Mass spectrometry imaging is the method of scanning a sample of interest and generating an ‘image’ of the intensity distribution of a specific analyte/ion. This method results in a large number of spectra which are typically acquired with identical measurement parameters. The application of MS imaging is rapidly growing with a constantly increasing number of different instrumental systems and software tools. COMPUTIS is an EU funded project that develops new and improved technologies for molecular imaging mass spectrometry. An important task within this project is the comparison of images generated by diverse types of mass spectrometers. Both the DICOM standard for in-vivo imaging data and the mzML standard by HUPO-PSI are not able to completely represent an imaging MS experiment. Therefore a standardized data format was developed to simplify the exchange of imaging MS data between different instrument and data analysis software.
Several previous imaging MS data formats utilize two separate files: a small (ini or XML) file for the metadata and a larger (binary) file for the mass spectral data (biomap, DCE, udp). We decided to keep this structure in order to ensure flexible and fast handling of the imaging MS data. We also decided that the metadata file should be based on the mass spectrometry standard mzML developed by HUPO-PSI. The MS data is stored in one of two binary formats in order to ensure the most efficient storage of these large data sets. A new controlled vocabulary was compiled for imzML to include parameters that are specific for imaging experiments. These parameters are stored in the imagingMS.obo file.

Why not use mzML?
First of all mzML was not available at the beginning of the COMPUTIS project (2006). At later stages we evaluated mzML with respect to imaging MS data. The main concern was the file size of converted data sets. Storing the MS data in a separate binary file is crucial for handling the very large imaging MS data sets. After it became clear that our main requirement (a separate binary file) was not possible in mzML, we decided to continue with our own format. However, we decided to store our metadata in the mzML format in order to be able to easily convert between the two formats.
During the last years we stayed in contact with HUPO-PSI at various occasions. The result of these discussions was that we call our format imzML (for imaging mzML) and that it will exist in parallel to mzML for specific use cases (e.g. large data sets). The structure of the XML metadata file will remain compatible with mzML 1.1. The imaging specific cv parameters will be kept in a separate imagingMS.obo file. A number of entries from this file (which were of general importance) have already been included in the psi-ms controlled vocabulary.

Selected imzML presentations and contact with HUPO-PSI:

• HUPO 2006: initial contact with HUPO-PSI (Randy Jullian)
• 2007: Post on (dataXML) discussion group (separate binary file)
• ASMS 2008: idea to name our format imzML (Randy Julian)
• HUPO 2008: imzML poster
• March 2009: ProDac workshop (Lennart Martens)
• March 2009: post on the mzML discussion group
• HUPO-PSI workshop in Turku (April 2009): discussion of imzML
• ASMS 2009: imzML poster