SVM classification of microaneurysms with imbalanced dataset based on borderline-SMOTE and data cleaning techniques

Qingjie Wang; Jingmin Xin; Jiayi Wu; Nanning Zheng

doi:10.1117/12.2268519

17 March 2017 SVM classification of microaneurysms with imbalanced dataset based on borderline-SMOTE and data cleaning techniques

Qingjie Wang, Jingmin Xin, Jiayi Wu, Nanning Zheng

Author Affiliations +

Proceedings Volume 10341, Ninth International Conference on Machine Vision (ICMV 2016); 103411S (2017) https://doi.org/10.1117/12.2268519
Event: Ninth International Conference on Machine Vision, 2016, Nice, France

Abstract

Microaneurysms are the earliest clinic signs of diabetic retinopathy, and many algorithms were developed for the automatic classification of these specific pathology. However, the imbalanced class distribution of dataset usually causes the classification accuracy of true microaneurysms be low. Therefore, by combining the borderline synthetic minority over-sampling technique (BSMOTE) with the data cleaning techniques such as Tomek links and Wilson’s edited nearest neighbor rule (ENN) to resample the imbalanced dataset, we propose two new support vector machine (SVM) classification algorithms for the microaneurysms. The proposed BSMOTE-Tomek and BSMOTE-ENN algorithms consist of: 1) the adaptive synthesis of the minority samples in the neighborhood of the borderline, and 2) the remove of redundant training samples for improving the efficiency of data utilization. Moreover, the modified SVM classifier with probabilistic outputs is used to divide the microaneurysm candidates into two groups: true microaneurysms and false microaneurysms. The experiments with a public microaneurysms database shows that the proposed algorithms have better classification performance including the receiver operating characteristic (ROC) curve and the free-response receiver operating characteristic (FROC) curve.

Citation Download Citation

Qingjie Wang, Jingmin Xin, Jiayi Wu, and Nanning Zheng "SVM classification of microaneurysms with imbalanced dataset based on borderline-SMOTE and data cleaning techniques", Proc. SPIE 10341, Ninth International Conference on Machine Vision (ICMV 2016), 103411S (17 March 2017); https://doi.org/10.1117/12.2268519

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available