Interactive Estimation of Heterogeneity from Mass Spectrometry Imaging

Area of Science:

• Biomedical Imaging
• Analytical Chemistry
• Computational Biology

Background:

• Mass spectrometry imaging (MSI) generates complex hyperspectral data.
• Assessing spatial structures and heterogeneity in MSI data is crucial for biological interpretation.
• Existing methods may lack interactive visualization for spatial analysis.

Purpose of the Study:

• To develop and demonstrate a novel interactive approach for assessing spatial heterogeneity in hyperspectral MSI data.
• To enable the visualization of similarity levels between different regions within a sample.
• To facilitate feature selection and biological interpretation of tissue samples using MSI.

Main Methods:

• Utilizing cosine distance to quantify similarity between mass spectra of a selected region and the entire image.
• Developing interactive visualization tools based on cosine distance maps.
• Applying the method to hyperspectral mass spectrometry images of mouse brain slices.

Main Results:

• Demonstrated the applicability of the cosine distance visualization for assessing spatial structures in MSI data.
• Showed that similarity maps reveal the spatial distribution of molecular features.
• Highlighted the potential for molecular histological annotation directly from MSI data.

Conclusions:

• The proposed approach provides an effective way to interactively assess spatial heterogeneity in hyperspectral MSI data.
• Cosine distance visualization aids in identifying distinct tissue regions and molecular features.
• This method supports feature selection and biological interpretation, potentially serving as molecular histology.