Automated segmentation of retinal blood vessels and identification of proliferative diabetic retinopathy

Box-Counting Fractal Analysis: A Primer for the Clinician

This chapter lays out the elementary principles of fractal geometry underpinning much of the rest of this book. It assumes a minimal mathematical background, defines the key principles and terms in context, and outlines the basics of a fractal analysis method known as box counting and how it is used to perform fractal, lacunarity, and multifractal analyses. As a standalone reference, this chapter grounds the reader to be able to understand, evaluate, and apply essential methods to appreciate and heal the exquisitely detailed fractal geometry of the brain.

Proliferative Diabetic Retinopathy Diagnosis Using Varying-Scales Filter Banks and Double-Layered Thresholding

Diabetic retinopathy is one of the abnormalities of the retina in which a diabetic patient suffers from severe vision loss due to an affected retina. Proliferative diabetic retinopathy (PDR) is the final and most critical stage of diabetic retinopathy. Abnormal and fragile blood vessels start to grow on the surface of the retina at this stage. It causes retinal detachment, which may lead to complete blindness in severe cases. In this paper, a novel method is proposed for the detection and grading of neovascularization. The proposed system first performs pre-processing on input retinal images to enhance the vascular pattern, followed by blood vessel segmentation and optic disc localization. Then various features are tested on the candidate regions with different thresholds. In this way, positive and negative advanced diabetic retinopathy cases are separated. Optic disc coordinates are applied for the grading of neovascularization as NVD or NVE. The proposed algorithm improves the quality of automated diagnostic systems by eliminating normal blood vessels and exudates that might cause hindrances in accurate disease detection, thus resulting in more accurate detection of abnormal blood vessels. The evaluation of the proposed system has been carried out using performance parameters such as sensitivity, specificity, accuracy, and positive predictive value (PPV) on a publicly available standard retinal image database and one of the locally available databases. The proposed algorithm gives an accuracy of 98.5% and PPV of 99.8% on MESSIDOR and an accuracy of 96.5% and PPV of 100% on the local database.

Improving foveal avascular zone segmentation in fluorescein angiograms by leveraging manual vessel labels from public color fundus pictures

In clinical routine, ophthalmologists frequently analyze the shape and size of the foveal avascular zone (FAZ) to detect and monitor retinal diseases. In order to extract those parameters, the contours of the FAZ need to be segmented, which is normally achieved by analyzing the retinal vasculature (RV) around the macula in fluorescein angiograms (FA). Computer-aided segmentation methods based on deep learning (DL) can automate this task. However, current approaches for segmenting the FAZ are often tailored to a specific dataset or require manual initialization. Furthermore, they do not take the variability and challenges of clinical FA into account, which are often of low quality and difficult to analyze. In this paper we propose a DL-based framework to automatically segment the FAZ in challenging FA scans from clinical routine. Our approach mimics the workflow of retinal experts by using additional RV labels as a guidance during training. Hence, our model is able to produce RV segmentations simultaneously. We minimize the annotation work by using a multi-modal approach that leverages already available public datasets of color fundus pictures (CFPs) and their respective manual RV labels. Our experimental evaluation on two datasets with FA from 1) clinical routine and 2) large multicenter clinical trials shows that the addition of weak RV labels as a guidance during training improves the FAZ segmentation significantly with respect to using only manual FAZ annotations.

Neovascularization Detection and Localization in Fundus Images Using Deep Learning

Proliferative Diabetic Retinopathy (PDR) is a severe retinal disease that threatens diabetic patients. It is characterized by neovascularization in the retina and the optic disk. PDR clinical features contain highly intense retinal neovascularization and fibrous spreads, leading to visual distortion if not controlled. Different image processing techniques have been proposed to detect and diagnose neovascularization from fundus images. Recently, deep learning methods are getting popular in neovascularization detection due to artificial intelligence advancement in biomedical image processing. This paper presents a semantic segmentation convolutional neural network architecture for neovascularization detection. First, image pre-processing steps were applied to enhance the fundus images. Then, the images were divided into small patches, forming a training set, a validation set, and a testing set. A semantic segmentation convolutional neural network was designed and trained to detect the neovascularization regions on the images. Finally, the network was tested using the testing set for performance evaluation. The proposed model is entirely automated in detecting and localizing neovascularization lesions, which is not possible with previously published methods. Evaluation results showed that the model could achieve accuracy, sensitivity, specificity, precision, Jaccard similarity, and Dice similarity of 0.9948, 0.8772, 0.9976, 0.8696, 0.7643, and 0.8466, respectively. We demonstrated that this model could outperform other convolutional neural network models in neovascularization detection.

