Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{201648,
        author = {Shruthi D and Dr. S. Babu},
        title = {Multimodal Deep Learning Based Screening Of Diabetes and Retinal Complication from Oct, Fundus Imaging and Clinical Operator},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {12},
        pages = {11543-11548},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=201648},
        abstract = {This research introduced a multimodal deep learning framework aimed at automating the screening of diabetes and its retinal complications by integrating Optical Coherence Tomography (OCT) images, fundus photographs, and clinical operator data. The methodology employed distinct convolutional neural network (CNN) architectures to process OCT and fundus images, enabling hierarchical feature extraction to detect intricate retinal abnormalities and surface-level signs of diabetic retinopathy. Image preprocessing techniques including normalization, resizing, and data augmentation were utilized to enhance model generalization and mitigate overfitting risks. Concurrently, clinical features such as blood glucose levels and duration of diabetes underwent standardization before being analyzed through fully connected networks to extract pertinent representations of disease progression. A late-fusion strategy combined image and clinical feature vectors, which were subsequently input into a Softmax-activated classification head for multi-class predictions. The model training utilized cross-entropy loss, optimized with the Adam optimizer, alongside transfer learning leveraging pre-trained networks like ResNet and VGG for improved feature configuration. Performance metrics, including accuracy, sensitivity, specificity, and AUC, indicated that the proposed multimodal approach significantly surpassed single-modality models, especially in the detection of early-stage complications, thus providing a robust decision-support system for diabetes-related retinal disease management.},
        keywords = {multimodal deep learning, convolutional neural networks, OCT imaging, diabetic retinopathy, clinical data fusion, transfer learning, predictive modeling},
        month = {May},
        }