The goal of normalisation is to scale the intensities of each pixel to remove systematic artefacts that affect intensity. For further reading on normalisation methods there are a number of articles that discuss this further
Preprocessing Workflow Editor
by selecting Edit
button in Spectral Preprocessing
panel.Normalisation
method (see Choosing the most appropriate method section below) and click the adjacent +
button.OK
to close the Edit Preprocesing Method
window and then OK
again to close the Preprocessing Workflow Editor
. The chosen method(s) will now be automatically applied to any viewed spectrum.Choosing an appropriate normalisation method is challenging and depends on the artefacts that need to be removed. The methods included in SpectralAnalysis are described below.
This method normalises the data such that the sum of the squares of each spectrum will always be add up to 1
This normalises by dividing the intensities of each spectrum by the median intensity for that spectrum. In some datasets (particularly protein imaging) this is an estimation of the baseline of the data.
This method aims to estimate the noise level in the dataset using the method described by Deininger et al. and normalises to this. This assumes that the noise in the data should be constant.
p-norm is a generalisable variation on the l2 normalisation where the sum of the power p
specified by the user adds up to 1. In the case where p = 2, this is equivalent to the l2 norm, and p=1 is equivalent to the TIC norm
Root mean square normalisation scales the intensities to the square root of the the arithmetic mean of the squares of the intensities for each spectrum.
This method scales the intensities of each spectrum such that they sum to 1. This method assumes that each spectrum should have the same total number of ion present.