Automatic detection of red lesions in digital color fundus photographs

Fundus images analysis using deep features for detection of exudates, hemorrhages and microaneurysms

Background

Convolution neural networks have been considered for automatic analysis of fundus images to detect signs of diabetic retinopathy but suffer from low sensitivity.

Methods

This study has proposed an alternate method using probabilistic output from Convolution neural network to automatically and simultaneously detect exudates, hemorrhages and microaneurysms. The method was evaluated using two approaches: patch and image-based analysis of the fundus images on two public databases: DIARETDB1 and e-Ophtha. The novelty of the proposed method is that the images were analyzed using probability maps generated by score values of the softmax layer instead of the use of the binary output.

Results

The sensitivity of the proposed approach was 0.96, 0.84 and 0.85 for detection of exudates, hemorrhages and microaneurysms, respectively when considering patch-based analysis. The results show overall accuracy for DIARETDB1 was 97.3% and 86.6% for e-Ophtha. The error rate for image-based analysis was also significantly reduced when compared with other works.

Conclusion

The proposed method provides the framework for convolution neural network-based analysis of fundus images to identify exudates, hemorrhages, and microaneurysms. It obtained accuracy and sensitivity which were significantly better than the reported studies and makes it suitable for automatic diabetic retinopathy signs detection.

Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices

Artificial Intelligence (AI) has long promised to increase healthcare affordability, quality and accessibility but FDA, until recently, had never authorized an autonomous AI diagnostic system. This pivotal trial of an AI system to detect diabetic retinopathy (DR) in people with diabetes enrolled 900 subjects, with no history of DR at primary care clinics, by comparing to Wisconsin Fundus Photograph Reading Center (FPRC) widefield stereoscopic photography and macular Optical Coherence Tomography (OCT), by FPRC certified photographers, and FPRC grading of Early Treatment Diabetic Retinopathy Study Severity Scale (ETDRS) and Diabetic Macular Edema (DME). More than mild DR (mtmDR) was defined as ETDRS level 35 or higher, and/or DME, in at least one eye. AI system operators underwent a standardized training protocol before study start. Median age was 59 years (range, 22–84 years); among participants, 47.5% of participants were male; 16.1% were Hispanic, 83.3% not Hispanic; 28.6% African American and 63.4% were not; 198 (23.8%) had mtmDR. The AI system exceeded all pre-specified superiority endpoints at sensitivity of 87.2% (95% CI, 81.8–91.2%) (>85%), specificity of 90.7% (95% CI, 88.3–92.7%) (>82.5%), and imageability rate of 96.1% (95% CI, 94.6–97.3%), demonstrating AI’s ability to bring specialty-level diagnostics to primary care settings. Based on these results, FDA authorized the system for use by health care providers to detect more than mild DR and diabetic macular edema, making it, the first FDA authorized autonomous AI diagnostic system in any field of medicine, with the potential to help prevent vision loss in thousands of people with diabetes annually. ClinicalTrials.gov NCT02963441

A Random Forest classifier-based approach in the detection of abnormalities in the retina

Classification of abnormalities from medical images using computer-based approaches is of growing interest in medical imaging. Timely detection of abnormalities due to diabetic retinopathy and age-related macular degeneration is required in order to prevent the prognosis of the disease. Computer-aided systems using machine learning are becoming interesting to ophthalmologists and researchers. We present here one such technique, the Random Forest classifier, to aid medical practitioners in accurate diagnosis of the diseases. A computer-aided diagnosis system is proposed for detecting retina abnormalities, which combines K means-based segmentation of the retina image, after due preprocessing, followed by machine learning techniques, using several low level and statistical features. Abnormalities in the retina that are classified are caused by age-related macular degeneration and diabetic retinopathy. Performance measures used in the analysis are accuracy, sensitivity, specificity, F-measure, and Mathew correlation coefficient. A comparison with another machine learning technique, the Naïve Bayes classifier shows that the classification achieved by Random Forest classifier is 93.58% and it outperforms Naïve Bayes classifier which yields an accuracy of 83.63%. Graphical abstract Random Forest classifier for abnormality detection in retina images.

