A preliminary study on the importance of normalization methods in Infrared Micro-Spectroscopy for biomedical applications

Fourier transformed infrared (FT-IR) micro spectroscopy has become a reliable, non-destructive and automatic tool to analyze the chemical differences in biological samples associated with healthy or diseased states for diagnostic or biomedical research purpose. To extract useful information from the huge number of spectra usually recorded, statistical multivariate analysis is applied and careful data pre-treatment to reduce or eliminate possible sources of error is necessary. In this regard, a quantitative criterion to evaluate the performance of different methods applied in the most sensitive pre-processing step, normalization, is proposed here. This could constitute a promising approach for validating the significance of chemical variations in complex samples of biomedical interest.