scholarly journals In Search for the Optimal Preprocessing Technique for Deep Learning Based Diabetic Retinopathy Stage Classification from Fundus Images

2021 ◽  
Author(s):  
Nilarun Mukherjee ◽  
Souvik Sengupta
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Vision Loss ◽  
Fundus Images ◽  
New Combinations ◽  
Z Score ◽  
Kappa Score ◽  
Stage Classification ◽  
Classification Tasks ◽  
Intensity Normalization

Abstract Background: Diabetic retinopathy (DR) is a complication of diabetes mellitus, which if left untreated may lead to complete vision loss. Early diagnosis and treatment is the key to prevent further complications of DR. Computer-aided diagnosis is a very effective method to support ophthalmologists, as manual inspection of pathological changes in retina images are time consuming and expensive. In recent times, Machine Learning and Deep Learning techniques have subsided conventional rule based approaches for detection, segmentation and classification of DR stages and lesions in fundus images. Method: In this paper, we present a comparative study of the different state-of-the-art preprocessing methods that have been used in deep learning based DR classification tasks in recent times and also propose a new unsupervised learning based retinal region extraction technique and new combinations of preprocessing pipelines designed on top of it. Efficacy of different existing and new combinations of the preprocessing methods are analyzed using two publicly available retinal datasets (EyePACS and APTOS) for different DR stage classification tasks, such as referable DR, DR screening, and five-class DR grading, using a benchmark deep learning model (ResNet-50). Results: It has been observed that the proposed preprocessing strategy composed of region of interest extraction through K-means clustering followed by contrast and edge enhancement using Graham’s method and z-score intensity normalization achieved the highest accuracy of 98.5%, 96.51% and 90.59% in DR-screening, referable-DR, and DR gradation tasks respectively and also achieved the best quadratic weighted kappa score of 0.945 in DR grading task. It achieved best AUC-ROC of 0.98 and 0.9981 in DR grading and DR screening tasks respectively. Conclusion: It is evident from the results that the proposed preprocessing pipeline composed of the proposed ROI extraction through K-means clustering, followed by edge and contrast enhancement using Graham’s method and then z-score intensity normalization outperforms all other existing preprocessing pipelines and has proven to be the most effective preprocessing strategy in helping the baseline CNN model to extract meaningful deep features.

An Explainable Deep Learning Ensemble Model for Robust Diagnosis of Diabetic Retinopathy Grading

2021 ◽  
Vol 17 (3s) ◽  
pp. 1-24
Author(s):  
Mohammad Shorfuzzaman ◽  
M. Shamim Hossain ◽  
Abdulmotaleb El Saddik
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Vision Loss ◽  
Final Diagnosis ◽  
Learning Rate ◽  
Ensemble Model ◽  
Fundus Images ◽  
Robust Detection ◽  
Retinal Fundus Images ◽  
Retinal Fundus

Diabetic retinopathy (DR) is one of the most common causes of vision loss in people who have diabetes for a prolonged period. Convolutional neural networks (CNNs) have become increasingly popular for computer-aided DR diagnosis using retinal fundus images. While these CNNs are highly reliable, their lack of sufficient explainability prevents them from being widely used in medical practice. In this article, we propose a novel explainable deep learning ensemble model where weights from different models are fused into a single model to extract salient features from various retinal lesions found on fundus images. The extracted features are then fed to a custom classifier for the final diagnosis of DR severity level. The model is trained on an APTOS dataset containing retinal fundus images of various DR grades using a cyclical learning rates strategy with an automatic learning rate finder for decaying the learning rate to improve model accuracy. We develop an explainability approach by leveraging gradient-weighted class activation mapping and shapely adaptive explanations to highlight the areas of fundus images that are most indicative of different DR stages. This allows ophthalmologists to view our model's decision in a way that they can understand. Evaluation results using three different datasets (APTOS, MESSIDOR, IDRiD) show the effectiveness of our model, achieving superior classification rates with a high degree of precision (0.970), sensitivity (0.980), and AUC (0.978). We believe that the proposed model, which jointly offers state-of-the-art diagnosis performance and explainability, will address the black-box nature of deep CNN models in robust detection of DR grading.

scholarly journals A Survey: Deep Learning Methods on Diabetic Retinopathy

2021 ◽  
pp. 131-135
Author(s):  
Kamlesh Raghuwanshi ◽  
Vipin Tiwari
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Vision Loss ◽  
The Body ◽  
Fundus Images ◽  
Learning Approaches ◽  
Severe Vision Loss ◽  
Pros And Cons ◽  
Retinal Fundus ◽  
Aided Diagnosis