Retinal Microaneurysms Detection Using Local Convergence Index Features

Retinal microaneurysms (MAs) are the earliest clinical sign of diabetic retinopathy disease. Detection of MAs is crucial for the early diagnosis of diabetic retinopathy and prevention of blindness. In this paper, a novel and reliable method for automatic detection of MAs in retinal images is proposed. In the first stage of the proposed method, several preliminary microaneurysm candidates are extracted using a gradient weighting technique and an iterative thresholding approach. In the next stage, in addition to intensity and shape descriptors, a new set of features based on local convergence index filters is extracted for each candidate. Finally, the collective set of features is fed to a hybrid sampling/boosting classifier to discriminate the MAs from non-MAs candidates. The method is evaluated on images with different resolutions and modalities (color and scanning laser ophthalmoscope) using six publicly available data sets including the retinopathy online challenges (ROC) data set. The proposed method achieves an average sensitivity score of 0.471 on the ROC data set outperforming state-of-the-art approaches in an extensive comparison. The experimental results on the other five data sets demonstrate the effectiveness and robustness of the proposed MAs detection method regardless of different image resolutions and modalities.

Retinal hemorrhage detection by rule-based and machine learning approach

Robust detection of hemorrhages (HMs) in color fundus image is important in an automatic diabetic retinopathy grading system. Detection of the hemorrhages that are close to or connected with retinal blood vessels was found to be challenge. However, most methods didn't put research on it, even some of them mentioned this issue. In this paper, we proposed a novel hemorrhage detection method based on rule-based and machine learning methods. We focused on the improvement of detection of the hemorrhages that are close to or connected with retinal blood vessels, besides detecting the independent hemorrhage regions. A preliminary test for detecting HM presence was conducted on the images from two databases. We achieved sensitivity and specificity of 93.3% and 88% as well as 91.9% and 85.6% on the two datasets.

Clinical Report Guided Retinal Microaneurysm Detection With Multi-Sieving Deep Learning

Timely detection and treatment of microaneurysms is a critical step to prevent the development of vision-threatening eye diseases such as diabetic retinopathy. However, detecting microaneurysms in fundus images is a highly challenging task due to the low image contrast, misleading cues of other red lesions, and the large variation of imaging conditions. Existing methods tend to fail in face of the large intra-class variation and small inter-class variations for microaneurysm detection in fundus images. Recently, hybrid text/image mining computer-aided diagnosis systems have emerged to offer a promise of bridging the semantic gap between images and diagnostic information. In this paper, we focus on developing an interleaved deep mining technique to cope intelligently with the unbalanced microaneurysm detection problem. Specifically, we present a clinical report guided multi-sieving convolutional neural network, which leverages a small amount of supervised information in clinical reports to identify the potential microaneurysm regions via the image-to-text mapping in the feature space. These potential microaneurysm regions are then interleaved with fundus image information for multi-sieving deep mining in a highly unbalanced classification problem. Critically, the clinical reports are employed to bridge the semantic gap between low-level image features and high-level diagnostic information. We build an efficient microaneurysm detection framework based on the hybrid text/image interleaving and validate its performance on challenging clinical data sets acquired from diabetic retinopathy patients. Extensive evaluations are carried out in terms of fundus detection and classification. Experimental results show that our framework achieves 99.7% precision and 87.8% recall, comparing favorably with the state-of-the-art algorithms. Integration of expert domain knowledge and image information demonstrates the feasibility of reducing the difficulty of training classifiers under extremely unbalanced data distributions.

New hierarchical approach for microaneurysms detection with matched filter and machine learning

Microaneurysms are regarded as the first signs of diabetic retinopathy (DR), but the microaneurysms are not clear in the color retinal images, and many researches were studied to detect and locate these lesions. In this paper, a new hierarchical computing-aided diagnosis approach is proposed for the microaneurysms detection by using the multi-scale and multi-orientation sum of matched filter (MMMF) and machine learning, where 37 dimensional features are extracted from each candidate. Furthermore, several classifiers such as the k-nearest neighbor (kNN), local linear discrimination analysis (LLDA) and support vector machine (SVM) are modified to distinguish the true microaneurysms from the false ones, which is a typical unbalanced classification problem. The effectiveness of the proposed method is verified through the training set of a publicly available database, and the experiment results show that the proposed method has better detection performance including the receiver operating characteristic (ROC) curve and the free-response receiver operating characteristic (FROC) curve. Moreover, the proposed method with 37 dimensional features outperforms that with other features and has a sensitivity from 1/8 to 8 with the average of all seven points being 0.286 tested on the same database.

Automated coronary artery tree segmentation in X-ray angiography using improved Hessian based enhancement and statistical region merging

BACKGROUND AND OBJECTIVE

Coronary artery segmentation is a fundamental step for a computer-aided diagnosis system to be developed to assist cardiothoracic radiologists in detecting coronary artery diseases. Manual delineation of the vasculature becomes tedious or even impossible with a large number of images acquired in the daily life clinic. A new computerized image-based segmentation method is presented for automatically extracting coronary arteries from angiography images.

