Title: Detection and classification of calcifications in digital mammograms by multi-scale and multi-position

Authors: Z. Lifeng; Chen Ying; Zhang Lu

Addresses: Department of Biomedical Engineering, School of Medicine, Shanghai Jiao Tong University, Chongqing Rd. 227, Shanghai, 200025, China. ' Department of Biomedical Engineering, School of Medicine, Shanghai Jiao Tong University, Chongqing Rd. 227, Shanghai, 200025, China. ' Department of Biomedical Engineering, School of Medicine, Shanghai Jiao Tong University, Chongqing Rd. 227, Shanghai, 200025, China

Abstract: In this paper, we present a novel multi-scale and multi-position classification method. In this method, mammograms are divided into sub-images from which the image features can be extracted. The proposed method uses a cascaded support vector machine (SVM) classifier to detect and classify calcifications. Using proposed method, we can robotically detect and classify calcifications into different types. This method appropriately meets the basic requirements in CAD-based mammographic screening. The experiments data of the method from digital database for screening mammography (DDSM) show that the detection rate of calcifications can reach up to 98% with 23.75% FP.

Keywords: multi-scale classification; multi-position classification; cascaded SVM; support vector machines; breast cancer; cancer detection; feature extraction; pattern recognition; mammograms; breast screening; calcification; CAD.

DOI: 10.1504/IJCSE.2012.048242

International Journal of Computational Science and Engineering, 2012 Vol.7 No.3, pp.224 - 231

Received: 03 Nov 2011
Accepted: 04 Feb 2012

Published online: 22 Sep 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article