Diabetes Mellitus (DM) is a metabolic condition that arises because of the elevated level of blood sugar in the body which triggers eye deficiency, also known as Diabetic Retinopathy (DR) which causes severe vision loss. An effective and efficient tool for early DR diagnosis and assisting experts is a computer-aided diagnosis (CAD) device focused on retinal fundus images that can detect this problem. A CAD method requires different phases in fundus images, such as identification, segmentation and lesion classification. Recent advancement of deep learning (DL) and its definitive victory over conventional ML approaches inspired researchers for implementation of many deep-learning-based techniques using different phases of fundus images. This paper highlights these deep learning approaches along with their pros and cons.

scholarly journals Deep Learning for Diabetic Retinopathy Detection and Classification Based on Fundus Images: A Review

2021 ◽  
pp. 104599
Author(s):  
Nikos Tsiknakis ◽  
Dimitris Theodoropoulos ◽  
Georgios Manikis ◽  
Emmanouil Ktistakis ◽  
Ourania Boutsora ◽  
...  
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Fundus Images

Diabetic Retinopathy Detection and Classification from Fundus Images Using Deep Learning

2021 ◽  
Author(s):  
Pratiksha Shetgaonkar ◽  
Shailendra Aswale ◽  
Saurabh Naik ◽  
Amey Gaonkar ◽  
Swapnil Gawade ◽  
...  
Keyword(s):  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Fundus Images

scholarly journals Classification among Microaneurysms, Exudates, and Lesion free Retinal Regions in the Eye Images using Transfer Learned CNNs

2019 ◽  
Vol 9 (2) ◽  
pp. 5508-5512
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Diabetic Retinopathy ◽  
Glucose Level ◽  
Vision Loss ◽  
Early Stage ◽  
Training Data ◽  
Fundus Images ◽  
Diabetic Mellitus ◽  
Start Process

When pancreas fails to secrete sufficient insulin in the human body, the glucose level in blood either becomes too high or too low. This fluctuation in glucose level affects different body organs such as kidney, brain, and eye. When the complications start appearing in the eyes due to Diabetic Mellitus (DM), it is called Diabetic Retinopathy (DR). DR can be categorized in several classes based on the severity, it can be Microaneurysms (ME), Haemorrhages (HE), Hard and Soft Exudates (EX and SE). DR is a slow start process that starts with very mild symptoms, becomes moderate with the time and results in complete vision loss, if not detected on time. Early-stage detection may greatly bolster in vision loss. However, it is impassable to detect the symptoms of DR with naked eyes. Ophthalmologist harbor to the several approaches and algorithm which makes use of different Machine Learning (ML) methods and classifiers to overcome this disease. The burgeoning insistence of Convolutional Neural Network (CNN) and their advancement in extracting features from different fundus images captivate several researchers to strive on it. Transfer Learning (TL) techniques help to use pre-trained CNN on a dataset that has finite training data, especially that in under developing countries. In this work, we propose several CNN architecture along with distinct classifiers which segregate the different lesions (ME and EX) in DR images with very eye-catching accuracies.

Construction of benchmark retinal image database for diabetic retinopathy analysis

2020 ◽  
Vol 234 (9) ◽  
pp. 1036-1048
Author(s):  
Jaskirat Kaur ◽  
Deepti Mittal
Keyword(s):  
Diabetic Retinopathy ◽  
Vision Loss ◽  
Medical Condition ◽  
Diagnostic Technique ◽  
Performance Comparison ◽  
Fundus Images ◽  
Benchmark Database ◽  
Computer Aided ◽  
Retinal Fundus Images ◽  
Retinal Fundus

Diabetic retinopathy, a symptomless medical condition of diabetes, is one of the significant reasons of vision impairment all over the world. The prior detection and diagnosis can decrease the occurrence of acute vision loss and enhance efficiency of treatment. Fundus imaging, a non-invasive diagnostic technique, is the most frequently used mode for analyzing retinal abnormalities related to diabetic retinopathy. Computer-aided methods based on retinal fundus images support quick diagnosis, impart an additional perspective during decision-making, and behave as an efficient means to assess response of treatment on retinal abnormalities. However, in order to evaluate computer-aided systems, a benchmark database of clinical retinal fundus images is required. Therefore, a representative database comprising of 2942 clinical retinal fundus images is developed and presented in this work. This clinical database, having varying attributes such as position, dimensions, shapes, and color, is formed to evaluate the generalization capability of computer-aided systems for diabetic retinopathy diagnosis. A framework for the development of benchmark retinal fundus images database is also proposed. The developed database comprises of medical image annotations for each image from expert ophthalmologists corresponding to anatomical structures, retinal lesions and stage of diabetic retinopathy. In addition, the substantial performance comparison capability of the proposed database aids in analyzing candidature of different methods, and subsequently its usage in medical practice for real-time applications.

