The number of unique cluster templates which could be generated was limited by the number of mastectomy samples available during the period of preparation for the stud y. However, it was preferable to use a large number of clusters in the study as this is expected to increase the statistical power. Therefore, to increase the numbers available the calcifications within each cluster were rearranged. When rearranging the calcifications the orientations of the calcifications were unchanged but the positions of their centroids were swapped. The rearranged clusters were inspected by a radiologist who confirmed that they looked sufficiently different in the image to the original clusters, and so could be treated as unique clusters. Including the original 33 clusters, this gave a total of 66 clusters. Forty-seven of the clusters were repeated twice within the image set; however, each repeated cluster was rotated and/or flipped horizontally or vertically in relation to the original cluster. The final number of clusters inserted into the clinical images was 113.