Abstract
We present an automatic method based on transfer learning for the identification of dry age-related macular degeneration (AMD) and diabetic macular edema (DME) from retinal optical coherence tomography (OCT) images. The algorithm aims to improve the classification performance of retinal OCT images and shorten the training time. Firstly, we remove the last several layers from the pre-trained Inception V3 model and regard the remaining part as a fixed feature extractor. Then, the features are used as input of a convolutional neural network (CNN) designed to learn the feature space shifts. The experimental results on two different retinal OCT images datasets demonstrate the effectiveness of the proposed method.
Highlights
The macula, which is located in the central part of the retina, is the most sensitive area of vision.Its healthiness can be affected by a number of pathologies such as age-related macular degeneration (AMD) and diabetic macular edema (DME)
Sugruk et al [24] segmented the optical coherence tomography (OCT) images to find the retinal pigment epithelium (RPE) layer and used a binary classification to classify between AMD and DME
To address the problem of insufficient training data and time-consuming propagation, we propose a method for automatic diagnosis of AMD, DME, and NOR in retinal OCT B-scans which can effectively identify different pathologies
Summary
The macula, which is located in the central part of the retina, is the most sensitive area of vision. Sugruk et al [24] segmented the OCT images to find the retinal pigment epithelium (RPE) layer and used a binary classification to classify between AMD and DME. Kermany et al [34] proposed an image-based deep learning (IBDL) method, which fixed the weights of the network and was used as a feature extractor. Fang et al [36] presented a method utilizing the principal component analysis network (PCANet) to extract the features from each B-scan of the 3-D retinal OCT images and fusing multiple kernels as a composite kernel to exploit the strong correlations among features of the 3-D OCT images for classification. To address the problem of insufficient training data and time-consuming propagation, we propose a method for automatic diagnosis of AMD, DME, and NOR in retinal OCT B-scans which can effectively identify different pathologies. Experimental results and analysis on two SD-OCT are presented in Section 3, and Section 4 outlines conclusions and suggests some future work
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
