counting in multiple ROIs

[Pim Dekker]

Hello,

 

I'm using ImageJ to analyse colonies of cells in plates. I've succeeded
in getting digital pictures of a good enough quality and resolution to
be able to see separate cells within the colonies. So with the 'Analyze
particles' command I can count the total number of cells in the picture.
But what I would like to do is to count the number of cells within each
colony. It seems to me that it shouldn't be too difficult to write a
plugin or macro to be able to do this and I could imagine that someone
has already done this.

 

Does anybody know of a macro or plugin to do this? Or could anyone help
me fiure out a way to count particles in multiple ROIs?

 

Many thanks in advance,

 

Pim Dekker

PhD student

Leiden University Medical Center
Department of General Internal Medicine
Postbox 9600
2300 RC Leiden
The Netherlands
Tel.: [phone]
Fax.: [phone]
E-mail: [hidden email]

 

[Justin McGrath]

Hi Pim,
I don't know if someone else has answered your question yet, but I
would make a selection  around each colony and then use "Analyze
Particles..." then it will only look for particles in that ROI.  If
you select each colony then store them with the ROI manager, you could
have a macro go through each selection, but I don't think that would
speed anything up.

If your multiple colonies will always be in about the same positions
in the image then you could automate it more.  For instance, if you
had a colony in each corner you could use a macro to make a selection
in the upper left quadrant then run "Analyze Particles...", then the
upper right etc.

And if they weren't always in the same position but were well
separated from eachother and you really wanted to automate it, you
could probably figure out some math to determine which particles
belong to the same colony.  My first thought is that you would run
"Analyze Particles..." and measure their coordinates, as well as
whatever else you want, then do some grouping to minimize the variance
of the distance between particles.

Justin

On 2/23/07, Pim Dekker <[hidden email]> wrote:

> Hello,
>
>
>
> I'm using ImageJ to analyse colonies of cells in plates. I've succeeded
> in getting digital pictures of a good enough quality and resolution to
> be able to see separate cells within the colonies. So with the 'Analyze
> particles' command I can count the total number of cells in the picture.
> But what I would like to do is to count the number of cells within each
> colony. It seems to me that it shouldn't be too difficult to write a
> plugin or macro to be able to do this and I could imagine that someone
> has already done this.
>
>
>
> Does anybody know of a macro or plugin to do this? Or could anyone help
> me fiure out a way to count particles in multiple ROIs?
>
>
>
> Many thanks in advance,
>
>
>
> Pim Dekker
>
> PhD student
>
> Leiden University Medical Center
> Department of General Internal Medicine
> Postbox 9600
> 2300 RC Leiden
> The Netherlands
> Tel.: [phone]
> Fax.: [phone]
> E-mail: [hidden email]
>
>
>

[Pim Dekker]

Hi Justin,

Thanks for your reaction. The only other reaction I've had is from Jim
Quinn, suggesting that I should put the image somewhere on a URL and
tell people where it is so they can try to do what I'm after. I will try
to do this some time soon.

As you suggest I did find out that it is possible to manually draw the
ROIs around the colonies. You can then measuere the number and area of
the colonies (i.e. ROIs) and this is already very nice. But I haven't
managed to figure out how to count the cells within the ROIs
automatically, i.e. to make a macro that will give in one table the
total number of colonies and the results of each colony i.e. the area
and number of cells it consists of.

Unfortunately I'm not in a position in which my ROI's are in the same
position: the colonies could be anywhere on the plate.

Your last idea sounds good, but I don't think my math knowledge is up to
scratch to do what you suggest. Overlapping colonies would be a problem
as well.

Thanks again.

Pim

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of
Justin McGrath
Sent: Thursday, March 01, 2007 3:27 PM
To: [hidden email]
Subject: Re: counting in multiple ROIs

Hi Pim,
I don't know if someone else has answered your question yet, but I
would make a selection  around each colony and then use "Analyze
Particles..." then it will only look for particles in that ROI.  If
you select each colony then store them with the ROI manager, you could
have a macro go through each selection, but I don't think that would
speed anything up.

If your multiple colonies will always be in about the same positions
in the image then you could automate it more.  For instance, if you
had a colony in each corner you could use a macro to make a selection
in the upper left quadrant then run "Analyze Particles...", then the
upper right etc.

And if they weren't always in the same position but were well
separated from eachother and you really wanted to automate it, you
could probably figure out some math to determine which particles
belong to the same colony.  My first thought is that you would run
"Analyze Particles..." and measure their coordinates, as well as
whatever else you want, then do some grouping to minimize the variance
of the distance between particles.

Justin

On 2/23/07, Pim Dekker <[hidden email]> wrote:
> Hello,
>
>
>
> I'm using ImageJ to analyse colonies of cells in plates. I've
succeeded
> in getting digital pictures of a good enough quality and resolution to
> be able to see separate cells within the colonies. So with the
'Analyze
> particles' command I can count the total number of cells in the
picture.
> But what I would like to do is to count the number of cells within
each
> colony. It seems to me that it shouldn't be too difficult to write a
> plugin or macro to be able to do this and I could imagine that someone
> has already done this.
>
>
>
> Does anybody know of a macro or plugin to do this? Or could anyone
help

> me fiure out a way to count particles in multiple ROIs?
>
>
>
> Many thanks in advance,
>
>
>
> Pim Dekker
>
> PhD student
>
> Leiden University Medical Center
> Department of General Internal Medicine
> Postbox 9600
> 2300 RC Leiden
> The Netherlands
> Tel.: [phone]
> Fax.: [phone]
> E-mail: [hidden email]
>
>
>

[Jim Cant]

This is probably overkill for this situation but it seems to fit right in.  One could define the ROIs using a clustering technique once the cell (particles) have been located.  Here's the link where I discovered this
http://www.ucl.ac.uk/oncology/MicroCore/HTML_resource/Hier_Clust.htm 
; it's part of this tutorial on microarrays http://www.ucl.ac.uk/oncology/MicroCore/HTML_resource/TUTORIAL_welcome.htm

Overlap would still be an issue.  In fact, how would you do this by hand; do you just divide any overlapped region in two through the isthmus?

jim

<quote author='Pim Dekker'>
Hi Justin,

Thanks for your reaction. The only other reaction I've had is from Jim
Quinn, suggesting that I should put the image somewhere on a URL and
tell people where it is so they can try to do what I'm after. I will try
to do this some time soon.

As you suggest I did find out that it is possible to manually draw the
ROIs around the colonies. You can then measuere the number and area of
the colonies (i.e. ROIs) and this is already very nice. But I haven't
managed to figure out how to count the cells within the ROIs
automatically, i.e. to make a macro that will give in one table the
total number of colonies and the results of each colony i.e. the area
and number of cells it consists of.

Unfortunately I'm not in a position in which my ROI's are in the same
position: the colonies could be anywhere on the plate.

Your last idea sounds good, but I don't think my math knowledge is up to
scratch to do what you suggest. Overlapping colonies would be a problem
as well.

Thanks again.

Pim
<---snip--->