Studies on PCA for Breast Tissue Segmentation

Breast conserving surgery is mainly limited by the accuracy of tumor margin delineations. One over five is the estimated risk to undergo a second procedure following the failure of the first breast cancer removal attempt. In order to lessen this ratio, several teams take part in a general effort to develop a THz-based surgery room tissue type recognizer. However, at this time, no robust definition of a numerical tissue segmentation has been provided by the community. In an attempt to draw the basis for such a numerical tissue boarder description, we report here the investigation led on freshly excised breast tissues with Principal Component Analysis (PCA) to recognize tissue types and their associated margins.