PROCESSING ION MOBILITY SPECTROMETRY DATA TO CHARACTERIZE GROUP DIFFERENCES IN A MULTIPLE CLASS COMPARISON

The large amount of data, obtained from measurements with an ion mobility spectrometer (IMS) coupled to a multi-capillary column (MCC), leads to the need of an efficient data reduction to the essential information for the specific problem. A common task is the comparison of several groups of objects to find some characteristic differences between them. In IMS data those variations can only become manifest in different peak structures. Therefore, it seems suitable to merge a raster to the peak positions occurred in the measurements of the existing sample. Afterwards data reduction can be enhanced by computing a centroid value for each detected peak region. So after data processing, the only left is a vector with its length corresponding to the number of different peaks in the sample. This data vector can be the base for a following statistical analysis, which aims at selecting those constructed variables, which show a significant difference between the groups. In conclusion, a linear discriminant analysis including differential expressed variables results in a decision rule, which ideally allows the correct distinction between the different groups.