Performance and Limitation of Machine Learning Algorithms for Diabetic Retinopathy Screening: Meta-analysis

Background

Diabetic retinopathy (DR), whose standard diagnosis is performed by human experts, has high prevalence and requires a more efficient screening method. Although machine learning (ML)–based automated DR diagnosis has gained attention due to recent approval of IDx-DR, performance of this tool has not been examined systematically, and the best ML technique for use in a real-world setting has not been discussed.

Objective

The aim of this study was to systematically examine the overall diagnostic accuracy of ML in diagnosing DR of different categories based on color fundus photographs and to determine the state-of-the-art ML approach.

Methods

Published studies in PubMed and EMBASE were searched from inception to June 2020. Studies were screened for relevant outcomes, publication types, and data sufficiency, and a total of 60 out of 2128 (2.82%) studies were retrieved after study selection. Extraction of data was performed by 2 authors according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses), and the quality assessment was performed according to the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2). Meta-analysis of diagnostic accuracy was pooled using a bivariate random effects model. The main outcomes included diagnostic accuracy, sensitivity, and specificity of ML in diagnosing DR based on color fundus photographs, as well as the performances of different major types of ML algorithms.

Results

The primary meta-analysis included 60 color fundus photograph studies (445,175 interpretations). Overall, ML demonstrated high accuracy in diagnosing DR of various categories, with a pooled area under the receiver operating characteristic (AUROC) ranging from 0.97 (95% CI 0.96-0.99) to 0.99 (95% CI 0.98-1.00). The performance of ML in detecting more-than-mild DR was robust (sensitivity 0.95; AUROC 0.97), and by subgroup analyses, we observed that robust performance of ML was not limited to benchmark data sets (sensitivity 0.92; AUROC 0.96) but could be generalized to images collected in clinical practice (sensitivity 0.97; AUROC 0.97). Neural network was the most widely used method, and the subgroup analysis revealed a pooled AUROC of 0.98 (95% CI 0.96-0.99) for studies that used neural networks to diagnose more-than-mild DR.

Conclusions

This meta-analysis demonstrated high diagnostic accuracy of ML algorithms in detecting DR on color fundus photographs, suggesting that state-of-the-art, ML-based DR screening algorithms are likely ready for clinical applications. However, a significant portion of the earlier published studies had methodology flaws, such as the lack of external validation and presence of spectrum bias. The results of these studies should be interpreted with caution.

Artificial Intelligence in the assessment of diabetic retinopathy from fundus photographs

Background: Over the next 25 years, the global prevalence of diabetes is expected to grow to affect 700 million individuals. Consequently, an unprecedented number of patients will be at risk for vision loss from diabetic eye disease. This demand will almost certainly exceed the supply of eye care professionals to individually evaluate each patient on an annual basis, signaling the need for 21st century tools to assist our profession in meeting this challenge. Methods: Review of available literature on artificial intelligence (AI) as applied to diabetic retinopathy (DR) detection and predictionResults: The field of AI has seen exponential growth in evaluating fundus photographs for DR. AI systems employ machine learning and artificial neural networks to teach themselves how to grade DR from libraries of tens of thousands of images and may be able to predict future DR progression based on baseline fundus photographs. Conclusions: AI algorithms are highly promising for the purposes of DR detection and will likely be able to reliably predict DR worsening in the future. A deeper understanding of these systems and how they interpret images is critical as they transition from the bench into the clinic.

Local characterization of neovascularization and identification of proliferative diabetic retinopathy in retinal fundus images

Neovascularization (NV) is a characteristic of the onset of sight-threatening stage of DR, called proliferative DR (PDR). Identification of PDR requires modeling of these unregulated ill-formed vessels, and other associated signs of PDR. We present an approach that models the micro-pattern of local variations (using texture based analysis) and quantifies structural changes in vessel patterns in localized patches, to arrive at a score of neovascularity. The distribution of patch-level confidence scores is collated into an image-level decision of presence or absence of PDR. Evaluated on a dataset of 779 images combining public data and clinical data from local hospitals, the patch-level neovascularity prediction has a sensitivity of 92.4% at 92.6% specificity. For image-level PDR identification our method is shown to achieve sensitivity of 83.3% at a high specificity operating point of 96.1% specificity, and specificity of 83% at high sensitivity operating point of 92.2% sensitivity. Our approach could have potential application in DR grading where it can localize NVE regions and identify PDR images for immediate intervention.

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.

