EAD-Net: A Novel Lesion Segmentation Method in Diabetic Retinopathy Using Neural Networks

Abstract

Diabetic retinopathy (DR) is a common chronic fundus disease, which has four different kinds of microvessel structure and microvascular lesions: microaneurysms (MAs), hemorrhages (HEs), hard exudates, and soft exudates. Accurate detection and counting of them are a basic but important work. The manual annotation of these lesions is a labor-intensive task in clinical analysis. To solve the problem, we proposed a novel segmentation method for different lesions in DR. Our method is based on a convolutional neural network and can be divided into encoder module, attention module, and decoder module, so we refer it as EAD-Net. After normalization and augmentation, the fundus images were sent to the EAD-Net for automated feature extraction and pixel-wise label prediction. Given the evaluation metrics based on the matching degree between detected candidates and ground truth lesions, our method achieved sensitivity of 92.77%, specificity of 99.98%, and accuracy of 99.97% on the e_ophtha_EX dataset and comparable AUPR (Area under Precision-Recall curve) scores on IDRiD dataset. Moreover, the results on the local dataset also show that our EAD-Net has better performance than original U-net in most metrics, especially in the sensitivity and F1-score, with nearly ten percent improvement. The proposed EAD-Net is a novel method based on clinical DR diagnosis. It has satisfactory results on the segmentation of four different kinds of lesions. These effective segmentations have important clinical significance in the monitoring and diagnosis of DR.

Emerging Trends and Research Foci in Artificial Intelligence for Retinal Diseases: Bibliometric and Visualization Study

“…It is the leading cause of vision loss and preventable blindness in adults aged 20-74 years in middle- and high-income countries [ 30 ]. Using a combination of digital retinal image analysis and telemedicine assessment to help identify people at risk of cardiovascular disease and cognitive impairment may have benefits beyond sight-threatening diseases prevention [ 11 , 31 ]. Aamir et al [ 32 ] built a hierarchical deep convolutional neural network (CNN) for glaucoma recognition and prevention using an advanced deep-learning technique.…”

A deeply supervised maximum response texton basedSegNetfor simultaneous multi retinal lesion segmentation

“…Biswal et al 28 utilized the concept of capsule network and implemented an MCaps‐Net for the segmentation of exudates. Cheng Wan et al 29 developed EAD‐Net that extracts the pertinent features using a dual attention module. Caixia et al 30 developed a model by combining U‐Net++ with a residual module that enables the network to segment the lesions more accurately by regaining the relevant low‐level features.…”

Development and Application of an Intelligent Diagnosis System for Retinal Vein Occlusion Based on Deep Learning

“…Li et al [ 15 ] proposed an OCT image segmentation algorithm based on a 3D neural network to solve the problem of retinal fluid segmentation. Wan [ 16 ] presented a convolutional neural network named EAD-Net that can achieve pixel-level accuracy for different types of lesions in diabetic retinopathy. Xu et al [ 17 ] proposed two biomarker segmentation schemes integrating the semiautomatic localization technique and the low-rank and sparse decomposition theory to locate the leakage area in laser surgery of chronic central serous chorioretinopathy.…”