Statistical Procedure for IMS Data Analysis

In MALDI-IMS of tissue samples, since the tissue contains an enormous variety of biomolecules, a complex mass spectrum with hundreds of to a thousand peaks can be obtained from a single data point. Furthermore, several thousands of spectra with spatial data are obtained at one IMS experiment. Because of the complexity and enormousness of the IMS dataset, manual processing of the dataset to obtain significant information (e.g., identification of disease-specific mass signature) is not a realistic procedure. In this regard, today, multivariate analysis becomes a powerful tool in IMS data analysis. In this chapter, we describe an unsupervised multivariate data analysis technique that enables us to sort the data sets without any reference information. Particularly, two major methods that are related to IMS, namely, hierarchical clustering and principal component analysis (PCA), are described in detail with examples. Finally a basic procedure for PCA with familiar software (such as Microsoft Excel) is introduced.