117828224

Chromatographic Fingerprints ofTwenty Salvia Species 521

lyptus oil, and juniper o ii (Apotheca Pacis, Rybnik, Poland) as the sets of the volatile standards [33].

Results and Discussion

In our study, two sets of herbal fingerprints of the Salina species were proc-essed and compared using chemometric methods. Chemometrics has many approaches that can be used to expłore and to evaluate differences among chromatographic fingerprints. Irrespective of the method used, proper sig-nal preprocessing is crucial for the further comparative and exploratory analysis. Preprocessing of instrumental signals is performed to meet three objectives:

(i)    to enhance signal quality;

(ii)    to make the signal analysis possible; and

(iii)    to enhance interprętation of the finał results.

These objectives are usually accomplished in four consecutive steps of signal preprocessing and in the next sections they are discussed in detail.

Step 1: Preliminary Preparation of Chromatographic

Fingerprints

The first step of signal preprocessing can be regarded as a preliminarily step. At this stage, individual fingerprints are prepared to make further preprocessing possible. For example, adjustment of the length of the chro-matograms and/or removal of uninformative parts can prove necessary; this was necessary for the herbal fingerprints studied in this work.

With the analyzed HPLC fingerprints, different acquisition rates were used from one sample to another. Ln that way, peaks were better shaped but the time axis of the chromatograms is not uniformiy sampled. As a conse-quence, within each time interval the number of the data points is different. The length of the analyzed HPLC chromatograms varied between 4380 and 10 859 sampling points. To organize these data into a matrix, it is obligatory to record all chromatograms within the same time interval and sample their time axis at the same points. To take uneven sampling of the time axis into account, signal intensities measured at the points of the original time axis can be linearly interpolated to the points of a new time axis that are uniformiy distributed over the entire time rangę; this was done for the HPLC fingerprints studied.