Py-MS data

Different applications regarding the evaluation of Py-MS data as shown for one single peak (univariate data) can be extended to the data matrix. One such application is the rejection of an extreme value in a set of replicates for the same specimen. For this purpose, a variable Hj should first be calculated  

The peaks predominant in glycogen will show positive, while those predominant in cellulose will show negative values. The smaller are the differences in the graph, the less difference is expected between the samples. 

More complicated techniques derived from multivariate data analysis are also utilized for data comparison, and these can be based on different types of measures such as  

A similarity index for two Py-MS results (see also Section 5.2), A and B, each with a maximum of k peaks, can be calculated based on the ratios of peak intensities for each m/Zi ion. Assuming that Ai and B, i =1, 2. . . k, are the pair peak intensities in the two samples, first are calculated the ratios  

Another possibility [69] (which was also used for the comparison of chromatograms) applies the formula  

---

The Py-MS data was obtained on the MS-25 spectrometer using a direct heating probe designed in this laboratory. The maceral samples were deposited as a slurry onto a fine platinum grid... 

Collections of Py-MS spectra were published [47], and also specific interpretation techniques were adapted to process Py-MS data, mainly developed for providing pertinent comparisons. Most techniques are oriented toward comparing multi-component fingerprint information. Therefore, the stability of the results (reproducibility) is an important quality that must be maintained when performing Py-MS work. It was shown [47] that variability of 1-3% in peak intensities can be noticed for replicates within 1 day of work and up to 10-11% in long term (one month). 

Py-MS data analysis with univariate statistical techniques. 

Multivariate data analysis has been developed as an independent field of statistics and has numerous practical and theoretical applications. This explains the existence of a variety of printed materials and of different computer programs available for multivariate data processing [71a], Only some aspects of multivariate data analysis with application to the processing of Py-MS data will be discussed here. 

The above described procedure can be combined with moving" points in a certain dimension to maintain the order of the distances. For example, if dij > dm but < dm, the coordinates Xij of a point j can be corrected (in dimension i ) to achieve dij < dki. This type of technique was extensively used for processing Py-MS data for the classification of microorganisms and bioorganic samples [76, 76b],... 

Figure 7.1.10 shows the score plot of the first two discriminant functions of the Curie point Py-MS data on several glucans and one mannan (five replicates of each sample). A good separation of each polysaccharide is obtained. 

The degree of methylation (DM) of pectin [60] can be estimated using a Py-MS technique. The application of principal component analysis and canonical variate analysis to the Py-MS data (see Section 5.5) showed a linear relationship between DM and the first canonical variate score of the data as shown in Figure 7.5.3. 

Fibers are another important source of forensic material. Both synthetic and natural fibers were analyzed by pyrolysis, the pyrograms or Py-MS data generating a fingerprint that can be diagnostic for each type of fiber [13]. 

The reason this method is so attractive to pyrolysis-mass spectroscopy (Py-MS) data is that it has been shown mathematically that a neural network consisting of only one hidden layer, with an arbitrary large number of nodes, can learn any arbitrary (and hence nonlinear) mapping to an arbitrary degree of accuracy. ANNs are also considered to be robust to noisy data, such as that which may be generated by Py-MS. 

Analysis was performed by desorbing the organics from the traps with a Curie-point pyrolyzer unit in series with a quadrupole mass spectrometer. The data produced were similar to Py-MS data, although, quite likely, thermal desorption was taking place rather than pyrolysis. The typical mass spectrum obtained from the contaminated areas was dominated by the major ions of PCE. Table 7.5 shows the various compounds that were identified in spectra obtained from the 25 samples spaced around the contaminated area. Table 7.6 shows the compounds identified by static trapping from a particular location and by purge-and-trap GC/MS analysis of water from an adjacent well. 

Unsupervised learning (that is, factor analysis and nonlinear mapping) showed that the Py-MS data contained enough distinguishing chemical information that digested and undigested pollens could be differentiated. Also, samples that were... 

Progress in the design of mass spectrometers and the availability of online computer systems has allowed the integration of Py-MS data acquisition with multivariate mathematical data reduction methods into a single analysis technique. Such an approach combines rapid analysis capability with expert system or pattern recognition based data evaluation (Figure 13). 

---