Problems with intensity measurement for pictures


#1

Hi, everybody,

I am trying to measure intensity for two pictures by simply using two modules in a pipeline: LoadSingleImage, followed by MeasureImageIntensity. Strangely, two pictures with obvious different intensity by naked eys (see the attached pictures), resulted in almost same intensity from CP for either total intensity or mean intensity. Did I do anything wroing when measuring the intensity? By the way, the two pictures were taken with the same exposure time, so I assume that the value for background should be same as well.

Thank you for your help.





#2

hi paradise

You’d probably want to use the module Applythreshold to set the background greys to blacks, then when you use the measure module, you’ll only be measuring your objects of interest.
If you want to segment individual cells, you’d need to use the correctillumination modules (calculate & apply) and then an IdentifyPrimAutomatic to get your intensity per cell.

hope that helps! :stuck_out_tongue:

Regards


#3

Hi, Pan,

Thank you for your reply. Yes, I realized I should use ‘threshold’.

But the question followed is that how to set the number where it says “Pixels below this value will be set to zero” ? I found this number is crucial for the measurement: the larger you set, the higher chance of loss of information (less cells will be identified); the lower, the less chance of noise removal.

Is there anyway to calculate the threshold or this number is set in an empirical way?

Thanks.


#4

Hi,

[quote=“paradise”]
But the question followed is that how to set the number where it says “Pixels below this value will be set to zero” ? I found this number is crucial for the measurement: the larger you set, the higher chance of loss of information (less cells will be identified); the lower, the less chance of noise removal. [/quote]

Based on this requirement, it sounds like you may want to take a closer look at the IdentifyPrimAutomatic module. It takes an image as input and produces a set of identified objects as output, with the threshold being set automatically.

Here, you have a wide selection of thresholding algorithms to choose from, with several parameters you can set. It may seem intimidating at first, but try each of the global thresholding methods and see which one gives you the best results, then the adaptive ones if the global don’t do the job (It looks like background or robustbackground may do well). But overall, this module allows you to find the trade-off between object detection and noise removal in a fairly flexible way.

Let us know if you have any problems!
-Mark


#5

One thing that I recently came to the realisation was the fact that when you choose an algorithm for the automatic thresholding, it infact calculates the threshold for you (I ignorantly ignored the automatic part for about 8 months :open_mouth: ). This somewhat negates the manual input of min,max a couple of options down (after threshold correction factor). At the same time, knowing what the threshold correction factor does also helped me optimise the settings even further.

In the end, I think its all there in the help guides, just a matter of piecing it all together - which makes CP even more powerful.

Paradise - if you load up your image, through a module then open a separate window, use the show/hide pixel data and use your mouse to hover over your background to find your threshold.

Regards,