Comparison of Classifier Strength for Detection of Retinal Hemorrhages
Sreeja K.A1, S.S. Kumar2
1Sreeja K.A., Associate Professor, Department of Electronics and Communication Engineering, SCMS School of Engineering & Technology, Karukutty, Ernakulam, Kerala, India.
2S.S. Kumar, Associate Professor, Department of Electronics and Instrumentation Engineering, Noorul Islam University, Kanyakumari, Tamil Nadu, India.
Manuscript received on 08 April 2019 | Revised Manuscript received on 15 April 2019 | Manuscript Published on 24 May 2019 | PP: 688-693 | Volume-8 Issue-6S3 April 2019 | Retrieval Number: F11370486S319/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Diabetes Mellitus(DM) which is the root cause of diabetic retinopathy(DR) diseases such as occlusion, microaneurysms, retinal hemorrhage, etc. Hemorrhage is considered the most dangerous among these, as it can accelerate the occurrence of vision loss. Hence, the severity of hemorrhages is analyzed in most of the recent studies of diabetic retinopathy detection. This paper focusses on the best classification approach by comparing different machine learning approach using supervised classifiers. Fundus image collected from publically available database are preprocessed and enhanced. Using splat based method, ground truth is established with the help of a retinal expert. Supervised classifiers are trained from the GLCM features extracted from the segmented images and validated on clinical images. The experimental results were verified by the Area Under Curve(AUC) for the three classifiers that were trained and results are verified and tabulated.
Keywords: Hence, the Severity of Hemorrhages is Analyzed in Most of the Recent Studies of Diabetic Retinopathy Detection
Scope of the Article: Communication