Detection of spectral differences between normal and cancerous oral tissue using near-infrared spectroscopy

Abstract

A system was developed using a Fourier Transform spectrometer configured for the near-infrared (2.0- 2.5 pm) range to investigate spectral differences between cancerous and normal oral tissue. Cancer was induced in the left or right cheek of hamsters. The animals were euthanized and cancerous as well as normal tissue was excised and analyzed in vitro. A total of 20 spectra were recorded 10 from cancerous tissue and 10 from normal tissue. Polynomial subtraction was used for preprocessing the raw spectral data. The preprocessed data were entered into a principal component analysis (PCA) algorithm which showed the ability to discriminate the tissue types across animals using this wavelength range.