutional neural network (DCNN) for automated retinal disease classification from OCT images. The proposed framework combines an effective preprocessing pipeline with a customized network architecture that learns discriminative retinal representations through hierarchical feature extraction. Furthermore, Gradient-weighted Class Activation Mapping (Grad-CAM) is added to improve the transparency and clinical trustworthiness of the model. Experimental results show that FocusNet achieves an accuracy of 97.50%, precision of 97.58%, sensitivity of 97.50%, specificity of 99.17%, an F1-score of 97.50%, and an AUC of 99.77%, outperforming several representative deep learning models. These findings demonstrate the effectiveness of FocusNet and highlight its potential as a reliable computer-aided diagnostic tool for retinal disease classification from OCT images.

Cite

APA

Alamin, M., Hossain, M. S., & Khan, M. A. (2026). FocusNet: A deep convolutional neural network for automated retinal disease classification using OCT images. Array, 32, 101411. https://doi.org/10.1016/j.array.2026.101411

Vancouver

Alamin M, Hossain MS, Khan MA. FocusNet: A deep convolutional neural network for automated retinal disease classification using OCT images. Array. 2026;32:101411. doi:10.1016/j.array.2026.101411

BibTeX
@article{Alamin2026,
  author = {Alamin, Md. and Hossain, Md. Safayat and Khan, Mohammad Azam},
  title = {FocusNet: A deep convolutional neural network for automated retinal disease classification using OCT images},
  journal = {Array},
  year = {2026},
  volume = {32},
  pages = {101411},
  doi = {10.1016/j.array.2026.101411}
}
Start for free →