We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
GD-STFA: Gradient Descent Sea Turtle Foraging Algorithm enabled Deep Q Network for Diabetic Retinopathy Detection.
- Authors
Sabeena, A. S.; Jeyakumar, M. K.
- Abstract
Diabetic Retinopathy (DR) refers to a retinal disease that happens due to high glucose levels in the blood, and it leads to blindness globally. In order to delay or avoid vision loss and degradation, early detection, and treatment are necessary. Hence, the conception of an automated technique for precise DR recognition is necessary. Thus, a Deep Q network (DQN) with a Gradient Descent-Sea Turtle Foraging Algorithm (GD-STFA) is proposed that inherits Gradient Descent (GD) in Sea Turtle Foraging Algorithm (STFA), to detect DR from images. This method involves various phases, like pre-processing, segmentation, as well as DR detection. Here, the DR image quality is enhanced in the pre-processing phase using a median filter. After that, the segmentation is done by utilizing Swin-UNet and it is trained using the proposed GD-STFA algorithm. Finally, the DQN trained with GD-STFA finds if a person has DR or normal from his retinal images. The experimentation is performed in detail and validated on the Indian Diabetic Retinopathy Image Dataset benchmark dataset. Ultimately, the experimental outcomes demonstrate that the model is better than traditional techniques regarding optimal accuracy of 0.98, sensitivity of 0.984 and specificity of 0.971. Also, regarding the segmentation accuracy, the proposed model is around 0.906 accurate with a Dice coefficient of 0.963.
- Subjects
SEA turtles; DIABETIC retinopathy; ALGORITHMS; BLOOD sugar; VISION disorders; RETINAL diseases; RETROLENTAL fibroplasia
- Publication
Multimedia Tools & Applications, 2024, Vol 83, Issue 18, p53817
- ISSN
1380-7501
- Publication type
Article
- DOI
10.1007/s11042-023-17507-y