Retinal OCT Dataset for Disease-Specific Optical Reflectivity Analysis Using Deep Learning

This dataset contains retinal Optical Coherence Tomography (OCT) images used in the study "Revealing Disease-Specific Optical Reflectivity Patterns in Retinal OCT Images Through Explainable Deep Learning". The dataset comprises retinal OCT images from eight diagnostic categories: Age-related Macular Degeneration (AMD), Choroidal Neovascularization (CNV), Central Serous Retinopathy (CSR), Diabetic Macular Edema (DME), Diabetic Retinopathy (DR), DRUSEN, Macular Hole (MH), and NORMAL. The images were used to develop and evaluate an explainable deep learning framework based on EfficientNetB0 with transfer learning for multiclass retinal disease classification. Beyond classification, the dataset was utilized to investigate disease-specific optical reflectivity characteristics through Grad-CAM-based explainability analysis, depth-resolved intensity profiling, reflectivity heterogeneity assessment, mean reflectivity mapping, and differential reflectivity analysis. All images were resized to 224 × 224 pixels and prepared for deep learning analysis using TensorFlow and EfficientNet preprocessing. The dataset supports research in retinal disease diagnosis, medical image analysis, explainable artificial intelligence (XAI), optical characterization of retinal tissues, and OCT-based deep learning applications.