Unsupervised machine learning for exploratory data analysis in imaging mass spectrometry.

Area of Science:

• Analytical Chemistry
• Computational Biology
• Data Science

Background:

• Imaging mass spectrometry (IMS) offers high chemical specificity for molecular distribution mapping without target tagging.
• IMS generates large, high-dimensional datasets, necessitating automated computational analysis for effective exploration.
• Existing computational research covers preprocessing, dimensionality reduction, classification, and data fusion for IMS.

Purpose of the Study:

• To review unsupervised machine learning methods for exploratory analysis of imaging mass spectrometry data.
• To provide an entry point for analytical chemists and computer scientists interested in IMS data analysis.
• To cover a range of IMS modalities, including MALDI, DESI, and SIMS.

Main Methods:

• Focus on unsupervised machine learning techniques: factorization, clustering, and manifold learning.
• Review computational methods applied to various IMS modalities.
• Synthesize existing research on IMS data analysis.

Main Results:

• Unsupervised machine learning methods are crucial for handling high-dimensional IMS data.
• Factorization, clustering, and manifold learning are key approaches for exploratory IMS analysis.
• The review covers diverse IMS techniques and computational strategies.

Conclusions:

• Unsupervised machine learning provides essential tools for exploratory analysis of complex IMS data.
• This review serves as a guide for researchers entering the field of IMS computational analysis.
• Bridging analytical chemistry and computer science is vital for advancing IMS data interpretation.