Table 1 Comparison of existing methodologies.
Reference | Methodology | Dataset | Evaluation Measures | Results |
|---|---|---|---|---|
A practical approach for the diagnosis of DME | Duke Dataset | SVM classifier, Naive Bayes, KNN | 79.25% Accuracy | |
Convolutional Neural Network (CNN) | Labeled (OCT) images Mendeley Data, 2018 | Accuracy, Precision, Recall, and F1-Score | 95.66% Accuracy | |
Deep Neural Network Feature Extraction & Handcrafted Feature Extraction | Mendeley Data, 2018 | Accuracy, Precision, Recall, F1-Score | Testing accuracy of DenseNet: 0.88% & ResNet: 0.89% | |
Deep learning techniques, such as convolutional neural networks (CNN) | Mendeley Data, 2018 | Sensitivity, Specificity, Accuracy, F1-Score, And Confusion Matrix | 96% Accuracy | |
Residual neural network (ResNet) | Mendeley Data, 2018 | Accuracy, Mean Accuracy, Standard deviation. | 98% Accuracy | |
OCT-NET deep learning model | SERI Dataset | Accuracy, sensitivity, specificity | 93% Accuracy | |
Ensemble Deep Neural Networks | AREDS dataset | Accuracy | 86% Accuracy | |
Deep convolutional neural networks (AlexNet) | Labeled (OCT) images Mendeley Data, 2018 | Accuracy, Sensitivity, and Specificity | 93.8% Accuracy | |
Efficient global attention block (GAB) for feed-forward convolutional neural networks (CNN) | Mendeley Data, 2018 | Accuracy, Precision, Recall, F1-Score, and Confusion Matrix | 97% Accuracy | |
Deep Multi-Scale Fusion CNN | (UCSD) dataset | Accuracy | 96% Accuracy | |
Machine Learning (CNN) method | ODIR dataset | Precision, Recall, F1-score, & Accuracy | 99.21% Accuracy | |
Interleaved DenseNet with SENet (IDSNet) | BreakHis dataset | Accuracy | Binary Classification | |
Convolutional Block Attention Module (CBAM) | ImageNet1K, MS COCO, and VOC 2007 | Top-1 Error & Top-5 Error | Object detection | |
Modeling normal OCT images using Gaussian Mixture Model (GMM) | SERI Dataset, and Duke Dataset | Sensitivity, specificity | 80% and 93% Sensitivity100% and 80% Specificity | |
Binary Residual Feature fusion (BARF) | OCT Retinal Structural Changes | Recall, Precision, F1-Score, & Accuracy | 98% Accuracy | |
Deep convolutional neural network | Labeled (OCT) images Mendeley Data, 2018 | Recall, Precision, F1-Score, and Accuracy | 97.85% Accuracy | |
Deep learning & Convolutional neural Network | Kermany Mendeley Data, 2018 | Accuracy, Sensitivity, Specificity | 98.20% Accuracy | |
Label smoothing generative adversarial network (LSGAN) | OCT Bscan images Mendeley Data, 2018 | Precision, SE, SP, & F1 | Highest Score: 87.25% Precision, 87.21% SE, 95.09% SP, 87.11% F1 |
