International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 10 Issue: 05 | May 2023
p-ISSN: 2395-0072
www.irjet.net
An automated severity classification model for diabetic retinopathy M Sai Praneeth,V Sujay Keertan T.Samved, Dr.V SOWMYA DEVI Assistant Professor, Department of Computer Science and Engineering, SNIST, Hyderabad-501301, India
1,2,3 B. Tech Scholars, Department of Computer Science and Engineering, SNIST, Hyderabad-501301, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Diabetic Retinopathy (DR) - a complication
Eye injury, but there are only 200,000 ophthalmologists in the globe [3]. Grading inconsistencies, a serious shortage of ophthalmologists in the field, as well as the problematic lab procedure, continue to be barriers to the early diagnosis of diabetic retinopathy. To lessen the heavy burden on healthcare systems, retinopathy tests should be automated. This has spurred major attempts to improve computer-aided medical diagnostics (CAMD) systems.
developed due to heightened blood glucose levels- is deemed one of the most sight-threatening diseases. Unfortunately, an ophthalmologist, a process that can be considered incorrect and time consuming, manually acquires the DR screening. In view of the huge increase in diabetic patients over recent years, automated diagnostic tests for diabetes have also become an increasingly important research topic. Additionally, Convolutional Neural Networks (CNN) have proven themselves to be state-of-the-art for DR stage diagnosis in recent times. This study offers a fresh, automatically learning-based method for determining severity from a single Colour Fundus Photograph (CFP). The suggested method builds a visual embedding using DenseNet169's encoder.. Convolutional Block Attention Module (CBAM) is also added on top of the encoder to boost its ability to discriminate. The model is then trained using the Kaggle Asia Pacific TeleOphthalmology Society (APTOS) dataset using cross-entropy loss. In comparison to state-of-the-art performance on the binary classification test, we achieved (97% accuracy, 97% sensitivity, 98.3% specificity, and 0.9455, Quadratic Weighted Kappa score (QWK)). Additionally, for severity grading, Our network demonstrated high proficiency (82% accuracy - 0.888 (QWK)). The suggested approach makes a substantial contribution by accurately classifying the degree of diabetic retinopathy severity while requiring less time and spatial complexity, making it a promising contender for autonomous diagnosis.
DR grading methods can be divided into two groups: separating diabetic retinas from healthy ones (binary classification problem) and determining the severity of afflicted retinas from class 0 (healthy) to class 4 proliferative DR (PDR) (multi-class classification task). Algorithms used for traditional machine learning (ML) are examples of artificial intelligence (AI) methods that gain knowledge by being exposed to data. Gradient Boosted Trees (GBT) [6] as a classification model and Artificial Flora Algorithm (AFA) [5] for feature selection. Furthermore, by using the feature engineering method and Support Vector Machines (SVM) as a classifier for DR detection, Gharaibeh et al. took use of this in [7] and [8]. Despite their effectiveness, ML algorithms require individualised experience and subject expertise to find the most informative representation. Through the representation of the universe as a layered hierarchy of concepts, each concept in Deep Learning (DL) has established a foothold in a variety of domains [10].
Key Words: Diabetic retinopathy, convolutional neural networks (CNN), attention mechanism, deep learning.
Utilising the capabilities of convolutional neural networks in the medical field has led to the development of more durable treatments, particularly in the DR sector. The efficacy of such a method for retinal vascular segmentation was shown in [16] and [17]. Similarly, Zhao et al. [18] were able to create fundus images using generative advertisement networks (GANs). For the early detection of Micro-aneurysms (MA), Dai et al. [19] used a multi-sieving convolutional neural network and picture to text mapping. [20] assessed the effectiveness of VGG16, VGG19, and InceptionV3 as three well-known CNN architectures. using transfer learning and adjusting for binary and multi-class classification [21, 22]. In order to categorise fundus images into two grades, Zeng et al. [23] proposed Siamese-like architecture [24] trained with transfer learning.
1.INTRODUCTION Hyperglycemia, a symptom of the chronic metabolic disorder diabetes mellitus, has a long-term negative impact on the body's blood vessels on both a micro and macro level. The World Health Organisation (WHO) estimates that there were 422 million diabetics worldwide in 2014, and that figure is expected to rise to 700 million by 2045 [1], [2]. diabetes retinopathy, a progressive anomaly exposed and recognised through ocular pathologies, which results in blocking and bleeding of the retinal capillaries, is one of the long-term diabetes micro-vascular consequences. Fortunately, vision damage can be avoided by early identification. Without routine inspection, it could cause irreparable harm. According to the International Diabetes Federation (IDF), there are 93 million diabetics worldwide.
© 2023, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 1769