A Comparative Analysis of Glaucoma Feature Extraction and Classification Techniques in Fundus Images

Abstract

Glaucoma is an asymptomatic chronic eye disease that, if not treated in the early stages, can lead to blindness. Therefore, detection in the early stages is essential to preserve the patient’s quality of life. Thus, it is crucial to have a noninvasive method capable of detecting this disease through images in the fundus examination. In the literature, datasets are available with fundus images; however, only a few have glaucoma images and labels. Learning from an imbalanced dataset challenges machine learning, which limits supervised learning algorithms. We compared approaches to extract and classify three public datasets with 2390 images: ACRIMA, REFUGE, and RIM-ONE DL. First, we evaluated extracted features non-structural from HOG, LBP, Zernike, and Gabor filters and features obtained from transfer learning. Then, we classified them with Multilayer Perceptron (MLP), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGB). Finally, each classifier was evaluated individually and in a voting classifier (VOT). We extracted and classified features from transfer learning models in the same process. Also, they were classified using traditional machine learning. Due to class imbalance, we undersampled the majority class normal by applying the following methods: random choice, near miss, and cluster centroid. We also evaluated our model using a cross-dataset approach. Therefore, we efficiently identified glaucoma in different fundus images using network VGG19 and a voting classifier. In addition, balancing classes reduced false negatives and improved model quality. Our approach achieved an average F1-score equals to 94.69%, accuracy rate of 94.77%, precision of 96.10%, recall of 93.45%, and specificity of 96.08%.

Enhancing Medical Image Classification with an Advanced Feature Selection Algorithm: A Novel Approach to Improving the Cuckoo Search Algorithm by Incorporating Caputo Fractional Order

“…Table 5 demonstrates the numerical results of our comparison analysis. The obtained results confirm the efficacy of CFO-CS compared to methods used in both [ 41 , 42 ].…”

“…Furthermore, we compared our approach to [ 41 ], which uses identical benchmark datasets without feature selection techniques. We also performed a comprehensive assessment in comparison to [ 42 ], where similar techniques were applied.…”

“…To further validate the effectiveness of our proposed method, we conducted a comparison with existing methods outlined in the literature, as shown in Table 5 . Interestingly, our method works better than similar ones that use the whale optimization algorithm (WOA) for feature selection [ 42 ], which uses the same dataset but does not use any feature selection techniques [ 41 ]. This comparison serves to further validate the superiority of our method for enhancing performance.…”

“…To further validate the efficacy of CFO-CS, we conducted a comparative analysis with existing approaches documented in the literature, as summarized in Table 5 . Notably, our method exhibits superior performance compared to similar techniques that utilize the whale optimization algorithm (WOA) for feature selection [ 42 ], which employs the same dataset without any feature selection methods [ 41 ]. This comparative evaluation further validates the superiority of our approach in enhancing overall performance, thereby underscoring its potential as a robust and effective strategy for optimizing classification accuracy and reliability.…”

Automated Glaucoma Detection in Retinal Fundus Images Using Machine Learning Models

“…The study by de Assis and Cortez (2023) [20] provided a comparative analysis of glaucoma feature extraction and classification techniques. They explored both structural and non-structural feature extraction methods, combined with machine learning classifiers.…”