Rouzbahani, Sina (2022) Automatic classification of macular pathologies using deep learning techniques. Master's Thesis / Essay, Computing Science.
|
Text
mCS_2022_RouzbahaniS.pdf Download (20MB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
toestemming.pdf Restricted to Registered users only Download (127kB) |
Abstract
Diabetes is one of the most common diseases worldwide, affecting 9.3% of people during their lifetime. Diabetic Retinopathy is a complication of diabetes caused by elevated blood sugar levels. This project focuses on learning-supported visual pathology analysis in Diabetic Macular Edema. Deep learning methods are used to build CNN models that classify eight morphological features in a supervised learning setup. Eight different morphological features are studied. This project utilizes deep learning and machine learning algorithms to investigate the relationship between eight morphological features. The trained models are used to classify and detect the presence of individual and combined features against the novel data. The results are further evaluated to establish the accuracy and reliability of the model to diagnose DME in OCT images with the usage of transfer learning. The experiment results illustrate the most substantial relationship between Thickening and Macular Volume. The results indicate that Cyst is the prominent singleton feature that positively reinforces the accuracy of the trained DME detection model. Similarly, the combination of Thickening and Macular Volume is the most prominent two-feature combination that positively reinforces the accuracy of the model. The average sensitivity and specificity for the prediction of DME is approximately 96% for using single features and combinations of features.
| Item Type: | Thesis (Master's Thesis / Essay) |
|---|---|
| Supervisor name: | Frey, S.D. and Kosinka, J. |
| Degree programme: | Computing Science |
| Thesis type: | Master's Thesis / Essay |
| Language: | English |
| Date Deposited: | 15 Nov 2022 13:46 |
| Last Modified: | 15 Nov 2022 13:46 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/28904 |
Actions (login required)
 |
View Item |