Breast tumor segmentation in high resolution x-ray phase contrast analyzer based computed tomography
- Institute for Clinical Radiology, Ludwig-Maximilians-University Hospital Munich, 81377 Munich (Germany)
- Department of Physics, Harvard University, Cambridge, Massachusetts 02138 (United States)
- Department of Physics, Ludwig-Maximilians University, Garching 85748, Germany and Institute for Clinical Radiology, Ludwig-Maximilians-University Hospital Munich, 81377 Munich (Germany)
- European Synchrotron Radiation Facility (ESRF), Grenoble 380000 (France)
Purpose: Phase contrast computed tomography has emerged as an imaging method, which is able to outperform present day clinical mammography in breast tumor visualization while maintaining an equivalent average dose. To this day, no segmentation technique takes into account the specificity of the phase contrast signal. In this study, the authors propose a new mathematical framework for human-guided breast tumor segmentation. This method has been applied to high-resolution images of excised human organs, each of several gigabytes. Methods: The authors present a segmentation procedure based on the viscous watershed transform and demonstrate the efficacy of this method on analyzer based phase contrast images. The segmentation of tumors inside two full human breasts is then shown as an example of this procedure’s possible applications. Results: A correct and precise identification of the tumor boundaries was obtained and confirmed by manual contouring performed independently by four experienced radiologists. Conclusions: The authors demonstrate that applying the watershed viscous transform allows them to perform the segmentation of tumors in high-resolution x-ray analyzer based phase contrast breast computed tomography images. Combining the additional information provided by the segmentation procedure with the already high definition of morphological details and tissue boundaries offered by phase contrast imaging techniques, will represent a valuable multistep procedure to be used in future medical diagnostic applications.
- OSTI ID:
- 22317951
- Journal Information:
- Medical Physics, Vol. 41, Issue 11; Other Information: (c) 2014 American Association of Physicists in Medicine; Country of input: International Atomic Energy Agency (IAEA); ISSN 0094-2405
- Country of Publication:
- United States
- Language:
- English
Similar Records
Estimation of breast percent density in raw and processed full field digital mammography images via adaptive fuzzy c-means clustering and support vector machine segmentation
A new background distribution-based active contour model for three-dimensional lesion segmentation in breast DCE-MRI