Data

MSI images must be in the imzML format. Various tools are available online to convert from proprietary format to mzML or imzML (e.g. imzMLConverter).

Complementary images can be in any ITK format (.png, .tif, .hdr…)

Processed vs. continuous spectra

We assume the mass list is the same for each spectrum in the MSI images, i.e. the storage type is continuous. However, if the mass list is different, i.e. each spectrum is processed, it is possible to convert it to continuous spectra.

For this purpose, use the examples/same_mz_axis.py script. This script selects peaks across all spectra and aligns peaks that are sufficiently close to a common m/z value. This also helps reduce the size of the image.

The script has various parameters:

• factor: the local prominence factor used for peak selection

• level: the noise level estimated across the spectra

• nbpeaks: the number of occurrence for a peak to appear and be considered as a valid reference peak in the alignment procedure

• step: the tolerance value (in m/z) to create groups of peaks that can be matched to a common m/z value

On another note, the spectra can be merged by preserving the information across all spectra, and simply removing duplicates in the mass list. See esmraldi.spectraprocessing.same_mz_axis().