The ability to rapidly and accurately discriminate between healthy and malignant tissue offers surgeons a tool for in vivo analysis that would potentially reduce operating time, facilitate quicker recovery, and improve patient outcomes. To this end, we investigate discrimination between diseased tissue and adjacent healthy controls from patients with head and neck cancer using near-infrared Raman spectroscopy. Our results indicate previously unreported peaks in the Raman spectra that lie outside the conventional "fingerprint" region (400 cm-1-1800 cm -1) played an important role in our analysis and in discriminating between the tissue classes. Preliminary multivariate statistical analyses of the Raman spectra indicate that discrimination between diseased and healthy tissue is possible based on these peaks.
Keywords: Raman spectroscopy; head and neck cancer; multivariate statistics; otolaryngology; principal component analysis; tissue diagnostics.