SEGMENTATION OF MULTI CLASS RETINAL LESIONS FROM FUNDUS IMAGES

2022-2-9
Çontar, Elif K
Diabetic retinopathy is a leading cause of preventable blindness among adults. Detection of diabetic retinopathy-related retinal lesions is essential for automatic detection of DR. There are different kinds of lesions related to the disease, namely microaneurysm, hemorrhage, hard exudate, and soft exudate. Each lesion has different characteristics: color, size, and shape. In the literature, the detection of retinal lesions has been examined as a localization or segmentation problem. Besides traditional image processing methods, machine learning-based and neural network-based methods have been proposed widely in the last years. Most of the works focused on detecting only one type of lesion. These methods can not be transferred to detect another kind of lesion because of the different characteristics of the lesions. Additionally, segmentation of retinal lesion task is an imbalanced classification problem. Task includes both foreground-background imbalance and imbalance between positive classes. In this study, we developed a new instance-based intersection over union(IB_IoU) objective function to segment multi-class retinal lesions from fundus images. The loss has targeted the following two problems. Firstly, it aims to solve the imbalance problem by averaging intersection over union(IoU) scores across the classes. Secondly, IoU score is calculated separately for every instance with a closed contoured shape. The aim is to improve the detection performance of lesions with small pixel areas. The connected component analysis is applied to find instances on a union of prediction and ground truth labels. The results show that the proposed algorithm is comparable to state-of-the-art methods focused on detecting single lesions. Additionally, the proposed loss function has improved detection performance of microaneurysm and exudate lesions over other loss functions used in multi-class retinal lesion segmentation.

Suggestions

Curcumin ameliorates impaired insulin/IGF signalling and memory deficit in a streptozotocin-treated rat model
IŞIK, AHMET TURAN; Celik, Turgay; Ulusoy, Gokhan; Ongoru, Onder; Elibol, Birsen; Doruk, Huseyin; Bozoglu, Ergun; Kayir, Hakan; Mas, Mehmet Refik; Akman, Serif (Springer Science and Business Media LLC, 2009-03-01)
Increased serum insulin levels and reduced peripheral insulin activities seen in insulin resistance syndrome are associated with age-dependent cognitive impairment and Sporadic Alzheimer's Disease (SAD), suggesting a disturbance in the insulin signalling system in the brain and possibly being one of the causes of dementia. Therefore, the streptozotocin (STZ)-induced animal may be an appropriate model for the investigation of SAD and related dementia. This study was designed to investigate the beneficial eff...
Early alterations in myocardia and vessels of the diabetic rat heart: an FTIR microspectroscopic study
Toyran, Neslihan; Lasch, Peter; Naumann, Dieter; Turan, Belma; Severcan, Feride (2006-08-01)
Diabetes mellitus is associated with a high incidence and poor prognosis of cardiovascular disease. The aim of the present study was to examine the effect of relatively short-term (5 weeks) Type I diabetes on the left ventricle, the right ventricle and the vessel (vein) on the left ventricle of the myocardium, at molecular level by FTIR (Fourier-transform infrared) microspectroscopy. The rats were categorized into two groups: control group (for the left ventricle myocardium, n = 8; for the right ventricle m...
Gene expressions of Mn-SOD and GPx-1 in streptozotocin-induced diabetes: effect of antioxidants
Sadi, Goekhan; Güray, Nülüfer Tülün (Springer Science and Business Media LLC, 2009-07-01)
Increased oxidative stress and impaired antioxidant defense mechanisms are believed to be the important factors contributing to the pathogenesis and progression of diabetes mellitus. In this study, we have reported the effects of the streptozotocin-induced diabetes on the gene expression and the activities of two antioxidant enzymes, manganese superoxide dismutase (MnSOD) and glutathione peroxidase (GPx). We also studied the effects of two antioxidants, vitamin C and DL-alpha-lipoic acid (LA), on the system...
Nanometer-Thick Mn:NiO and Co:NiO Films for High Performance Nonenzymatic Biosensors
Tunca, Sensu; Tutel, Yusuf; Koylan, Serkan; Ünalan, Hüsnü Emrah (2021-01-01)
People with diabetes require regular blood sugar level monitoring, using commercial enzyme-based biosensors. There is a considerable need to develop biosensors with nonenzymatic electrodes to eliminate the drawbacks of enzymes. Nanostructured nickel oxide (NiO) thin films are highly promising materials for the development of nonenzymatic glucose and hydrogen peroxide (H2O2) biosensors. Although the biosensor performance can be easily attained with nonenzymatic electrodes, their commercialization still requi...
Prediction of insulin resistance by statistical tool MARS
Örsçelik, Simge Gökçe; Weber, Gerhard Wilhelm; Osterhoff, Martin; Department of Bioinformatics (2014)
Recently, following the rise in obesity prevalence, the incidence of type 2 diabetes rose remarkably. Diabetes is a serious disorder, accompanied by increased risk of developing heart disease, kidney failure, and new cases of blindness. Dietary habits are strongly related to type 2 diabetes. We sought to observe how dietary protein and glycemic index patterns, weight change and/or other predictors we selected relate to insulin resistance change. First, we applied multiple linear regression, and then statist...
Citation Formats
E. K. Çontar, “SEGMENTATION OF MULTI CLASS RETINAL LESIONS FROM FUNDUS IMAGES,” M.S. - Master of Science, Middle East Technical University, 2022.