scholarly journals Early Detection of Diabetic Retinopathy Using PCA-Firefly Based Deep Learning Model

Electronics ◽  
2020 ◽  
Vol 9 (2) ◽  
pp. 274 ◽  
Author(s):  
Thippa Reddy Gadekallu ◽  
Neelu Khare ◽  
Sweta Bhattacharya ◽  
Saurabh Singh ◽  
Praveen Kumar Reddy Maddikunta ◽  
...  
Keyword(s):  
Machine Learning ◽  
Diabetic Retinopathy ◽  
Deep Learning ◽  
Early Detection ◽  
Dimensionality Reduction ◽  
Vision Loss ◽  
Principal Component ◽  
Screening Methods ◽  
Learning Approaches ◽  
Time Period

Diabetic Retinopathy is a major cause of vision loss and blindness affecting millions of people across the globe. Although there are established screening methods - fluorescein angiography and optical coherence tomography for detection of the disease but in majority of the cases, the patients remain ignorant and fail to undertake such tests at an appropriate time. The early detection of the disease plays an extremely important role in preventing vision loss which is the consequence of diabetes mellitus remaining untreated among patients for a prolonged time period. Various machine learning and deep learning approaches have been implemented on diabetic retinopathy dataset for classification and prediction of the disease but majority of them have neglected the aspect of data pre-processing and dimensionality reduction, leading to biased results. The dataset used in the present study is a diabetes retinopathy dataset collected from the UCI machine learning repository. At its inceptions, the raw dataset is normalized using the Standardscalar technique and then Principal Component Analysis (PCA) is used to extract the most significant features in the dataset. Further, Firefly algorithm is implemented for dimensionality reduction. This reduced dataset is fed into a Deep Neural Network Model for classification. The results generated from the model is evaluated against the prevalent machine learning models and the results justify the superiority of the proposed model in terms of Accuracy, Precision, Recall, Sensitivity and Specificity.

CURVELET FUSION ENHACEMENT BASED EVALUATION OF DIABETIC RETINOPATHY BY THE IDENTIFICATION OF EXUDATES IN OPTIC COLOR FUNDUS IMAGES

2016 ◽  
Vol 28 (06) ◽  
pp. 1650046
Author(s):  
V. Ratna Bhargavi ◽  
Ranjan K. Senapati
Keyword(s):  
Diabetic Retinopathy ◽  
Vision Loss ◽  
Detection System ◽  
Clinical Signs ◽  
Posterior Part ◽  
Curvelet Transform ◽  
Optic Disk ◽  
Support Vector ◽  
Fundus Images ◽  
Statistical Measures

Rapid growth of Diabetes mellitus in people causes damage to posterior part of eye vessel structures. Diabetic retinopathy (DR) is an important hurdle in diabetic people and it causes lesion formation in retina due to retinal vessel structures damage. Bright lesions (BLs) or exudates are initial clinical signs of DR. Early BLs detection can help avoiding vision loss. The severity can be recognized based on number of BLs formed in the color fundus image. Manually diagnosing a large amount of images is time consuming. So a computerized DR grading and BLs detection system is proposed. In this paper for BLs detection, curvelet fusion enhancement is done initially because bright objects maps to largest coefficients in an image by utilizing the curvelet transform, so that BLs can be recognized in the retina easily. Then optic disk (OD) appearance is similar to BLs and vessel structures are barriers for lesion exact detection and moreover OD falsely classified as BLs and that increases false positives in classification. So these structures are segmented and eliminated by thresholding techniques. Various features were obtained from detected BLs. Publicly available databases are used for DR severity testing. 260 fundus images were used for the performance evaluation of proposed work. The support vector machine classifier (SVM) used to separate fundus images in various levels of DR based on feature set extracted. The proposed system that obtained the statistical measures were sensitivity 100%, specificity 95.4% and accuracy 97.74%. Compared to existing state-of-art techniques, the proposed work obtained better results in terms of sensitivity, specificity and accuracy.

Sign in / Sign up

Export Citation Format

Share Document