METHODS

A combination of a multiscale-based adaptive Hessian-based enhancement method and a statistical region merging technique provides a simple and effective way to improve the complex vessel structures as well as thin vessel delineation which often missed by other segmentation methods. The methodology was validated on 100 patients who underwent diagnostic coronary angiography. The segmentation performance was assessed via both qualitative and quantitative evaluations.

RESULTS

Quantitative evaluation shows that our method is able to identify coronary artery trees with an accuracy of 93% and outperforms other segmentation methods in terms of two widely used segmentation metrics of mean absolute difference and dice similarity coefficient.

CONCLUSIONS

The comparison to the manual segmentations from three human observers suggests that the presented automated segmentation method is potential to be used in an image-based computerized analysis system for early detection of coronary artery disease.

An observational study to assess if automated diabetic retinopathy image assessment software can replace one or more steps of manual imaging grading and to determine their cost-effectiveness

BACKGROUND

Diabetic retinopathy screening in England involves labour-intensive manual grading of retinal images. Automated retinal image analysis systems (ARIASs) may offer an alternative to manual grading.

OBJECTIVES

To determine the screening performance and cost-effectiveness of ARIASs to replace level 1 human graders or pre-screen with ARIASs in the NHS diabetic eye screening programme (DESP). To examine technical issues associated with implementation.

DESIGN

Observational retrospective measurement comparison study with a real-time evaluation of technical issues and a decision-analytic model to evaluate cost-effectiveness.

SETTING

A NHS DESP.

PARTICIPANTS

Consecutive diabetic patients who attended a routine annual NHS DESP visit.

INTERVENTIONS

Retinal images were manually graded and processed by three ARIASs: iGradingM (version 1.1; originally Medalytix Group Ltd, Manchester, UK, but purchased by Digital Healthcare, Cambridge, UK, at the initiation of the study, purchased in turn by EMIS Health, Leeds, UK, after conclusion of the study), Retmarker (version 0.8.2, Retmarker Ltd, Coimbra, Portugal) and EyeArt (Eyenuk Inc., Woodland Hills, CA, USA). The final manual grade was used as the reference standard. Arbitration on a subset of discrepancies between manual grading and the use of an ARIAS by a reading centre masked to all grading was used to create a reference standard manual grade modified by arbitration.

MAIN OUTCOME MEASURES

Screening performance (sensitivity, specificity, false-positive rate and likelihood ratios) and diagnostic accuracy [95% confidence intervals (CIs)] of ARIASs. A secondary analysis explored the influence of camera type and patients' ethnicity, age and sex on screening performance. Economic analysis estimated the cost per appropriate screening outcome identified.

RESULTS

A total of 20,258 patients with 102,856 images were entered into the study. The sensitivity point estimates of the ARIASs were as follows: EyeArt 94.7% (95% CI 94.2% to 95.2%) for any retinopathy, 93.8% (95% CI 92.9% to 94.6%) for referable retinopathy and 99.6% (95% CI 97.0% to 99.9%) for proliferative retinopathy; and Retmarker 73.0% (95% CI 72.0% to 74.0%) for any retinopathy, 85.0% (95% CI 83.6% to 86.2%) for referable retinopathy and 97.9% (95% CI 94.9 to 99.1%) for proliferative retinopathy. iGradingM classified all images as either 'disease' or 'ungradable', limiting further iGradingM analysis. The sensitivity and false-positive rates for EyeArt were not affected by ethnicity, sex or camera type but sensitivity declined marginally with increasing patient age. The screening performance of Retmarker appeared to vary with patient's age, ethnicity and camera type. Both EyeArt and Retmarker were cost saving relative to manual grading either as a replacement for level 1 human grading or used prior to level 1 human grading, although the latter was less cost-effective. A threshold analysis testing the highest ARIAS cost per patient before which ARIASs became more expensive per appropriate outcome than human grading, when used to replace level 1 grader, was Retmarker £3.82 and EyeArt £2.71 per patient.

LIMITATIONS

The non-randomised study design limited the health economic analysis but the same retinal images were processed by all ARIASs in this measurement comparison study.

CONCLUSIONS

Retmarker and EyeArt achieved acceptable sensitivity for referable retinopathy and false-positive rates (compared with human graders as reference standard) and appear to be cost-effective alternatives to a purely manual grading approach. Future work is required to develop technical specifications to optimise deployment and address potential governance issues.

FUNDING

The National Institute for Health Research (NIHR) Health Technology Assessment programme, a Fight for Sight Grant (Hirsch grant award) and the Department of Health's NIHR Biomedical Research Centre for Ophthalmology at Moorfields Eye Hospital and the University College London Institute of Ophthalmology.

