ABSTRACT
In the current standard mammography approach to breast cancer detection, the radiologist reads two orthogonal 2-D x-ray images of each breast, a top down, craniocaudal (CC) view and a sideways mediolateral oblique (MLO) view. In each view, all of the three-dimensional tissue and structure contained within the volume of the breast is superimposed and collapsed onto a 2-D projection image. Detecting cancer in this way is a daunting and o¬en extremely di–cult task because of the complexity in the many ways cancer can present and the many ways in which it can be hidden (see Chapter 7). Most daunting of all is that detection accuracy depends heavily on the radiologist’s ability to discern how the various patches of dense tissue or particles of calciœcation visible in the separate 2-D images are related across the separate images and, ultimately, how they are positioned in the breast volume and locally shaped or organized in depth. What is daunting is that with the present 2-D technique all of that abstraction of volumetric information requires cognitive merging of information collected separately from the two images. ™e critical contribution of stereo imaging is that it enables the reader to acquire that volumetric information directly, and gather it much more precisely-immediately perceiving the structure within the breast volume in depth.
