An Evaluation of Stereoscopic Digital Mammography for Earlier Detection of Breast Cancer and Reduced Rate of Recall

The clinical trial, begun in December, 2004, is being conducted at Emory University. Over the next three years, we will enroll and image about 2000 women who are at elevated risk for development of breast cancer (personal or family history of breast cancer) as they come in for their annual screening examinations. Each participating patient will be imaged with both systems, and the stereo and standard (non-stereo) images will be read independently by different mammographers. A stereoscopic digital mammogram consists of two x-ray images of the breast acquired from slightly different points of view on a digital mammography unit. The x-ray source is rotated 10-degrees between the two exposures while the position of the breast remains fixed in the compression device. The stereo pair of mammograms is viewed on a new third-generation, high-resolution stereo display workstation that was developed in the first part of this project. The mammographer views the stereo display wearing lightweight polarized glasses, enabling him or her to see in depth the internal structure within the breast. As a result, a subtle lesion that may be obscured by superimposed normal tissue in a standard 2D image, now becomes visible as the overlying and underlying normal tissue is separated in depth. Conversely, layers of tissue that may falsely resemble a lesion in a standard 2D image due to chance superimposition, are seen in the stereo mammogram to lie at different depths and, thus, will not be mistaken as a lesion. We anticipate that with stereo mammography the mammographers will detect subtle lesions in the stereo images that are missed in the non-stereo images. We also expect that there will be fewer false positive detections with the stereo images compared to the standard images. Finally, we also expect that the mammographers will be more confident in their judgments of what they see in the stereo images, resulting in a reduced rate of recall of patients for further work-up.