JavaScript is disabled for your browser. Some features of this site may not work without it.
| dc.contributor.author | Kalkidan, Mulatu | |
| dc.date.accessioned | 2022-11-18T08:24:41Z | |
| dc.date.available | 2022-11-18T08:24:41Z | |
| dc.date.issued | 2022-08 | |
| dc.identifier.uri | http://ir.bdu.edu.et/handle/123456789/14483 | |
| dc.description.abstract | Glaucoma is an eye condition that occurs as the result of increased fluid pressure inside the eye and injured optic nerve when intraocular pressure rises. If left untreated, this can permanently damage vision in the affected eye(s) and result in blindness. It cannot be cured; however, its progression can be slowed down by treatment. Therefore, early glaucoma detection is crucial. The available scanning methods for glaucoma approaches are pricy and require the assistance of qualified medical professionals and also the manual examination of the eye takes a lot of time, and different doctors' parameter measurements may not always be accurate. Because of this, it's critical to develop a machine learning method that can quickly and easily identify the disease to assist medical professionals or different non-expert users. We created a CNN model to automatically assess fundus images of the eye to determine if a person has glaucoma or not. Many studies have been conducted by utilizing machine learning to diagnose glaucoma, but the methods they employed for preprocessing and feature extraction are still limited the accuracy of the detection. To reduce this limitation an alternative glaucoma detection model is developed based on a machine learning method to analyze and classify the fundus image. The proposed system is implemented using the Python programming language on top of the Tensor flow and Keras API and it is tested with publicly available Large-scale Attention-based Glaucoma (LAG) dataset. Preprocessing methods like Bicubic interpolation, noise reduction with a median filter, and contrast enhancement with contrast limited adaptive histogram equalization (CLAHE) were used to speed up processing and improve the accuracy of the detection model. Additionally, after preprocessing, relevant features from fundus images were extracted using buildup CNN, GLCM and CNN-GLCM to compare the result. The obtained best features were then given to the SVM KNN and CNN classifiers. Our model is evaluated using performance metrics like accuracy, precision, recall, and f1-score. From the experiment, we noticed that the developed end-to-end CNN with adam optimizer achieved a significant classification performance using preprocessed images with training accuracy of 98.40% and test accuracy of 93% with a batch size of 32 and 100 epoch. Keywords: Glaucoma, Machine Learning, Preprocessing, CNN, GLCM, KNN | en_US |
| dc.language.iso | en_US | en_US |
| dc.subject | Faculty of Electrical and Computer Engineering | en_US |
| dc.title | Detecting Glaucoma from Optic Fundus Image Using Machine Learning Approach | en_US |
| dc.type | Thesis | en_US |
