Baseline-corrected spectral regions

Figu re 6.5 Two views of a 3-D univariate chemical map plotting integrated absorbance over the spectral region 1275-1190cm , after baseline correction. Red indicates areas of highest absorption, blue indicates areas of lowest absorption. The view is looking toward the section 1 side of the sampled volume in (a) and towards the section 4 side in (b). 

Piece-wise MSC Piece-wise Multiplicative Signal Correction is a mathematical treatment of spectra used prior to regression to correct for baseline and offset shifts caused by e.g. difference in particle size or in colour between samples. While in ordinary MSC the varying offset and slope is assumed to be constant over the spectral range, piece-wise MSC uses windows for different spectral regions to adjust offset and slope locally. 

In order to automatically correct the baseline of the spectrum, switch to the Select Method page (Fig. 10.5) after loading the spectra. Here, you have to specify the correction method and the number of baseline points to be used. As an option you can exclude the spectral region that contain CO2 bands. In this case, data points between 2400 and 2275 cm and between 680 and 660 cm are not taken into account in the calculation. This feature is available for the Automatic option only. 

Linear Regression Baseline Fitting. This is a very simple approach to baseline correction in that it requires no effort to set up. In this method, a least squares regression line is fit to the responses in each spectral region selected for calibration. This line is then subtracted from the response values in the region before using the data to perform the calibration model calculations (Ref. 52). 

Unfortunately, this is not always the best approach, especially when the selected spectral regions are primarily large bands from the constituents of interest. It tends to work better when the entire spectrum is used or when the selected regions are very broad. In some cases, this method actually degrades the performance to the calibration models more than if no baseline correction was used at all. In general, this method should be used only in situations in which baseline aberrations are severe and a limited number of training sample spectra are available. 

An example of how normalisation affects the differentiability of spectra from different tissue structures in classification analysis is given in Figure 6.8. This illustration shows representative FT-IR microspectra from nine different structures of the human colon spectra of the submucosa (green), crypts (red), and of seven other pre-defined tissue structures (black). Panel A depicts the raw spectral data, while spectra in panel B have been offset corrected and vector-normalised using the spectral region of 1200-1770 cm In this simple case, characteristic spectral features of crypts (mucin features) and of the submucosa (features of collagen) are easily discernible in the pre-processed spectra, but not in the raw data. Furthermore, univariate analysis of the spectra in panel B would, in principle, allow for the introduction of certain thresholds suitable for the identification of both classes, namely at 1450 cm for submucosa and at 1080 and 1733 cm for the class crypts. The illustration demonstrates that normalisation and baseline correction are essential prerequisites for classification analysis. 

PCA was employed to classify the differences in the spectra for each local area spectrum of the spherulite. Consequently, 20 single spectra were extracted from 20 points of groups A-C, but for group D only five spectra were obtained as the amorphous area is limited (as shown in Figure 22.23c). All spectra in the region of 5500-3300 cm were subjected to a linear baseline correction and a second derivative pretreatment to highlight subtle differences in the spectral features among the spectra before the PCA calculations. 

Whether the transmittance of the baseline increases or decreases when a sample is inserted in the beam, it is often good practice to divide the transmittance spectrum by its maximum ordinate in a spectral region where the sample does not absorb before converting the spectrum to absorbance. (This operation is sometimes known as single-point baseline correction-, see Section 10.1.)... 

Many automatic baseline correction routines that may be applied without operator intervention are available. These routines may be applied by default for operations such as spectral searching. Automatic basehne functions typically use linear or polynomial baseline fits in regions of the spectmm where no peaks are detected. 

Few spectral regions were selected (figure not shown), arotmd 12000-8800 cm range, probably selected for background/baseline correction, 8700-7890 and 7600-6260 cm , where CH and OH overtones are present, respectively. 

As discussed earlier, second-derivative spectra can be used to minimize effects of baseline spectral fluctuations. Images prepared in the same manner as in Fig. 4.20 except using the negative of the second derivative of the butter-bacteria data are shown in Fig. 4.21. In this figure, the bacterial contamination region is correctly... 

---