Multiple label-free detection of breast cancer biomarkers using Raman spectroscopy

We have investigated a multiple label-free detection method based on Raman spectroscopy and multivariate curve resolution (MCR) analysis to classify breast cancer. Twenty breast tissues collected from five participants during breast surgery were used as biological samples. Ten samples were from malignant tumor mass (cancer core area) and the others were from the safety margin outside of the tumor mass (two sample groups). For each breast tissue sample, twenty Raman spectra were collected using a fiber-optics Raman system consisting of a fiber-optic Raman probe, a low dark current deep-depletion CCD connected to a Czerny-Turner spectrograph and 785-nm laser source. Using MCR analysis iteratively optimized by an alternative least squares (ALS) algorithm, biomarker-dominated spectral data can be obtained from the preprocessed Raman spectra. This allows a more accurate classification between the two sample groups (normal and cancer). We expect that the proposed method based on biomarker analysis using MCR-ALS will more accurately classify breast cancer.