期刊
COMPUTERIZED MEDICAL IMAGING AND GRAPHICS
卷 33, 期 6, 页码 415-422出版社
PERGAMON-ELSEVIER SCIENCE LTD
DOI: 10.1016/j.compmedimag.2009.03.007
关键词
Breast cancer; Image analysis; Mammographic automatic mass detection; False positive reduction; Textural information; Local binary patterns
资金
- Ministerio de Educacion y Ciencia of Spain [TIN2007-60553]
- UdG [IdIBGi-UdG]
- CIRIT
- CUR of DlUiE of Generalitat de Catalunya [2008SALUT00029]
During the last decade several algorithms have been proposed for automatic mass detection in mammographic images. However, almost all these methods suffer from a high number of false positives. In this paper we propose a new approach for tackling this false positive reduction problem. The key point of our proposal is the use of Local Binary Patterns (LBP) for representing the textural properties of the masses. We extend the basic LBP histogram descriptor into a spatially enhanced histogram which encodes both the local region appearance and the spatial structure of the masses. Support Vector Machines (SVM) are then used for classifying the true masses from the ones being actually normal parenchyma. Our approach is evaluated using 1792 ROIs extracted from the DDSM database. The experiments show that LBP are effective and efficient descriptors for mammographic masses. Moreover, the comparison with current methods illustrates that our proposal obtains a better performance. (C) 2009 Elsevier Ltd. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据