A convolutional neural network for the screening and staging of diabetic retinopathy

Abstract

Diabetic retinopathy (DR) is a serious retinal disease and is considered as a leading cause of blindness in the world. Ophthalmologists use optical coherence tomography (OCT) and fundus photography for the purpose of assessing the retinal thickness, and structure, in addition to detecting edema, hemorrhage, and scars. Deep learning models are mainly used to analyze OCT or fundus images, extract unique features for each stage of DR and therefore classify images and stage the disease. Throughout this paper, a deep Convolutional Neural Network (CNN) with 18 convolutional layers and 3 fully connected layers is proposed to analyze fundus images and automatically distinguish between controls (i.e. no DR), moderate DR (i.e. a combination of mild and moderate Non Proliferative DR (NPDR)) and severe DR (i.e. a group of severe NPDR, and Proliferative DR (PDR)) with a validation accuracy of 88%-89%, a sensitivity of 87%-89%, a specificity of 94%-95%, and a Quadratic Weighted Kappa Score of 0.91-0.92 when both 5-fold, and 10-fold cross validation methods were used respectively. A prior pre-processing stage was deployed where image resizing and a class-specific data augmentation were used. The proposed approach is considerably accurate in objectively diagnosing and grading diabetic retinopathy, which obviates the need for a retina specialist and expands access to retinal care. This technology enables both early diagnosis and objective tracking of disease progression which may help optimize medical therapy to minimize vision loss.

Assessing the Performance of Artificial Intelligence Systems for the Screening of Diabetic Retinopathy: A Systematic Review and Meta-Analysis

“…Convolution neural networks (CNNs) form the base of deep learning (DL), a subset of machine learning (ML) where the algorithms are inspired by the structure of the human brain (60,61). CNNs take in data, train themselves to recognize the patterns in the data, then predict an output.…”

Systematic Comparison of Heatmapping Techniques in Deep Learning in the Context of Diabetic Retinopathy Lesion Detection

“…The quadratic weighted kappa was the reference metric in the Kaggle EyePacs challenge of 2016 20 and since then it has been used frequently in publications evaluating deep learning for scoring DR severity from fundus pictures. 27 – 29 …”

“…The quadratic weighted kappa was the reference metric in the Kaggle EyePacs challenge of 2016 20 and since then it has been used frequently in publications evaluating deep learning for scoring DR severity from fundus pictures. [27][28][29] The InceptionV3 model attained a quadratic weighted kappa of 0.82, and the ResNet50 and VGG16 models both had a quadratic weighted kappa of 0.79 on the test set. These scores are in the range of the top results attained in the KaggleDR competition.…”

“…Heatmaps were generated for the 54 images using the 10 heatmapping techniques in Table 1. 4,6,[27][28][29][30][31][32][33] The test set was used for testing possible overfitting (see Results section). For Grad-CAM the implementation in the keras-vis library was used.…”

Convolutional neural network model based on radiological images to support COVID-19 diagnosis: Evaluating database biases

“…In fact, CNN has already been successfully applied to medical tasks such as the diagnosis of retinopathy [ 28 ], pneumonia [ 29 ], cardiomegaly [ 30 ] as well as several types of cancer [ 31 ]. Due to its ability to extract information from visual features, CNN can be applied to the task of detecting COVID-19 in patients, based on chest CT and/or X-ray images.…”