The comparison between left and right breasts is based on the fact that both breasts have similar internal structure, as it is shown in Figure , where mammograms corresponding to left and right breasts of a woman are displayed as radiologists are used to. A thoroughly evaluated method for comparing mammographic images is known as bi-lateral subtraction. Both images are first aligned and subsequently subtracted. The alignment of both breasts is the critical component in this process. There are two main approaches to this:
The advantage of the former techniques over the latters is that the distortion of the internal structures is taken into account, thus registering correctly the breast interior. It should be noted that imaging conditions (breast compression and exposure, X-rays energy, etc.) are likely to differ for the left and right mammograms.
There are alternatives to this alignment and subtraction approach. Kok-Wiles et al. [] represent the bright zones of the breast as a tree-like-structure, which forms the basis for the comparison process. Stamatakis et al. [] developed a Multiple Image Comparison approach. This method starts with one pair of mammograms, then the system computes eight new pairs of images using a set of grey-level features. Finally each of those pairs is bilaterally compared with each other and the resulting difference images are recombined into a final result.