Optical spectroscopy of murine breast tumors to determine metastatic potential

We used Diffuse Reflectance spectroscopy (DRS) and Raman spectroscopy to investigate functional differences between murine breast tumor xenografts of varying metastatic potential. Once tumors volume reached 200mm3, in-vivo DRS and ex-vivo Raman spectroscopy were performed. Spectra and optical properties were used to train and test a leave-one-out random forest classifier. Significant differences between metastatic and non-metastatic tumors were observed. The study shows that random forest classifiers coupled with optical spectroscopy provides consistent predictions of metastatic phenotype and metastatic abilities from the primary tumor that can be translated into clinic.