Development of Digital Steroscopic Imaging Technique in Mammography
Abstract
The goal of this research is to develop stereoscopic techniques for mammographic imaging and to investigate the feasibility of using stereomammography to improve the sensitivity of mammography for breast cancer detection, especially in dense breasts. During this year, we conducted further studies to investigate the effects of imaging techniques on depth perception and depth measurement in stereomammography. A GE Senographe 20OOD full field digital mammography system was used for stereoscopic imaging. We developed a stereoscopic imaging technique in which the phantom was shifted instead of the focal spot for acquisition of the left-eye and right-eye images. In a preliminary observer study, we found that larger stereo angles and zooming facilitate depth measurement with a virtual cursor. In a separate observer experiment, we evaluated the effects of magnification, contact, and zooming on depth discrimination. We found that the accuracy of depth discrimination increased with increasing fibril depth separation and with increasing x-ray exposure. Zooming the contact stereo images by 2X did not improve the accuracy. Depth discrimination was superior with stereo images acquired using geometric magnification in comparison with images acquired using a contact technique. These studies indicate that stereoscopic imaging may improve the perception of image details and depth separation, and thus may be useful for differentiating overlapping tissues from masses and identifying 3D spatial distribution of microcalcifications in mammography. Based on the information from these observer studies, it will be possible to design effective imaging techniques for stereomammography. The improvement in perception of the details of mammographic features and the additional size and depth information are expected to improve diagnostic accuracy of mammographic abnormalities.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2001
- Accession Number
- ADA392070
Entities
People
- Heang P. Chan
Organizations
- University of Michigan