3.3. Data Analysis

The GCxIMS-plots were interpreted manually, using the G.A.S. software VOCal 0.1.4. Peaks were selected manually, and their position was determined by choosing a coverage area. This area consists of the retention (in seconds) and drift time (in milliseconds) range in which the selected peak is situated and then added to the data set for further use. The signal intensity (in V) was determined by the software automatically using the maximum of the selected area.

For data processing, an initial step of peak alignment was carried out. A drift time scale using a reactant ion peak (RIP) as reference was carried out automatically by the software. Then, peaks (to be called markers) were selected by visual exploration of the GCxIMS plots of each sample and their intensities were selected as the analytical signals.

A non-supervised principal component analysis using auto-scales was carried out in order to reduce the dimensionality. Then, PCA scores were taken to carry out a linear discriminant analysis. The LDA incorporated PCA components automatically, carrying out at least 5% of variance.