An ensemble deep learning based approach for red lesion detection in fundus images

BACKGROUND AND OBJECTIVES

Diabetic retinopathy (DR) is one of the leading causes of preventable blindness in the world. Its earliest sign are red lesions, a general term that groups both microaneurysms (MAs) and hemorrhages (HEs). In daily clinical practice, these lesions are manually detected by physicians using fundus photographs. However, this task is tedious and time consuming, and requires an intensive effort due to the small size of the lesions and their lack of contrast. Computer-assisted diagnosis of DR based on red lesion detection is being actively explored due to its improvement effects both in clinicians consistency and accuracy. Moreover, it provides comprehensive feedback that is easy to assess by the physicians. Several methods for detecting red lesions have been proposed in the literature, most of them based on characterizing lesion candidates using hand crafted features, and classifying them into true or false positive detections. Deep learning based approaches, by contrast, are scarce in this domain due to the high expense of annotating the lesions manually.

METHODS

In this paper we propose a novel method for red lesion detection based on combining both deep learned and domain knowledge. Features learned by a convolutional neural network (CNN) are augmented by incorporating hand crafted features. Such ensemble vector of descriptors is used afterwards to identify true lesion candidates using a Random Forest classifier.

RESULTS

We empirically observed that combining both sources of information significantly improve results with respect to using each approach separately. Furthermore, our method reported the highest performance on a per-lesion basis on DIARETDB1 and e-ophtha, and for screening and need for referral on MESSIDOR compared to a second human expert.

CONCLUSIONS

Results highlight the fact that integrating manually engineered approaches with deep learned features is relevant to improve results when the networks are trained from lesion-level annotated data. An open source implementation of our system is publicly available at https://github.com/ignaciorlando/red-lesion-detection.

Automatic non-proliferative diabetic retinopathy screening system based on color fundus image

Background

Non-proliferative diabetic retinopathy is the early stage of diabetic retinopathy. Automatic detection of non-proliferative diabetic retinopathy is significant for clinical diagnosis, early screening and course progression of patients.

Methods

This paper introduces the design and implementation of an automatic system for screening non-proliferative diabetic retinopathy based on color fundus images. Firstly, the fundus structures, including blood vessels, optic disc and macula, are extracted and located, respectively. In particular, a new optic disc localization method using parabolic fitting is proposed based on the physiological structure characteristics of optic disc and blood vessels. Then, early lesions, such as microaneurysms, hemorrhages and hard exudates, are detected based on their respective characteristics. An equivalent optical model simulating human eyes is designed based on the anatomical structure of retina. Main structures and early lesions are reconstructed in the 3D space for better visualization. Finally, the severity of each image is evaluated based on the international criteria of diabetic retinopathy.

Results

The system has been tested on public databases and images from hospitals. Experimental results demonstrate that the proposed system achieves high accuracy for main structures and early lesions detection. The results of severity classification for non-proliferative diabetic retinopathy are also accurate and suitable.

Conclusions

Our system can assist ophthalmologists for clinical diagnosis, automatic screening and course progression of patients.

Automated Detection of Red Lesions Using Superpixel Multichannel Multifeature

Red lesions can be regarded as one of the earliest lesions in diabetic retinopathy (DR) and automatic detection of red lesions plays a critical role in diabetic retinopathy diagnosis. In this paper, a novel superpixel Multichannel Multifeature (MCMF) classification approach is proposed for red lesion detection. In this paper, firstly, a new candidate extraction method based on superpixel is proposed. Then, these candidates are characterized by multichannel features, as well as the contextual feature. Next, FDA classifier is introduced to classify the red lesions among the candidates. Finally, a postprocessing technique based on multiscale blood vessels detection is modified for removing nonlesions appearing as red. Experiments on publicly available DiaretDB1 database are conducted to verify the effectiveness of our proposed method.

Automatic Microaneurysms Detection Based on Multifeature Fusion Dictionary Learning

Recently, microaneurysm (MA) detection has attracted a lot of attention in the medical image processing community. Since MAs can be seen as the earliest lesions in diabetic retinopathy, their detection plays a critical role in diabetic retinopathy diagnosis. In this paper, we propose a novel MA detection approach named multifeature fusion dictionary learning (MFFDL). The proposed method consists of four steps: preprocessing, candidate extraction, multifeature dictionary learning, and classification. The novelty of our proposed approach lies in incorporating the semantic relationships among multifeatures and dictionary learning into a unified framework for automatic detection of MAs. We evaluate the proposed algorithm by comparing it with the state-of-the-art approaches and the experimental results validate the effectiveness of our algorithm.

