Fluorescence intensity measurement in individual cells through time

Hello everybody,

I have some troubles about measuring fluorescence intensity in cells (timelapse experiment). My goal is “simple”: measure one fluorescence channel intensity in individual cells through time (for instance Hoechst and GFP stained cells).
I created a pipeline which give me a good segmentation of my cells (Primary object= Nuclei, Secondary object=Cells, Tertiary Object= Cytoplasm).
I used the function Track Object and I notice that this function works perfectly to attribute a specific number for individual cells (which never change during all the acquisition, despite cells motility, appearance of other cells etc).
But the problem appears when I export my data in spreadsheet: cell numbering change through the time. I noticed this problem by looking the intensity which highly differs sometimes between two time position. I upload 4 files to show you the problem: the two colored images show the “TrackedCells” which have a good numbering through all the time, I marked the cell n°41 with a white circle.
The two screenshots show the corresponding exported spreadsheet obtained, you will notice that the cell n°41 is perfectly measure for the first position (Intensity around 20) but not for the second position (Intensity= 0) and seams to be numbered as n°42 instead of 41…

Thus I have several questions:

  • Is it possible to obtain the fluorescence intensity of “Tracked cells” (I can save a color image with cells numbered, but this image can’t be use for the “MeasureIntensity” function because it’s a color image instead of grayscale).
  • Why is there a numbering difference between the function TrackedCells and the Spreadsheet export ?
  • and obviously… how can I resolve these troubles !

In optimal way I would like to create a file (spreadsheet file) when all the cells are identify by an individual invariant number with the corresponding fluorescence intensity through the time.
I also uploaded the pipeline I use.

Thanks a lot for your consideration and help!
Regards,
Baptiste




Fluorescence project.cpproj (774.9 KB)

Hi! I really don’t know a lot about this, but could it be that … Cellprofiler assigns a number to every cell based on the relative position in each image, i.e. from top-left to bottom right, as it usually does with objects? That way if a cell moves, you may not get the same object number depending on the relative position of other cells. IF this is the case…?

I wish I knew the answer :slight_smile: but maybe you can investigate

Fabba

Hi, I’d like to help. Providing your pipeline is very helpful. Would you please reply to me with an attachment containing a sample of your original images in a zip file? Would you mind sharing what is GFP labeled and when the Hoechst stain is added to your sample?

Thanks for your responses :wink:

Fabba123 => Indeed I think you’re right and I’ve the same impression, the objects can be followed and identified with a unique number (as shown in my prior pictures) but when the spreadsheets are exported this numbering is not taken into account and a “basic” indexation is used to number the cells, probably from top-left to bottom right. With the file “Objects relationships” and the column “TrackObject” I found that it’s possible to match the object numbering, but only from one image (and so one timepoint) to the next (i.e: yellow line in my enclosed snapshot), this manuel check is very time consuming and not adapted for long-term imaging with hundreds pictures.

Karhohs => I provide you 8 TIFF timepoints from this experiment to avoid a too huge file (because actually there are 98 timepoints, *3 channels… =1.56 Gygabytes!!). It will be sufficient to identify the problem.
First channel is Hoechst, added to my specimen just before imaging
Second channel is GFP, it correspond to a specific ERK localisation biosensor
Third channel is DIC
Here is a dropbox link to download the zip: https://www.dropbox.com/s/uooauvwxdw8viy2/Files%20CP.zip?dl=0

Many thanks for your help.
Baptiste

Hi, I just read up on the tracking module… Am I wrong in thinking “This ought to work though”??

Did you follow the module help to the letter? It seems very complex, lots of things could go wrong. Also, are you sure you are not overlooking a sort of “unique ID” for an entire “cell track”?

The reason I’m asking is I am very interested in timelapse tracking and going to be doing it in near future, so any pointers or experience you can share with me would be greatly appreciated.

I had the same exact issue and still trying to figure it out. I built the pipeline following the ‘track object’ example posted on the website but it seems that for every frame objects are assigned with different numbers. If anybody has a clue, please share. Thanks

It seems like a big part of your issue is that there’s a case where what’s segmented as 39 and 40 in your top image (2 objects) is segmented as only a single object (39) in your bottom image, which may be messing up the tracking. Given how many cells you have per image and that they’re relatively clumpy this seems like a great place to use LAP tracking instead of Overlap tracking- this will allow objects to merge and split (for example when a cell is sometimes segmented as one object or two), and will give you a static “TrackObjectID” that I think is what you’re looking for. Try it and see if it helps your pipeline- if I have time the next few days I’ll see if I can set it up for you but I can’t commit to it yet.

Actually, for your current tracking method I think what should be marked in your output spreadsheet as “TrackObjects_Label_50” may actually be the column you want- I still think given the clumpiness of your cells they’ll track better with LAP (which is designed for exactly this sort of scenario)- if you can upload your .csv outputs that’d help.

EDIT
Downloaded your images and pipeline and re-ran this myself- yup, you want the TrackObjects_Label column- if you sort your spreadsheet by that you’ll see that it’s closely related though not identical to just “ObjectNumber”, which just lists all the objects in an image and thus is susceptible to changing as cells appear and disappear. Also, you can absolutely track fluorescence, the problem is right now that you’re only measure intensity of the “Cytoplasm” objects, measure the “Cells” as well (or track “Cytoplasm” instead) and you’ll be good to go.

1 Like