Laurensia, Yunika and Young, Julio Christian and Suryadibrata, Alethea (2020) Early Detection of Diabetic Retinopathy Cases using Pre-trained EfficientNet and XGBoost. International Journal Advance Soft Computing, 12 (3). pp. 101-111. ISSN 2074-8523
|
Text
Early Detection of Diabetic Retinopathy Cases using Pre-trained EfficientNet and XGBoost.pdf Download (2MB) | Preview |
Abstract
Diabetic retinopathy (DR) has been a leading cause of global blindness and its diagnosis through color fundus images requires experienced clinicians which makes this a difficult and timeconsuming task. In this paper, a CNN approach is proposed to diagnose DR from digital color fundus images and create a preliminary system as an early detection of DR. Gaussian filter is applied to the images for filtering and resizing purpose before the images are processed any further. Furthermore, EfficientNet and XGBoost are used to extract the images’ features and classify the images correspondingly. The network is trained using a high-end GPU on the publicly available Asia Pacific Tele-Ophthalmology Society (APTOS) dataset. On the data set of 3662 Gaussian filtered and resized images used, the proposed model achieves an accuracy score of 98% for binary classification.
| Item Type: | Article |
|---|---|
| Keywords: | diabetic retinopathy, neural networks, efficientnet, xgboost |
| Subjects: | 000 Computer Science, Information and General Works > 000 Computer Science, Knowledge and Systems > 003 Systems (Computer Modeling and Simulation) 600 Technology (Applied Sciences) > 610 Medicine and Health > 610 Medicine and Health |
| Divisions: | Faculty of Engineering & Informatics > Informatics |
| Depositing User: | Administrator UMN Library |
| Date Deposited: | 03 Nov 2023 07:45 |
| Last Modified: | 03 Nov 2023 07:45 |
| URI: | https://kc.umn.ac.id/id/eprint/27023 |
Actions (login required)
 |
View Item |
