学位论文简介

In this work, the authors have conducted research to enhance the diagnosis of eye diseases, specifically glaucoma and eye tumors, using advanced techniques in medical image analysis. We performed two main experiments: the first focused on glaucoma, integrating feature extraction methods and using the cuckoo search algorithm for feature selection, which led to high classification accuracy. The second experiment tackled eye tumors by developing a deep learning-based classification system improved with a new optimizer combining Caputo’s fractional gradient descent and cuckoo search, resulting in notable performance improvements. The research emphasizes optimizing feature extraction, improving classification accuracy, and integrating advanced methods into clinical practice, aiming for more accurate and timely diagnoses of ocular diseases. The following main innovative research results have been achieved:

1) We developed a new optimizer, CS-CFGD, by combining Caputo fractional gradient descent with the cuckoo search algorithm. This integration enhances search space exploration and convergence speed, addressing local minima issues in non-convex problems.

2) We introduced a novel method for classifying ocular tumors using data augmentation and pre-trained models on fundus images.

3) We compared our optimization algorithm with existing approaches through twenty independent training runs for each pre-trained model, using performance metrics and statistical tests.

4) We proposed a new feature selection method, CFO-CS, which combining of Caputo fractional order with the cuckoo search algorithm to improve glaucoma fundus image classification.

5) We enhanced feature extraction for glaucoma fundus images by combining histogram of oriented gradients (HOGs), local binary pattern (LBP), and deep features from MobileNet and Vgg19 for more accurate diagnosis.

主要学术成果

1. Habeb AAAA, Zhu N, Taresh MM, Ahmed Ali Ali T. 2024. Deep ocular tumor classification model using cuckoo search algorithm and Caputo fractional gradient descent. PeerJ Comput. Sci. 10: e1923 http://doi.org/10.7717/peerj-cs.1923

2. Habeb, A.A.A.A.; Taresh, M.M.; Li, J.; Gao, Z.; Zhu, N. Enhancing Medical Image Classification with an Advanced Feature Selection Algorithm: A Novel Approach to Improving the Cuckoo Search Algorithm by Incorporating Caputo Fractional Order. Diagnostics 2024, 14, 1191. https://doi.org/10.3390/diagnostics14111191