This paper addresses the automatic detection of microaneurysms in color fundus images, which plays a key role in computer assisted diagnosis of diabetic retinopathy, a serious and frequent eye disease. The algorithm can be divided into four steps. The first step consists in image enhancement, shade correction and image normalization of the green channel. The second step aims at detecting candidates, i.e. all patterns possibly corresponding to MA, which is achieved by diameter closing and an automatic threshold scheme. Then, features are extracted, which are used in the last step to automatically classify candidates into real MA and other objects; the classification relies on kernel density estimation with variable bandwidth. A database of 21 annotated images has been used to train the algorithm. The algorithm was compared to manually obtained gradings of 94 images; sensitivity was 88.5% at an average number of 2.13 false positives per image.

Automatic detection of microaneurysms in color fundus images for screening of diabetic retinopathy.

A survey on medical image analysis in diabetic retinopathy

Diabetic retinopathy (DR), a chronic disease in which the retina is damaged due to small vessel damage caused by diabetes mellitus, is one of the leading causes of vision impairment in diabetic patients. Detection of the earliest clinical sign of the advent of DR is a critical requirement for intervention and effective treatment. Ophthalmologists are trained to identify DR, based on examining specific minute changes in the eye - microaneurysms, retinal haemorrhages, macular edema and changes in the retinal blood vessels. Segmentation of microaneurysms (MA) is a critical requirement for the early diagnosis of DR and has been the primary focus of the research community over the past few years. In this work, a systematic review of existing literature is carried out to examine the diagnostic use of automated MA detection and segmentation for early DR diagnosis. We mainly focus on existing early DR diagnosis techniques to understand their strengths and weaknesses. Though early diagnosis is performed using colour fundus photography, fluorescein angiography or optical coherence tomography angiography images, our study is limited to colour fundus based techniques. The early DR diagnosis methodologies reviewed in this article can be broadly classified into classical image processing, conventional machine learning (ML), and deep learning (DL) based techniques. Though significant progress has been achieved in these three classes of early DR diagnosis, several challenges and gaps still exist, underscoring a considerable scope for the development of fully automated, user-friendly early DR diagnosis and grading systems. We discuss in detail the challenges that need to be addressed in designing such effective, efficient, and robust algorithms for early DR diagnosis systems and also the ample scope for future research in this area.

Automated microaneurysms detection for early diagnosis of diabetic retinopathy: A Comprehensive review

Automatic microaneurysm detection in fundus image based on local cross-section transformation and multi-feature fusion

Automatic feature extraction is one of the most advantageous merits of deep neural network (DNN), meanwhile, it is an important part for intelligent bearing fault diagnosis. However, most of fault diagnosis methods based on DNN usually excavate the complex relations from original time sequence signals which only present the fault information in time domain. Convolutional Neural Network (CNN) has demonstrated powerful feature learning capabilities in bearing fault diagnosis and the deeper the diagnosis model is, the better the recognition performance is, which resulted in some problems. In order to enrich the fault information from different views and enhance the discrimination for features learned from diagnosis network, this paper proposed a bearing fault diagnosis method based on multi-domain information fusion and improved residual dense network. The original signal and its transformed signals composed the multi-channel input, which contained more comprehensive information and will benefit the deep learning. Then it designed a residual dense network and introduced the convolution attention mechanism which can discriminate the importance of features further improve the feature extraction capability and efficiency of diagnosis network. Finally, it achieved the fault classification, analyzed the effects of key parameters and compared with other diagnosis to verify the effectiveness by lots of experimental results.

Bearing Fault Diagnosis Based on Multiple Transformation Domain Fusion and Improved Residual Dense Networks

Microaneurysms (MAs) are the earliest pre-eminent indicators of diabetic retinopathy (DR) and are hard to distinguish for ophthalmologists on standard fundus images. This study proposes a method based on discrete orthonormal Stockwell transform and statistical features for discriminating between normal and diseased retinal images. Feature extraction by the two different approaches is consolidated and a total of 24 features are extracted for classifier models. Training and testing of the proposed method have been accomplished using 1140 retinal colour photographs. A comparative study using eight best-known classifiers is showcased for detection of MAs and the performance of the classifiers is evaluated using retinal images by performing ten-fold cross-validation procedure. Simulation results demonstrate the efficiency and adequacy of the proposed method which mainly characterise the textural features. The proposed method is compared with existing algorithms and the results show that the algorithm detects DR with high veracity. With the high accuracy and positive prediction, the proposed system assures promising results in early diagnosis of DR.

https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/iet-ipr.2018.5672

Automated detection of microaneurysms using Stockwell transform and statistical features

Diabetic retinopathy (DR) is one of the most frequent microvascular complications of diabetes mellitus, which damages micro‐ and macrovascular systems. Hence, early detection and grading are important for its effective treatment. This study presents a comprehensive micro‐macro feature extraction algorithm for the grading of DR using retinal images. The method employed is a mutliresolutional microtechnique, based on dual‐tree quaternion wavelet transform fused with local mesh patterns. Since the pixel level model is unable to capture macrolevel features and is difficult for efficient decision‐making, this process additionally proposes a macrolevel feature extraction technique based on feature gradients. The macrolevel descriptor considers a group of pixels to find feature gradients of macrolevel lesions. Feature extracted using the micro‐macro approaches is summarized, and a comparison study using three machine learning classifiers is considered. Performance of the classifiers is determined by conducting a 10‐fold cross‐validation procedure. Among the classifiers, the highest classification accuracy of 93.2% is exhibited by radial basis function kernel extreme learning machine. Simulation results illustrate the adaptability and competency of the novel micro‐macro approach with high accuracy and sensitivity. The promising result assures the excellence of the proposed method for automated DR grading over other state‐of‐the‐art techniques.

Fusing dual‐tree quaternion wavelet transform and local mesh based features for grading of diabetic retinopathy using extreme learning machine classifier

Diabetic retinopathy is one of the foremost causes of blindness and also a major cause of visual impairment around the world. Nowdays severity of vision problems can be diagnosed by ophthalmologists using retinal fundus images as the screening platform. For an effective treatment of diabetic retinopathy, early detection and grading are essential. This investigation focused on classification and grading of diabetic retinopathy images into different severity classes. We explore a machine learning technique based on speeded up robust features (SURF) and spatial local binary pattern for automated grading of diabetic retinopathy. A combination of SURF and spatial local binary pattern form the hybrid descriptor set. A dimensionality reduction approach is employed to reduce complexity and feature size. A comparative study using three machine learning classifiers is done and are evaluated using optimum performance measures. Among the classifiers, artificial neural network showed highest classification accuracy of 89.89%.

Sheena Susan Andrews

Papers

Automated detection of microaneurysms using Stockwell transform and statistical features

Fusing dual‐tree quaternion wavelet transform and local mesh based features for grading of diabetic retinopathy using extreme learning machine classifier

Automated Grading of Diabetic Retinopathy using Local-Spatial Descriptors