Submitted By: AHMAD NADEEM (MSCSF20M024) FAHAD BIN ASIF (MSCSF21M541)
Punjab University College of Information and Technology
University of the Punjab, Lahore. Project Title: Detection of Diabetic Retinopathy
Research Paper Title:
DIAGNOSIS OF DIABETIC RETINOPATHY USING MACHINE LEARNING TECHNIQUES Abstract: To diagnose diabetic retinopathy, In this paper, three models like Probabilistic Neural network (PNN), Bayesian Classification and Support vector machine (SVM) are described and their performances are compared. The amount of the disease spread in the retina can be identified by extracting the features of the retina. The features like blood vessels, haemmoraghes of non-proliferative diabetic retinopathy (NPDR) image and exudates of proliferative diabetic retinopathy (PDR) image are extracted from the raw images using the image processing techniques and fed to the classifier for classification. A total of 350 fundus images were used, out of which 100 were used for training and 250 images were used for testing. Experimental results show that PNN has an accuracy of 89.6 % Bayes Classifier has an accuracy of 94.4% and SVM has an accuracy of 97.6%. This infers that the SVM model outperforms all other models. Also their system is run on 130 images available from “DIARETDB0: Evaluation Database and Methodology for Diabetic Retinopathy” and the results show that PNN has an accuracy of 87.69% Bayes Classifier has an accuracy of 90.76% and SVM has an accuracy of 95.38%.
Research Paper Title:
DETECTING CLINICAL FEATURES OF DIABETIC RETINOPATHY USING IMAGE PROCESSING Abstract: This paper deals with detecting the diabetic retinopathy retinal changes. Detection of the disease at an early stage enables the patient to get treatment by advanced methods like laser treatment to prevent total blindness. The retinal images are first subjected to pre-processing techniques like colour normalization and enhancement process. Fundus image obtained is first resized into true colour 576×768 at 24 bits, images. The resized image undergoes enhancement operation and median filtering is applied to reduce noise. The enhanced colour image obtained is converted into grayscale. This gray image is subtracted from the median filtered image. By selecting a suitable threshold value, the image is binarized which contains only blood vessels and hemorrhages. There exist different kinds of lesions caused by diabetic retinopathy in a diabetic patient’s eye such as micro aneurysm, hard exudates, soft exudates, hemorrhage etc. Automated analysis of the fundus (retinal image) image is very much essential and will be of help to facilitate the clinical diagnosis. Research Paper Title: DIAGNOSIS OF DIABETIC RETINOPATHY USING DEEP NEURAL NETWORKS Abstract: To automate the diagnosis of Diabetic Retinopathy and provide appropriate suggestions to DR patients, they built a dataset of Diabetic Retinopathy fundus images that have been labeled by the proper treatment method that is required. Using this dataset, they trained deep convolutional neural network models to grade the severities of DR fundus images. They were achieved an accuracy of 88.72% for a four-degree classification(Normal, Moderate, Heavy, Severe) task in the experiments. They deployed models on a cloud computing platform and provided pilot DR diagnostic services for several hospitals; in the clinical evaluation, the system achieved a consistency rate of 91.8% with ophthalmologists, demonstrating the effectiveness of their work.
Research Paper Title:
IDENTIFICATION AND CLASSIFICATION OF MICROANEURYSMS FOR EARLY DETECTION OF DIABETIC RETINOPATHY Abstract: Diabetic retinopathy is a progressive eye disease which may cause blindness if not detected and treated in time. Eye abnormality in which the human retina is affected due to an increasing amount of insulin in blood. This paper proposes a three-stage system for early detection of microaneurysms using filter banks. This is very critical step for detecting diabetic retinopathy on early stages because they appear as the first sign of disease. Feature vector for each region depending upon certain properties, i.e. shape, color, intensity and statistics presents a classifier GMM, SVM. The proposed system is assessed using standard fundus image databases with the help of performance parameters, such as, sensitivity, specificity, accuracy, and the Receiver Operating Characteristics curves for statistical analysis. An evaluation methodology is expressed for corresponding problem domain, analyse the maturity for technology transfer from the research laboratories to the medical practices.
Research Paper Title:
A DEEP LEARNING ENSEMBLE APPROACH FOR DIABETIC RETINOPATHY DETECTION Abstract: Diabetic Retinopathy (DR) is an ophthalmic disease that damages retinal blood vessels. This paper deals with the causes & detection of DR that might lead to blindness. While explaining 5- different stages (normal, mild, moderate, severe and PDR (Proliferative Diabetic Retinopathy) the diagnosis of this fatel disease is also explained after the consultation of experts. Paper also shows how the manual processes can replaced by automated ones. Dataset is also publicaly available. Comparisons for different DCNN (Deep Covolutional Neural Networks) Models (Resnet50, Inceptionv3, Xception, Dense121, Dense169) with different accuracy levels to improve tendancey and experimental observations. Single automated model that proposed all the stages of DR unlike the current methods and performs better is explained.