Interactive Visual Analysis of Mass Spectrometry Imaging Data Using Linear and Non-Linear Embeddings

Abstract

Mass spectrometry imaging (MSI) is an imaging technique used in analytical chemistry to study the molecular distribution of various compounds at a micro-scale level. For each pixel, MSI stores a mass spectrum obtained by measuring signal intensities of thousands of mass-to-charge ratios (m/z-ratios), each linked to an individual molecular ion species. Traditional analysis tools focus on few individual m/z-ratios, which neglects most of the data. Recently, clustering methods of the spectral information have emerged, but faithful detection of all relevant image regions is not always possible. We propose an interactive visual analysis approach that considers all available information in coordinated views of image and spectral space visualizations, where the spectral space is treated as a multi-dimensional space. We use non-linear embeddings of the spectral information to interactively define clusters and respective image regions. Of particular interest is, then, which of the molecular ion species cause the formation of the clusters. We propose to use linear embeddings of the clustered data, as they allow for relating the projected views to the given dimensions. We document the effectiveness of our approach in analyzing matrix-assisted laser desorption/ionization (MALDI-2) imaging data with ground truth obtained from histological images.

Fingerprints of Element Concentrations in Infective Endocarditis Obtained by Mass Spectrometric Imaging and t-Distributed Stochastic Neighbor Embedding

“…To preserve the full image resolution in the analysis of all detected elements in the ICP-MS imaging data and to obtain correlations of the distribution maps, we next employed t-SNE , implemented in the visual analytics of mass spectroscopy imaging (VAMSI) tool (available upon request) . t-SNE is a nonlinear embedding method that maps higher-dimensional data to a typically two-dimensional visual space trying to preserve neighborhoods.…”

“…To perform quantitative analysis and identify the elements that contributed most to cluster separation, we performed linear embedding as recently described . Using star coordinates (SC), the linear embedding can be interactively adjusted.…”

“…The resulting data set comprised two spatial and seven spectral dimensions. Data were either prepared as heat maps for each element individually, converted to star coordinates, , and represented as one 7D map (Figure S2) or subjected to t-SNE analysis. ,,, …”

“…t-SNE analysis of the nine (two spatial and seven spectral) dimensional LA-ICP-MS data was performed using the program VAMSI . The purpose of this procedure was to summarize information from the seven elemental distribution maps in one image, which takes similarities of the individual pixel into account.…”

“…For this purpose, t-distributed stochastic neighbor embedding (t-SNE) , has previously been used for hyperspectral visualization of molecular MS imaging data. − This approach maps the high-dimensional hyperspectral information of MS imaging data to a two-dimensional scatter plot visualization, preserving high-dimensional patterns, which allows for the detection of clusters. Mapping these clusters back onto the original sample, for example, via color coding, opens the possibility to identify typical molecular or elemental distribution patterns in previously inconspicuous tissue regions . More recently, automated segmentation has also been applied to LA-ICP-MS data using t-SNE and other clustering algorithms. − …”