Volume 10 Issue 3 - May 2018

  • 1. a comparative analysis of different techniques for diabetic retinopathy detection using fundus images

    Authors : D. Ashok Kumar, A.sankari

    Pages : 25-34

    DOI : http://dx.doi.org/10.21172/1.103.05

    Keywords : Diabetic Retinopathy, Segmentation, Feature Extraction, Classification.

    Abstract :

    In recent decays, diabetic retinopathy is an important eye disorder that may cause low vision if its diagnosis in late. Different feature extraction and classification methods have been studies in literature survey for the purpose of improving diabetic retinopathies accuracy in the screening test. In this paper, three hybrid types of diabetic detection approaches are analyzed namely 1) Hybrid MinIMas With Sparse Principal Component Analysis(HM: SPCA), 2).Hybrid Morphological-based Scanning Window Analysis and Hit-or-Miss Transformation (HMSWA-HMT),3.Low rank based dictionary based learning (HMSWA-HMT). To this end, in this comparative analysis of different algorithm is performed to select most appropriate method for retinopathy detection with various hybrid segmentation, feature extraction and classification method. Each technique is compared with evaluation metric of accuracy, sensitivity and specificity. Simulation results show that performance of individual procedure and can be used to decide the factor in algorithm selection for future research.

    Citing this Journal Article :

    D. Ashok Kumar, A.sankari , " a comparative analysis of different techniques for diabetic retinopathy detection using fundus images ", Volume 10 Issue 3 - May 2018, 25-34