Peak threshold

We would like to measure the number of peaks in the profile. For this we somehow need to define what a significant peak is.

In terms of intensity, there are several factors that affect the intensity of a peak. This includes expression level, exposure time, detector gain, excitation light intensity.

Ideal would be if we find a way to define a peak that is robust to changes in the above factors.

One idea is to measure the noise in the background, hoping that this is proportional to the above factors and define a peak relative to that. However, to determine whether this actually works we would need to acquire samples with different settings; and see how this affects BG noise and peak intensity.

Other ideas could be to do an automated thresholding on the whole image (e.g. Otsu). This would give us some overall intensity of the structure and we could try to define a peak threshold somehow relative to this.

We could also manually set one fixed threshold. However, this would then certainly need to be adapted if any of the above mentioned factors is changed.

Another idea could be to not use so much the height of the peak as a criterion, but rather whether it is the brightest pixel in a certain spatial range. So there's nothing brighter, let's say, within a radius of 10 micrometer than we say this is a significant peak. This would of course also give us peaks in the background noise, however there we could actually meaningfully use the background noise as a threshold to filter out such very dim peaks.