Purpose. A characterization of the geometrical or morphological aspect of m
icrocalcifications is not sufficient to confirm a diagnosis of cancer. On t
he other hand, it seems that the shape of the cluster is a pertinent and he
lpful criterion for diagnosis. The ai m of our study is to work out a meth
od for 3D reconstructions of clusters of microcalcifications.
Materials and methods. After having tested different algorithms for extract
ing microcalcifications, we have developed a method of 3D reconstructions o
f the shape of the cluster from three radiographs. The technique, available
using a standard mammography unit, consists of three principal steps:
segmentation of three images acquired under different view-points,
matching of microcalcifications,
reconstruction of microcalcifications positions in 3D space.
Results. The technique has been tested with two simple tridimensional phant
oms and has given results with sufficient precision for possible clinical u
se.
Conclusion. We plan now to apply the mettled in vivo with adaptation on a d
edicated mammographic stereotactic unit with digital acquisition.