E. Badeka, Cristina I. Papadopoulou, G. Papakostas
{"title":"Evaluation of LBP Variants in Retinal Blood Vessels Segmentation Using Machine Learning","authors":"E. Badeka, Cristina I. Papadopoulou, G. Papakostas","doi":"10.1109/ISCV49265.2020.9204176","DOIUrl":null,"url":null,"abstract":"In this paper, the problem of retinal image segmentation is examined. The segmentation task is handled as a binary classification problem and it is solved by applying handcrafted texture features and traditional machine learning models. In this context, the paper studies the segmentation performance of the Local Binary Pattern (LBP) texture descriptor and nine of its variants. For the needs of the evaluation, a segmentation methodology and a corresponding experimental protocol is proposed, which are applied along with a benchmark retinal image dataset. The simulation results revealed that not all the LBP variants are appropriate for accurate extraction of the retinal arteries/veins morphology. The derived segmentation accuracy varies between 78%-91% (Support Vector Machine- SVM), S6%-92% (Decision Tree-DT), and 84%-92% (k-Nearest Neighbors - k-NN), with the Extended LBP (E-LBP) being the most informative descriptor.","PeriodicalId":313743,"journal":{"name":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"7","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2020 International Conference on Intelligent Systems and Computer Vision (ISCV)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ISCV49265.2020.9204176","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 7
Abstract
In this paper, the problem of retinal image segmentation is examined. The segmentation task is handled as a binary classification problem and it is solved by applying handcrafted texture features and traditional machine learning models. In this context, the paper studies the segmentation performance of the Local Binary Pattern (LBP) texture descriptor and nine of its variants. For the needs of the evaluation, a segmentation methodology and a corresponding experimental protocol is proposed, which are applied along with a benchmark retinal image dataset. The simulation results revealed that not all the LBP variants are appropriate for accurate extraction of the retinal arteries/veins morphology. The derived segmentation accuracy varies between 78%-91% (Support Vector Machine- SVM), S6%-92% (Decision Tree-DT), and 84%-92% (k-Nearest Neighbors - k-NN), with the Extended LBP (E-LBP) being the most informative descriptor.