Elimination of Baseline Deformation from Infrared Spectra Prior to Protids Determination

We have recently showed by using Principal Component Analysis and Principal Component Regression on Mid-infrared spectra of biological samples that their protids concentrations could be determined with a good precision (CADET F. Spectrosc Lett 29(5):919-936, 1996). However the precision of the results can be improved; one way is to correct baseline deformations. Baseline is often assumed to be low-degreed polynomials. In this paper, the terms with weak n values were subtracted in order to improve the precision of the quantitative determination of protids from the Legendre polynomial functions that were obtained from the decomposition of the spectra. The filtering parameters range from n =0 to n= 8, by increments of 2. The mean of the difference between reference and predicted values are 0 before correction and -0.01 after correction, while standard deviation values are 0.12 and 0.11 before and after correction for n=6 or n=8.