how to compare intensity ratios (membran <> cytoplasm)?

[Michael Weber-4]

Dear list,

I need your help. The tasks is the following one: I would like to
compare the intensity ratios between membran and cytoplasm stainings on
yeast cells. Here's an example image:

http://www.mpi-cbg.de/~weber/yeast-crop.tif

Measuring the cytoplasm mean intensity is not so difficult, by using a
ROI and the measure function of ImageJ.

But how to get the mean value of the membran only? Maybe someone could
help me with this (simple?) question? This would be great.

Thanks in advance,
Michael

[Volker Baecker]

Hello,
if I understand you right the problem is to segment the membrane only.
You can use "find edges" followed by a smoothing filter. This should
yield you an image with almost only the membranes.
You could then for example use the threshold tool together with the
magic wand (alt click to subtract the black middle part from the roi).
Is this what you wanted to do?
Volker

Michael Weber a écrit :

> Dear list,
>
> I need your help. The tasks is the following one: I would like to
> compare the intensity ratios between membran and cytoplasm stainings
> on yeast cells. Here's an example image:
>
> http://www.mpi-cbg.de/~weber/yeast-crop.tif
>
> Measuring the cytoplasm mean intensity is not so difficult, by using a
> ROI and the measure function of ImageJ.
>
> But how to get the mean value of the membran only? Maybe someone could
> help me with this (simple?) question? This would be great.
>
> Thanks in advance,
> Michael
>

--
passerelle antivirus du campus CNRS de Montpellier
--

[Darren Scully]

I don't know if this will help, but could you not just get the mean intensity of the cytoplasm. Then get the mean intensity of the whole cell.  Once you have this, just subtract the cytoplasm mean intensity from the whole cell mean intensity.

Hope it helps.

Darren
  ----- Original Message -----
  From: Michael Weber
  To: [hidden email]
  Sent: Friday, September 29, 2006 2:48 PM
  Subject: how to compare intensity ratios (membran <> cytoplasm)?


  Dear list,

  I need your help. The tasks is the following one: I would like to
  compare the intensity ratios between membran and cytoplasm stainings on
  yeast cells. Here's an example image:

  http://www.mpi-cbg.de/~weber/yeast-crop.tif

  Measuring the cytoplasm mean intensity is not so difficult, by using a
  ROI and the measure function of ImageJ.

  But how to get the mean value of the membran only? Maybe someone could
  help me with this (simple?) question? This would be great.

  Thanks in advance,
  Michael

[Gabriel Landini]

> But how to get the mean value of the membran only? Maybe someone could
> help me with this (simple?) question? This would be great.

This assumes foreground =255 and background =0
You have to calculate the intensities separately.
I would suggest to first segment the cells (as a binary image).
Then find out how "thick" your membranes are (I would guess that 1-2 pixels or  
so, depending on the bleeding of the stain intensity).

Now you create 2 images from the binary one:
Keep the original greyscale (call it "original")
Keep the binary with the nuclei segmented (call it "binary")
Erode the binary by one pixel (or the number of pixels that you need to delete
the membrane). Call this "cytoplasm".
Do: binary-cytoplasm. This would keep the "membrane" pixels (still a binary
image) and call it "cytoplasm"
Now isolate the pixels of the 2 compartments:

original AND membrane = membrane pixels, the rest will be 0
original AND cytoplasm = cytoplasm pixels, the rest will be 0

Use the particle analyzer to find out the greyscale statistics of these blobs
(but be careful how you process the membrane image, so you do not consider
the hole pixels of the membrane image.)

You will have to do this cell by cell, rather than all in one go because it
would be a bit convoluted to find out the corresponding bits when there are
many cells in the image.

I hope it helps

G.

[lechristophe]

That's exactly the way we did it for confocal equatorial images of
spheric HEK cells, using an Object-Image macro :

Manually draw a ROI around the whole cell (ROI1)
Measure Total Fluorescence of the Cell and it's Area
Stretch ROI (Inset) 2-6 pixels (ROI2)
Measure Total Fluorescence of the Cytoplasm and it's Area
Substract Total Fluorescence of ROI2 from Total Fluorescence of ROI1 and
get also Area : Values corresponding to the membrane

all values corrected with background fluo from a ROI traced in the nucleus.

see

Constitutive endocytic cycle of the CB1 cannabinoid receptor.
Leterrier C, Bonnard D, Carrel D, Rossier J, Lenkei Z
J Biol Chem. 2004 Aug 20; 279(34): 36013-21


Christophe Leterrier

Ionic Channels and Neuronal Polarity Team :: Bénédicte Dargent
INSERM UMR 641
Neurobiology of Ionic Channels Lab :: Michael Seagar
Institut Jean Roche :: Université de la Méditerranée
51, boulevard Pierre Dramard :: 13916 Marseille Cedex 20
Telephone [phone] :: Fax [phone]
Website : http://ifrjr.nord.univ-mrs.fr
Email : [hidden email]


Darren Scully a écrit :

> I don't know if this will help, but could you not just get the mean intensity of the cytoplasm. Then get the mean intensity of the whole cell.  Once you have this, just subtract the cytoplasm mean intensity from the whole cell mean intensity.
>
> Hope it helps.
>
> Darren
>   ----- Original Message -----
>   From: Michael Weber
>   To: [hidden email]
>   Sent: Friday, September 29, 2006 2:48 PM
>   Subject: how to compare intensity ratios (membran <> cytoplasm)?
>
>
>   Dear list,
>
>   I need your help. The tasks is the following one: I would like to
>   compare the intensity ratios between membran and cytoplasm stainings on
>   yeast cells. Here's an example image:
>
>   http://www.mpi-cbg.de/~weber/yeast-crop.tif
>
>   Measuring the cytoplasm mean intensity is not so difficult, by using a
>   ROI and the measure function of ImageJ.
>
>   But how to get the mean value of the membran only? Maybe someone could
>   help me with this (simple?) question? This would be great.
>
>   Thanks in advance,
>   Michael
>