Vessel segmentation and microaneurysm detection using discriminative dictionary learning and sparse representation

Diabetic retinopathy (DR) is a major cause of visual impairment, and the analysis of retinal image can assist patients to take action earlier when it is more likely to be effective. The accurate segmentation of blood vessels in the retinal image can diagnose DR directly. In this paper, a novel scheme for blood vessel segmentation based on discriminative dictionary learning (DDL) and sparse representation has been proposed. The proposed system yields a strong representation which contains the semantic concept of the image. To extract blood vessel, two separate dictionaries, for vessel and non-vessel, capable of providing reconstructive and discriminative information of the retinal image are learned. In the test step, an unseen retinal image is divided into overlapping patches and classified to vessel and non-vessel patches. Then, a voting scheme is applied to generate the binary vessel map. The proposed vessel segmentation method can achieve the accuracy of 95% and a sensitivity of 75% in the same range of specificity 97% on two public datasets. The results show that the proposed method can achieve comparable results to existing methods and decrease false positive vessels in abnormal retinal images with pathological regions. Microaneurysm (MA) is the earliest sign of DR that appears as a small red dot on the surface of the retina. Despite several attempts to develop automated MA detection systems, it is still a challenging problem. In this paper, a method for MA detection, which is similar to our vessel segmentation approach, is proposed. In our method, a candidate detection algorithm based on the Morlet wavelet is applied to identify all possible MA candidates. In the next step, two discriminative dictionaries with the ability to distinguish MA from non-MA object are learned. These dictionaries are then used to classify the detected candidate objects. The evaluations indicate that the proposed MA detection method achieves higher average sensitivity about 2-15%, compared to existing methods.

Detecting retinal microaneurysms and hemorrhages with robustness to the presence of blood vessels

BACKGROUND AND OBJECTIVES

Diabetic Retinopathy is the leading cause of blindness in developed countries in the age group 20-74 years. It is characterized by lesions on the retina and this paper focuses on detecting two of these lesions, Microaneurysms and Hemorrhages, which are also known as red lesions. This paper attempts to deal with two problems in detecting red lesions from retinal fundus images: (1) false detections on blood vessels; and (2) different size of red lesions.

METHODS

To deal with false detections on blood vessels, novel filters have been proposed which can distinguish between red lesions and blood vessels. This distinction is based on the fact that vessels are elongated while red lesions are usually circular blob-like structures. The second problem of the different size of lesions is dealt with by applying the proposed filters on patches of different sizes instead of filtering the full image. These patches are obtained by dividing the original image using a grid whose size determines the patch size. Different grid sizes were used and lesion detection results for these grid sizes were combined using Multiple Kernel Learning.

RESULTS

Experiments on a dataset of 143 images showed that proposed filters detected Microaneurysms and Hemorrhages successfully even when these lesions were close to blood vessels. In addition, using Multiple Kernel Learning improved the results when compared to using a grid of one size only. The areas under receiver operating characteristic curve were found to be 0.97 and 0.92 for Microaneurysms and Hemorrhages respectively which are better than the existing related works.

CONCLUSIONS

Proposed filters are robust to the presence of blood vessels and surpass related works in detecting red lesions from retinal fundus images. Improved lesion detection using the proposed approach can help in automatic detection of Diabetic Retinopathy.

Automated detection of diabetic retinopathy in retinal images

Diabetic retinopathy (DR) is a disease with an increasing prevalence and the main cause of blindness among working-age population. The risk of severe vision loss can be significantly reduced by timely diagnosis and treatment. Systematic screening for DR has been identified as a cost-effective way to save health services resources. Automatic retinal image analysis is emerging as an important screening tool for early DR detection, which can reduce the workload associated to manual grading as well as save diagnosis costs and time. Many research efforts in the last years have been devoted to developing automatic tools to help in the detection and evaluation of DR lesions. However, there is a large variability in the databases and evaluation criteria used in the literature, which hampers a direct comparison of the different studies. This work is aimed at summarizing the results of the available algorithms for the detection and classification of DR pathology. A detailed literature search was conducted using PubMed. Selected relevant studies in the last 10 years were scrutinized and included in the review. Furthermore, we will try to give an overview of the available commercial software for automatic retinal image analysis.

Operational Components of Telemedicine Programs for Diabetic Retinopathy

Diabetic retinopathy is a leading cause of new-onset vision loss worldwide. Treatments supported by large clinical trials are effective in preserving vision, but many persons do not receive timely diagnosis and treatment of diabetic retinopathy, which is typically asymptomatic when most treatable. Telemedicine evaluation to identify diabetic retinopathy has the potential to improve access to care and improve outcomes, but incomplete implementation of published standards creates a risk to program utility and sustainability. In a prior article, we reviewed the literature regarding the impact of imaging device, number and size of retinal images, pupil dilation, type of image grader, and diagnostic accuracy on telemedicine assessment for diabetic retinopathy. This article reviews the literature regarding the impact of automated image grading, cost effectiveness, program standards, and quality assurance (QA) on telemedicine assessment of diabetic retinopathy. Telemedicine assessment of diabetic retinopathy has the potential to preserve vision, but greater attention to development and implementation of standards is needed to better realize its potential.

