Evaluation of Digital Image Recognition Methods for Mass Spectrometry Imaging Data Analysis

J Am Soc Mass Spectrom. 2018 Dec;29(12):2467-2470. doi: 10.1007/s13361-018-2073-0. Epub 2018 Oct 15.


Analyzing mass spectrometry imaging data can be laborious and time consuming, and as the size and complexity of datasets grow, so does the need for robust automated processing methods. We here present a method for comprehensive, semi-targeted discovery of molecular distributions of interest from mass spectrometry imaging data, using widely available image similarity scoring algorithms to rank images by spatial correlation. A fast and powerful batch search method using a MATLAB implementation of structural similarity (SSIM) index scoring with a pre-selected reference distribution is demonstrated for two sample imaging datasets, a plant metabolite study using Artemisia annua leaf, and a drug distribution study using maraviroc-dosed macaque tissue. Graphical Abstract ᅟ.

Keywords: Image recognition; Mass spectrometry imaging; SSIM.

MeSH terms

  • Algorithms
  • Animals
  • Image Processing, Computer-Assisted / methods*
  • Lymph Nodes / chemistry
  • Lymph Nodes / metabolism
  • Macaca mulatta
  • Maraviroc / pharmacokinetics
  • Mass Spectrometry / methods*
  • Models, Theoretical
  • Molecular Imaging / methods*
  • Plant Leaves / chemistry
  • Tissue Distribution


  • Maraviroc