Deep Feature Vectors Concatenation for Eye Disease Detection Using Fundus Image

Abstract

Fundus image is an image that captures the back of the eye (retina), which plays an important role in the detection of a disease, including diabetic retinopathy (DR). It is the most common complication in diabetics that remains an important cause of visual impairment, especially in the young and economically active age group. In patients with DR, early diagnosis can effectively help prevent the risk of vision loss. DR screening was performed by an ophthalmologist by analysing the lesions on the fundus image. However, the increasing prevalence of DR is not proportional to the availability of ophthalmologists who can read fundus images. It can lead to delayed prevention and management of DR. Therefore, there is a need for an automated diagnostic system as it can help ophthalmologists increase the efficiency of the diagnostic process. This paper provides a deep learning approach with the concatenate model for fundus image classification with three classes: no DR, non-proliferative diabetic retinopathy (NPDR), and proliferative diabetic retinopathy (PDR). The model architecture used is DenseNet121 and Inception-ResNetV2. The feature extraction results from the two models are combined and classified using the multilayer perceptron (MLP) method. The method that we propose gives an improvement compared to a single model with the results of accuracy, and average precision and recall of 91% and 90% for the F1-score, respectively. This experiment demonstrates that our proposed deep-learning approach is effective for the automatic DR classification using fundus photo data.

Classification of Diabetic Retinopathy Images through Deep Learning Models - Color Channel Approach: A Review

“…Two stages are distinguished in clinical DR [34]: (i) the initial phase nonproliferative diabetic retinopathy (NPDR) is a representation of DR, and (ii) the advanced stage of DR is represented by diabetes-related proliferative retinopathy (PDR) [35] [36]. Fundus imaging can identify microaneurysms, haemorrhages, and hard exudates in the retinal vasculature throughout the NPDR stage, even though the patients may be asymptotic.…”

Endoscopic Image Classification Based on Explainable Deep Learning

“…A large number of people die from a lack of early detection, and approximately 70% of patients miss optimal treatment because their lesion is too small to find in time. A number of approaches have reached high detection and classification accuracy by using a deep learning model, but we propose a new approach to classify an endoscopic image by using the Inception-Resnet-v2 [ 24 ], ResNet-50 [ 25 ], MobileNetV2 [ 26 ], ResNet-152 [ 27 ], and VGG16 [ 28 ] models with a Grad–CAM model specific for explainable artificial intelligence. In addition, we use a data augmentation method to increase the medical image efficiency and use a noise reduction method to overcome the overfitting problem while using a small dataset.…”

Diabetic retinopathy classification using hybrid optimized deep-learning network model in fundus images

“…To effectively classify DR data, feature extraction accuracy must be increased while computational time and expense are simultaneously reduced. 18 All these facts are considered in the proposed work. It aims to create a network model using deep learning for classification and segmentation feature extraction and combine it with deep-learning network models for improved classification.…”

“…Yet, feature extraction techniques are challenging to learn. To effectively classify DR data, feature extraction accuracy must be increased while computational time and expense are simultaneously reduced 18 . All these facts are considered in the proposed work.…”