Automated diabetic retinopathy imaging in Indian eyes: a pilot study

Aim:

To evaluate the efficacy of an automated retinal image grading system in diabetic retinopathy (DR) screening.

Materials and Methods:

Color fundus images of patients of a DR screening project were analyzed for the purpose of the study. For each eye two set of images were acquired, one centerd on the disk and the other centerd on the macula. All images were processed by automated DR screening software (Retmarker). The results were compared to ophthalmologist grading of the same set of photographs.

Results:

5780 images of 1445 patients were analyzed. Patients were screened into two categories DR or no DR. Image quality was high, medium and low in 71 (4.91%), 1117 (77.30%) and 257 (17.78%) patients respectively. Specificity and sensitivity for detecting DR in the high, medium and low group were (0.59, 0.91); (0.11, 0.95) and (0.93, 0.14).

Conclusion:

Automated retinal image screening system for DR had a high sensitivity in high and medium quality images. Automated DR grading software's hold promise in future screening programs.

Method of Quantifying Size of Retinal Hemorrhages in Eyes with Branch Retinal Vein Occlusion Using 14-Square Grid: Interrater and Intrarater Reliability

Purpose. To describe a method of quantifying the size of the retinal hemorrhages in branch retinal vein occlusion (BRVO) and to determine the interrater and intrarater reliabilities of these measurements. Methods. Thirty-five fundus photographs from 35 consecutive eyes with BRVO were studied. The fundus images were analyzed with Power-Point® software, and a grid of 14 squares was laid over the fundus image. Raters were asked to judge the percentage of each of the 14 squares that was covered by the hemorrhages, and the average of the 14 squares was taken to be the relative size of the retinal hemorrhage. Results. Interrater reliability between three raters was higher when a grid with 14 squares was used (intraclass correlation coefficient (ICC), 0.96) than that when a box with no grid was used (ICC, 0.78). Intrarater reliability, which was calculated by the retinal hemorrhage area measured on two different days, was also higher (ICC, 0.97) than that with no grid (ICC, 0.86). Interrater reliability for five fundus pictures with poor image quality was also good when a grid with 14 squares was used (ICC, 0.88). Conclusions. Although our method is subjective, excellent interrater and intrarater reliabilities indicate that this method can be adapted for clinical use.

A review on automatic analysis techniques for color fundus photographs

In this paper, we give a review on automatic image processing tools to recognize diseases causing specific distortions in the human retina. After a brief summary of the biology of the retina, we give an overview of the types of lesions that may appear as biomarkers of both eye and non-eye diseases. We present several state-of-the-art procedures to extract the anatomic components and lesions in color fundus photographs and decision support methods to help clinical diagnosis. We list publicly available databases and appropriate measurement techniques to compare quantitatively the performance of these approaches. Furthermore, we discuss on how the performance of image processing-based systems can be improved by fusing the output of individual detector algorithms. Retinal image analysis using mobile phones is also addressed as an expected future trend in this field.

Automated microaneurysm detection in diabetic retinopathy using curvelet transform

Microaneurysms (MAs) are known to be the early signs of diabetic retinopathy (DR). An automated MA detection system based on curvelet transform is proposed for color fundus image analysis. Candidates of MA were extracted in two parallel steps. In step one, blood vessels were removed from preprocessed green band image and preliminary MA candidates were selected by local thresholding technique. In step two, based on statistical features, the image background was estimated. The results from the two steps allowed us to identify preliminary MA candidates which were also present in the image foreground. A collection set of features was fed to a rule-based classifier to divide the candidates into MAs and non-MAs. The proposed system was tested with Retinopathy Online Challenge database. The automated system detected 162 MAs out of 336, thus achieved a sensitivity of 48.21% with 65 false positives per image. Counting MA is a means to measure the progression of DR. Hence, the proposed system may be deployed to monitor the progression of DR at early stage in population studies.