Automated multi-lesion detection for referable diabetic retinopathy in indigenous health care

Diabetic Retinopathy (DR) is a complication of diabetes mellitus that affects more than one-quarter of the population with diabetes, and can lead to blindness if not discovered in time. An automated screening enables the identification of patients who need further medical attention. This study aimed to classify retinal images of Aboriginal and Torres Strait Islander peoples utilizing an automated computer-based multi-lesion eye screening program for diabetic retinopathy. The multi-lesion classifier was trained on 1,014 images from the São Paulo Eye Hospital and tested on retinal images containing no DR-related lesion, single lesions, or multiple types of lesions from the Inala Aboriginal and Torres Strait Islander health care centre. The automated multi-lesion classifier has the potential to enhance the efficiency of clinical practice delivering diabetic retinopathy screening. Our program does not necessitate image samples for training from any specific ethnic group or population being assessed and is independent of image pre- or post-processing to identify retinal lesions. In this Aboriginal and Torres Strait Islander population, the program achieved 100% sensitivity and 88.9% specificity in identifying bright lesions, while detection of red lesions achieved a sensitivity of 67% and specificity of 95%. When both bright and red lesions were present, 100% sensitivity with 88.9% specificity was obtained. All results obtained with this automated screening program meet WHO standards for diabetic retinopathy screening.

Genetic algorithm based feature selection combined with dual classification for the automated detection of proliferative diabetic retinopathy

Proliferative diabetic retinopathy (PDR) is a condition that carries a high risk of severe visual impairment. The hallmark of PDR is the growth of abnormal new vessels. In this paper, an automated method for the detection of new vessels from retinal images is presented. This method is based on a dual classification approach. Two vessel segmentation approaches are applied to create two separate binary vessel map which each hold vital information. Local morphology features are measured from each binary vessel map to produce two separate 4-D feature vectors. Independent classification is performed for each feature vector using a support vector machine (SVM) classifier. The system then combines these individual outcomes to produce a final decision. This is followed by the creation of additional features to generate 21-D feature vectors, which feed into a genetic algorithm based feature selection approach with the objective of finding feature subsets that improve the performance of the classification. Sensitivity and specificity results using a dataset of 60 images are 0.9138 and 0.9600, respectively, on a per patch basis and 1.000 and 0.975, respectively, on a per image basis.

Automated detection of proliferative diabetic retinopathy using a modified line operator and dual classification

Proliferative diabetic retinopathy (PDR) is a condition that carries a high risk of severe visual impairment. The hallmark of PDR is neovascularisation, the growth of abnormal new vessels. This paper describes an automated method for the detection of new vessels in retinal images. Two vessel segmentation approaches are applied, using the standard line operator and a novel modified line operator. The latter is designed to reduce false responses to non-vessel edges. Both generated binary vessel maps hold vital information which must be processed separately. This is achieved with a dual classification system. Local morphology features are measured from each binary vessel map to produce two separate feature sets. Independent classification is performed for each feature set using a support vector machine (SVM) classifier. The system then combines these individual classification outcomes to produce a final decision. Sensitivity and specificity results using a dataset of 60 images are 0.862 and 0.944 respectively on a per patch basis and 1.00 and 0.90 respectively on a per image basis.

Extraction of Blood Vessels in Retinal Images Using Four Different Techniques

A variety of blood vessel extraction (BVE) techniques exist in the literature, but they do not always lead to acceptable solutions especially in the presence of anomalies where the reported work is limited. Four techniques are presented for BVE: (1) BVE using Image Line Cross-Sections (ILCS), (2) BVE using Edge Enhancement and Edge Detection (EEED), (3) BVE using Modified Matched Filtering (MMF), and (4) BVE using Continuation Algorithm (CA). These four techniques have been designed especially for abnormal retinal images containing low vessel contrasts, drusen, exudates, and other artifacts. The four techniques were applied to 30 abnormal retinal images, and the success rate was found to be (95 to 99%) for CA, (88–91%) for EEED, (80–85%) for MMF, and (74–78%) for ILCS. Application of these four techniques to 105 normal retinal images gave improved results: (99-100%) for CA, (96–98%) for EEED, (94-95%) for MMF, and (88–93%) for ILCS. Investigations revealed that the four techniques in the order of increasing performance could be arranged as ILCS, MMF, EEED, and CA. Here we demonstrate these four techniques for abnormal retinal images only. ILCS, EEED, and CA are novel additions whereas MMF is an improved and modified version of an existing matched filtering technique. CA is a promising technique.

Diabetic retinopathy grading by digital curvelet transform

One of the major complications of diabetes is diabetic retinopathy. As manual analysis and diagnosis of large amount of images are time consuming, automatic detection and grading of diabetic retinopathy are desired. In this paper, we use fundus fluorescein angiography and color fundus images simultaneously, extract 6 features employing curvelet transform, and feed them to support vector machine in order to determine diabetic retinopathy severity stages. These features are area of blood vessels, area, regularity of foveal avascular zone, and the number of micro-aneurisms therein, total number of micro-aneurisms, and area of exudates. In order to extract exudates and vessels, we respectively modify curvelet coefficients of color fundus images and angiograms. The end points of extracted vessels in predefined region of interest based on optic disk are connected together to segment foveal avascular zone region. To extract micro-aneurisms from angiogram, first extracted vessels are subtracted from original image, and after removing detected background by morphological operators and enhancing bright small pixels, micro-aneurisms are detected. 70 patients were involved in this study to classify diabetic retinopathy into 3 groups, that is, (1) no diabetic retinopathy, (2) mild/moderate nonproliferative diabetic retinopathy, (3) severe nonproliferative/proliferative diabetic retinopathy, and our simulations show that the proposed system has sensitivity and specificity of 100% for grading.

