Biomedical Engineering Reference
In-Depth Information
8
BREAST STRAIN IMAGING:
A CAD FRAMEWORK
Ruey-Feng Chang
Department of Computer Science and Information Engineering
Graduate Institute of Biomedical Electronics and Bioinformatics
Graduate Institute of Networking and Multimedia
National Taiwan University, Taipei, Taiwan
Chii-Jen Chen, Chia-Ling Tsai,
and Wei-Liang Chen
Department of Computer Science and Information Engineering
National Chung Cheng University, Chiayi, Taiwan
Jasjit S. Suri
Biomedical Research Institute, Idaho State University,
Pocatello, Idaho, USA
In 2D ultrasound Computer-Aided Diagnosis (CAD), the main emphasis is extraction of
tumor boundaries and its classification into benign and malignant types. This provides a
direct tool for breast radiologists and can even prevent breast biopsies, thereby reducing the
number of false positives. The prerequisite for accurate breast boundary estimation in 2D
breast ultrasound images is accurate segmentation of breast tumors and shape modeling.
But this is a challenging task, because there is no set pattern of progression of tumors in the
spatiotemporal domain. This chapter adapts a methodology based on geometric deformable
models such as the level set, which has the ability to extract the topology of shapes of breast
tumors. Using this framework, we extract several features of breast tumors and feed this
set of information into a vector machine-based classifier for classification of breast disease.
Our system demonstrates accuracy, sensitivity, specificity, PPV, and NPV values of 87, 85,
88, 82, and 89%, respectively.