Retinal Microaneurysms Detection Using Gradient Vector Analysis and Class Imbalance Classification

Retinal microaneurysms (MAs) are the earliest clinically observable lesions of diabetic retinopathy. Reliable automated MAs detection is thus critical for early diagnosis of diabetic retinopathy. This paper proposes a novel method for the automated MAs detection in color fundus images based on gradient vector analysis and class imbalance classification, which is composed of two stages, i.e. candidate MAs extraction and classification. In the first stage, a candidate MAs extraction algorithm is devised by analyzing the gradient field of the image, in which a multi-scale log condition number map is computed based on the gradient vectors for vessel removal, and then the candidate MAs are localized according to the second order directional derivatives computed in different directions. Due to the complexity of fundus image, besides a small number of true MAs, there are also a large amount of non-MAs in the extracted candidates. Classifying the true MAs and the non-MAs is an extremely class imbalanced classification problem. Therefore, in the second stage, several types of features including geometry, contrast, intensity, edge, texture, region descriptors and other features are extracted from the candidate MAs and a class imbalance classifier, i.e., RUSBoost, is trained for the MAs classification. With the Retinopathy Online Challenge (ROC) criterion, the proposed method achieves an average sensitivity of 0.433 at 1/8, 1/4, 1/2, 1, 2, 4 and 8 false positives per image on the ROC database, which is comparable with the state-of-the-art approaches, and 0.321 on the DiaRetDB1 V2.1 database, which outperforms the state-of-the-art approaches.

Automatic detection of microaneurysms in retinal fundus images

Diabetic retinopathy (DR) is one of the leading causes of new cases of blindness. Early and accurate detection of microaneurysms (MAs) is important for diagnosis and grading of diabetic retinopathy. In this paper, a new method for the automatic detection of MAs in eye fundus images is proposed. The proposed method consists of four main steps: preprocessing, candidate extraction, feature extraction and classification. A total of 27 characteristic features which contain local features and profile features are extracted for KNN classifier to distinguish true MAs from spurious candidates. The proposed method has been evaluated on two public database: ROC and e-optha. The experimental result demonstrates the efficiency and effectiveness of the proposed method, and it has the potential to be used to diagnose DR clinically.

Localizing Microaneurysms in Fundus Images Through Singular Spectrum Analysis

GOAL

Reliable recognition of microaneurysms (MAs) is an essential task when developing an automated analysis system for diabetic retinopathy (DR) detection. In this study, we propose an integrated approach for automated MA detection with high accuracy.

METHODS

Candidate objects are first located by applying a dark object filtering process. Their cross-section profiles along multiple directions are processed through singular spectrum analysis. The correlation coefficient between each processed profile and a typical MA profile is measured and used as a scale factor to adjust the shape of the candidate profile. This is to increase the difference in their profiles between true MAs and other non-MA candidates. A set of statistical features of those profiles is then extracted for a K-nearest neighbor classifier.

RESULTS

Experiments show that by applying this process, MAs can be separated well from the retinal background, the most common interfering objects and artifacts.

CONCLUSION

The results have demonstrated the robustness of the approach when testing on large scale datasets with clinically acceptable sensitivity and specificity.

SIGNIFICANCE

The approach proposed in the evaluated system has great potential when used in an automated DR screening tool or for large scale eye epidemiology studies.

Automated Axon Counting in Rodent Optic Nerve Sections with AxonJ

We have developed a publicly available tool, AxonJ, which quantifies the axons in optic nerve sections of rodents stained with paraphenylenediamine (PPD). In this study, we compare AxonJ's performance to human experts on 100x and 40x images of optic nerve sections obtained from multiple strains of mice, including mice with defects relevant to glaucoma. AxonJ produced reliable axon counts with high sensitivity of 0.959 and high precision of 0.907, high repeatability of 0.95 when compared to a gold-standard of manual assessments and high correlation of 0.882 to the glaucoma damage staging of a previously published dataset. AxonJ allows analyses that are quantitative, consistent, fully-automated, parameter-free, and rapid on whole optic nerve sections at 40x. As a freely available ImageJ plugin that requires no highly specialized equipment to utilize, AxonJ represents a powerful new community resource augmenting studies of the optic nerve using mice.

Fast Convolutional Neural Network Training Using Selective Data Sampling: Application to Hemorrhage Detection in Color Fundus Images

Convolutional neural networks (CNNs) are deep learning network architectures that have pushed forward the state-of-the-art in a range of computer vision applications and are increasingly popular in medical image analysis. However, training of CNNs is time-consuming and challenging. In medical image analysis tasks, the majority of training examples are easy to classify and therefore contribute little to the CNN learning process. In this paper, we propose a method to improve and speed-up the CNN training for medical image analysis tasks by dynamically selecting misclassified negative samples during training. Training samples are heuristically sampled based on classification by the current status of the CNN. Weights are assigned to the training samples and informative samples are more likely to be included in the next CNN training iteration. We evaluated and compared our proposed method by training a CNN with (SeS) and without (NSeS) the selective sampling method. We focus on the detection of hemorrhages in color fundus images. A decreased training time from 170 epochs to 60 epochs with an increased performance-on par with two human experts-was achieved with areas under the receiver operating characteristics curve of 0.894 and 0.972 on two data sets. The SeS CNN statistically outperformed the NSeS CNN on an independent test set.

