A lesion shape and margin characterization method in dynamic contrast enhanced magnetic resonance imaging of breast

Author

Liang, Xi ; Ramamohanarao, Kotagiri ; Frazer, Helen ; Yang, Qing

fYear

2012

fDate

2-5 May 2012

Firstpage

1783

Lastpage

1786

Abstract

This study presents a shape and margin characterization method of breast mass lesions in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). The overlap between a mass lesion and its minimum volume enclosing ellipsoid (MVEE) is used to capture the overall shape of a lesion. Various statistical measurements on distance from the lesion surface to its MVEE surface are computed to characterize margin (boundary) features of a lesion. In the evaluation study using 28 benign and 20 malignant manually segmented lesions, MVEE-based features are demonstrated to be significant (p <; 0.05). Our study shows that MVEE based features provide higher sensitivity and specificity in SVM and ANN classifications of benign and malignant lesions.

Keywords

biological organs; biomedical MRI; gynaecology; image classification; image segmentation; medical image processing; neural nets; statistical analysis; support vector machines; ANN classification; DCE-MRI; MVEE surface; MVEE-based feature; SVM classification; benign lesions; benign manually segmented lesions; breast mass lesions; dynamic contrast enhanced magnetic resonance imaging; lesion shape characterization method; lesion surface; malignant lesions; malignant manually segmented lesions; margin characterization method; minimum volume enclosing ellipsoid; statistical measurements; Artificial neural networks; Breast; Cancer; Ellipsoids; Lesions; Shape; Support vector machines; DCE-MRI; Mass lesion; Shape; classification;

fLanguage

English

Publisher

ieee

Conference_Titel

Biomedical Imaging (ISBI), 2012 9th IEEE International Symposium on

Conference_Location

Barcelona

ISSN

1945-7928

Print_ISBN

978-1-4577-1857-1

Type

conf

DOI

10.1109/ISBI.2012.6235927

Filename

6235927