A novel shape feature to classify microcalcifications

WCU Author/Contributor (non-WCU co-authors, if there are any, appear on document)
Yiming Ma (Creator)
Institution
Western Carolina University (WCU )
Web Site: http://library.wcu.edu/
Advisor
Peter Tay

Abstract: Clinical evident shows that the shape of mammographic calcification is an indicator of the pathology. Microcalcifications (MC) with rough shape are early signs of malignant breast cancer. This thesis proposed a shape metric to help radiologist in classifying regions of interest. Region growing and gradient vector flow algorithm are used to obtain the contour of MC to calculate the normalized distance signature. A three level wavelet decomposition with a Daubechies eight tap wavelet is used to provide a bandpass function and extract the desired shape feature of the MC. A comparison with previously used shape features such as compactness, moment, Fourier descriptors is provided. 58 malignant and 125 benign cases, totaling 368 individual MC, are tested by the proposed method and previously used shape features.

Additional Information

Publication
Thesis
Language: English
Date: 2010
Keywords
cancer, shape measurement
Subjects
Breast -- Cancer -- Diagnosis
Breast -- Radiography
Breast -- Imaging

Email this document to