Red Lesion Detection Using Dynamic Shape Features for Diabetic Retinopathy Screening

The development of an automatic telemedicine system for computer-aided screening and grading of diabetic retinopathy depends on reliable detection of retinal lesions in fundus images. In this paper, a novel method for automatic detection of both microaneurysms and hemorrhages in color fundus images is described and validated. The main contribution is a new set of shape features, called Dynamic Shape Features, that do not require precise segmentation of the regions to be classified. These features represent the evolution of the shape during image flooding and allow to discriminate between lesions and vessel segments. The method is validated per-lesion and per-image using six databases, four of which are publicly available. It proves to be robust with respect to variability in image resolution, quality and acquisition system. On the Retinopathy Online Challenge's database, the method achieves a FROC score of 0.420 which ranks it fourth. On the Messidor database, when detecting images with diabetic retinopathy, the proposed method achieves an area under the ROC curve of 0.899, comparable to the score of human experts, and it outperforms state-of-the-art approaches.

Automatic differentiation of color fundus images containing drusen or exudates using a contextual spatial pyramid approach

We developed an automatic system to identify and differentiate color fundus images containing no lesions, drusen or exudates. Drusen and exudates are lesions with a bright appearance, associated with age-related macular degeneration and diabetic retinopathy, respectively. The system consists of three lesion detectors operating at pixel-level, combining their outputs using spatial pooling and classification with a random forest classifier. System performance was compared with ratings of two independent human observers using human-expert annotations as reference. Kappa agreements of 0.89, 0.97 and 0.92 and accuracies of 0.93, 0.98 and 0.95 were obtained for the system and observers, respectively.

Brightness-preserving fuzzy contrast enhancement scheme for the detection and classification of diabetic retinopathy disease

The contrast enhancement of retinal image plays a vital role for the detection of microaneurysms (MAs), which are an early sign of diabetic retinopathy disease. A retinal image contrast enhancement method has been presented to improve the MA detection technique. The success rate on low-contrast noisy retinal image analysis shows the importance of the proposed method. Overall, 587 retinal input images are tested for performance analysis. The average sensitivity and specificity are obtained as 95.94% and 99.21%, respectively. The area under curve is found as 0.932 for the receiver operating characteristics analysis. The classifications of diabetic retinopathy disease are also performed here. The experimental results show that the overall MA detection method performs better than the current state-of-the-art MA detection algorithms.

Exploiting ensemble learning for automatic cataract detection and grading

Cataract is defined as a lenticular opacity presenting usually with poor visual acuity. It is one of the most common causes of visual impairment worldwide. Early diagnosis demands the expertise of trained healthcare professionals, which may present a barrier to early intervention due to underlying costs. To date, studies reported in the literature utilize a single learning model for retinal image classification in grading cataract severity. We present an ensemble learning based approach as a means to improving diagnostic accuracy. Three independent feature sets, i.e., wavelet-, sketch-, and texture-based features, are extracted from each fundus image. For each feature set, two base learning models, i.e., Support Vector Machine and Back Propagation Neural Network, are built. Then, the ensemble methods, majority voting and stacking, are investigated to combine the multiple base learning models for final fundus image classification. Empirical experiments are conducted for cataract detection (two-class task, i.e., cataract or non-cataractous) and cataract grading (four-class task, i.e., non-cataractous, mild, moderate or severe) tasks. The best performance of the ensemble classifier is 93.2% and 84.5% in terms of the correct classification rates for cataract detection and grading tasks, respectively. The results demonstrate that the ensemble classifier outperforms the single learning model significantly, which also illustrates the effectiveness of the proposed approach.

Automatic detection of microaneurysms in diabetic retinopathy fundus images using the L*a*b color space

We develop an automated image processing system for detecting microaneurysm (MA) in diabetic patients. Diabetic retinopathy is one of the main causes of preventable blindness in working age diabetic people with the presence of an MA being one of the first signs. We transform the eye fundus images to the L*a*b* color space in order to separately process the L* and a* channels, looking for MAs in each of them. We then fuse the results, and last send the MA candidates to a k-nearest neighbors classifier for final assessment. The performance of the method, measured against 50 images with an ophthalmologist's hand-drawn ground-truth, shows high sensitivity (100%) and accuracy (84%), and running times around 10 s. This kind of automatic image processing application is important in order to reduce the burden on the public health system associated with the diagnosis of diabetic retinopathy given the high number of potential patients that need periodic screening.