Improving vessel segmentation in ultra-wide field-of-view retinal fluorescein angiograms

Vessel segmentation on ultra-wide field-of-view fluorescein angiogram sequences of the retina is a challenging problem. Vessel appearance undergoes severe changes, as different portions of the vascular structure become perfused in different frames. This paper presents a method for segmenting vessels in such sequences using steerable filters and automatic thresholding. We introduce a penalization stage on regions with high vessel response in the filtered image, improving the detection of peripheral vessels and reducing false positives around the optic disc and in regions of choroidal vessels and lesions. Quantitative results are provided, in which the penalization stage improves the segmentation precision segmentation by 11.84%, the recall by 12.98% and the accuracy by 0.40%. To facilitate further evaluation, usage, and algorithm comparison, the algorithm, the data set used, the ground truth, and the results are made available on the internet.

Detection of new vessels on the optic disc using retinal photographs

Proliferative diabetic retinopathy is a rare condition likely to lead to severe visual impairment. It is characterized by the development of abnormal new retinal vessels. We describe a method for automatically detecting new vessels on the optic disc using retinal photography. Vessel-like candidate segments are first detected using a method based on watershed lines and ridge strength measurement. Fifteen feature parameters, associated with shape, position, orientation, brightness, contrast and line density are calculated for each candidate segment. Based on these features, each segment is categorized as normal or abnormal using a support vector machine (SVM) classifier. The system was trained and tested by cross-validation using 38 images with new vessels and 71 normal images from two diabetic retinal screening centers and one hospital eye clinic. The discrimination performance of the fifteen features was tested against a clinical reference standard. Fourteen features were found to be effective and used in the final test. The area under the receiver operator characteristic curve was 0.911 for detecting images with new vessels on the disc. This accuracy may be sufficient for it to play a useful clinical role in an automated retinopathy analysis system.

Analysis of foveal avascular zone in colour fundus images for grading of diabetic retinopathy severity

Diabetic retinopathy (DR) is a sight threatening complication due to diabetes mellitus that affects the retina. At present, the classification of DR is based on the International Clinical Diabetic Retinopathy Disease Severity. In this paper, FAZ enlargement with DR progression is investigated to enable a new and an effective grading protocol DR severity in an observational clinical study. The performance of a computerised DR monitoring and grading system that digitally analyses colour fundus image to measure the enlargement of FAZ and grade DR is evaluated. The range of FAZ area is optimised to accurately determine DR severity stage and progression stages using a Gaussian Bayes classifier. The system achieves high accuracies of above 96%, sensitivities higher than 88% and specificities higher than 96%, in grading of DR severity. In particular, high sensitivity (100%), specificity (>98%) and accuracy (99%) values are obtained for No DR (normal) and Severe NPDR/PDR stages. The system performance indicates that the DR system is suitable for early detection of DR and for effective treatment of severe cases.

Vessel network detection using contour evolution and color components

Automated retinal screening relies on vasculature segmentation before the identification of other anatomical structures of the retina. Vasculature extraction can also be input to image quality ranking, neovascularization detection and image registration. An extensive related literature often excludes the inherent heterogeneity of ophthalmic clinical images. The contribution of this paper consists in an algorithm using front propagation to segment the vessel network, including a penalty on the wait queue to the fast marching method, which minimizes leakage of the evolving boundary. The algorithm requires no manual labeling of seeds, a minimum number of parameters and it is capable of segmenting color ocular fundus images in real scenarios, where multi-ethnicity and brightness variations are parts of the problem.

Deformable registration of retinal fluorescein angiogram sequences using vasculature structures

State-of-the-art deformable registration algorithms do not perform as well with FA sequences because they are designed to deal with changes of content appearance (e.g., due to different sensors imaging the same organs) but not with content changes, which occur throughout a FA sequence as different portions or the vascular structure are visible (perfused) in different frames. This paper presents a frame-to-frame registration algorithm for ultra-wide-field-of-view (UWFV) fluorescein angiograms (FA) of the retina, based on deformable alignment of the retinal vasculature structure. Comparative experiments on an initial set of UWFV FAs indicate that, thanks to its specialization, our technique outperforms one of the best state-of-the-art methods for multimodal image registration when dealing with the demanding characteristics of the UWFV FA sequences.