ICA-guided delineation of breast cancer pathology.
In: Proceedings of the 9th IEEE International Symposium on Biomedical Imaging (From Nano to Macro ISBI 2012, Barcelona, 2-5 May 2012). Piscataway, NJ: IEEE, 2012. 1611-1614
A surgeon-guided independent component analysis from optical reflectance measurements is proposed for breast tumor delineation. Independent Component Analysis is first applied to extract the most relevant features from local measures of broadband reflectance and then a tumor probability indicator is obtained and provided utilizing surgeon assistance to resolve the inherent ambiguities in the independent component calculation. A set of 29 breast tissue samples have been diagnosed achieving a sensitivity of 90.57%, and specificity of 93.98%.