Deep learning assisted convolutional auto-encoders framework for glaucoma detection and anterior visual pathway recognition from retinal fundus images
Glaucoma is one of the world’s leading causes of visual disability. The optic nerve fibers can deteriorate over time and cannot be replaced until it achieves later stages. In aging populations, early detection is extremely significant. In this paper, the Deep Learning Assisted Convolutional Auto-Encoders Framework (DL-CAEF) has been proposed to detect glaucoma and Anterior Visual Pathway (AVP) recognition from retinal fundus images. It consists of an encoder with an encoder structure and a conventional Convolutional Neural Network (CNN). The DL-CAEF is designed to minimize both the reconstruction of an image and the classification error by a multi-model learning mechanism. This method proposes a fully automated AVP segmentation system driven by different MRI and deep learning features. This system, which uses deep learning representation, provides a jointly partitioned statistical model for stable and pathological AVP processing. Results demonstrated how the suggested localized model and sparse appearance- method improves current segmentation methods and is as stable as the manual segmentation approach.
Journal of Ambient Intelligence and Humanized Computing
Saravanan, Vijayalakshmi; Samuel, R. Dinesh Jackson; Krishnamoorthy, Sujatha; and Manickam, Adhiyaman, "Deep learning assisted convolutional auto-encoders framework for glaucoma detection and anterior visual pathway recognition from retinal fundus images" (2022). Kean Publications. 817.