Enhancing the Performance of Multi-Class Classification Systems for Retinal Diseases Using Deep Learning Models
Bibliographic record
Abstract
Early and accurate detection of retinal diseases is essential to prevent avoidable vision loss.However, manual assessment of fundus images is time-consuming and can vary across clinicians.Deep convolutional neural networks (DCNNs) have improved automated screening, but many models remain computationally demanding and provide limited interpretability.This study proposes a hybrid ensemble framework for multiclass retinal disease classification that balances accuracy, efficiency, and explainability.InceptionV3 and DenseNet121 were used as feature extractors on a public Eye Diseases Classification dataset comprising four categories: normal, cataract, glaucoma, and diabetic retinopathy.The extracted deep features were fused and classified using several ensemble strategies, including hard voting, soft voting, stacking, bagging with Random Forest, and gradient boosting.Performance was evaluated using accuracy, precision, recall, and F1-score, together with training time.DenseNet121 achieved higher accuracy than InceptionV3 while requiring shorter training time.Ensemble learning further improved performance.Bagging with Random Forest reached 99.4% accuracy, and the optimized boosting model achieved 100% accuracy on the held-out test set.Model interpretability was examined using Grad-CAM, which highlighted clinically plausible regions such as the optic disc, macula, and lesion areas.Although the results are promising, external validation on additional datasets is required before clinical deployment.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".