AIRCC's International Journal of Computer Science and Information Technology

Ayman Abu Baker ¹

Affiliations
1 Department of Electrical and Computer Engineering, Applied Science Private University, Amman, JO

Abstract

This paper describes the ongoing efforts by the author to provide efficient and accurate classification for mass lesions in mammogram images. A study of the characteristics of true masses compared to the falsely detected masses is carried out using wavelet decomposition transform combining with support vector machine (SVM). In this approach, four main wavelet features are extracted from different regions of interest in order to distinguish between TP and FP detected regions. A study of detecting regions of interest, extracting the wavelet features and choosing the optimal learning parameters for support vector machine are also presented in this paper. The combined between the wavelet features and SVM presented here can successfully reduces the FP ratio to 0.05 clusters/image, with accurate TP ratio 94%.

Keywords

Mammogram, Mass Lesions, Wavelet Transform, Support Vector Machine.

Full Text