ImageJ

Re: Counting particle (fluorescence labelled) using ImageJ

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

8 messages Options

Sindy Kueh

Re: Counting particle (fluorescence labelled) using ImageJ

Hi all

I am a newbie with using imageJ and am trying to count the number of
clusters and also to measure the cluster size of GABAA receptor a1
subunit (fluorescence labelled). I have tried looking up on the ImageJ
homepage and could not find or do not know where to look for information
on how to correctly set threshold etc on counting these
clusters/particles. I have downloaded a 2 page pdf file which explains
how to count particle etc, but have difficulty locating the file-embryos
in which is needed to do the "tutorial".
I read that in order to perform the analysis, I have to change the file
type to either 16-bit or 8 bit-gray scale, which I did... however, I had
difficulty trying to set the threshold and have returned results in the
order of thousands of clusters where only a few hundreds are present. I
have attached an example of a file of what I need to measure... the
clusters are labelled green.
I hope someone can help.

Many thanks

Best regards
Sindy

<<Example1.JPG>>

Example1.JPG (41K) Download Attachment

daschneider9

Re: Counting particle (fluorescence labelled) using ImageJ

Sindy,

Perhaps the use of a Gaussian blur filter might help. When I change your image to 16-bit, apply Gaussian blur filter with Sigma = 1.0, and apply auto default threshold for a dark background, then particle analyzer detects 615 particles. Using the generated ROIs to map these particles back to the original image looks like a good _first_ approximation. Perhaps further tweaking the Gaussian blur Sigma value, using watershed, and setting minimum size cutoffs might return values closer to your eyeball count.

In the end you might still need to "filter" the results in order to extract information only on those shape characteristics that you consider valid for actual clusters. For example, the results on the 615 include 350 with an "area" between 1-15 (I used the histogram function on the results window to get a feel for the particles detected). Perhaps you'd say these do not meet your threshold for inclusion in the counts (validated of course by your own manual counts). With this example of 615, you could re-run the particle analyzer and set "size" to exclude particles below 15. Now the analyzer detects 285 with minimum area of 15.

I have no idea if this is correct for your purposes but I hope it helps demonstrate some possibilities you can consider. I hope others might comment on this suggestion as I am also very new to this and eager to learn.

Good luck.
Dave
-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Sindy Kueh
Sent: Monday, February 01, 2010 9:26 PM
To: [hidden email]
Subject: Re: Counting particle (fluorescence labelled) using ImageJ

Hi all

I am a newbie with using imageJ and am trying to count the number of
clusters and also to measure the cluster size of GABAA receptor a1
subunit (fluorescence labelled). I have tried looking up on the ImageJ
homepage and could not find or do not know where to look for information
on how to correctly set threshold etc on counting these
clusters/particles. I have downloaded a 2 page pdf file which explains
how to count particle etc, but have difficulty locating the file-embryos
in which is needed to do the "tutorial".
I read that in order to perform the analysis, I have to change the file
type to either 16-bit or 8 bit-gray scale, which I did... however, I had
difficulty trying to set the threshold and have returned results in the
order of thousands of clusters where only a few hundreds are present. I
have attached an example of a file of what I need to measure... the
clusters are labelled green.
I hope someone can help.

Many thanks

Best regards
Sindy

<<Example1.JPG>>

dpoburko

Re: Counting particle (fluorescence labelled) using ImageJ

In reply to this post by Sindy Kueh

Hi Sindy,

It sounds like you have probably found the Analyze>Analyze Particles
function, which is a good place to start. Probably the most important
aspect that you will have to set, other than the threshold, will be your
minimum and maximum particle size. If you images are scaled (i.e. in
microns, etc), the units will be square microns, otherwise it will just
be in pixels (which is more intuitive). You can get an idea of the sizes
that make sense for you particles by tracing a few by hand and measuring
their area. (i.e. trace, add to the ROI manager ("T" is the shortcut),
repeat for other clusters, then set your measurements to include area).

As for the thresholding, there are a few things to consider.
1. Running a Gaussian filter (Process>Filters>Gaussian Blur...) will
help remove noise. This is a handy start to almost any process involving
thresholding. (I applied a Blur of Sigma=0.9 to your image and counted
~600 particles with a modest threshold)
2. Is you background uniform? If not, you could consider using
Process>Subtract Background.. with a ball size a little large than your
clusters. This effectively tries to locally subtract the background
while preserving bright puncta smaller than the size of the rolling ball.
3. Finally, you could consider a method that applied multiple thresholds
to your image and counts clusters/puncta that are restricted to certain
size or shape criteria. This becomes quite usefull in the case of
closely neighbouring puncta that will sometimes turn into one big glob
when using a single threshold that is low enough to let you capture
somewhat dim puncta. If you are interested I will send you just such a
macro.

Best of luck,
Damon

--

Damon Poburko, PhD
Postdoctoral Research Fellow
Stanford University School of Medicine
Dept. of Molecular & Cellular Physiology
279 Campus Dr., Beckman B103, Stanford, CA 94305
Ph: 650 725 7564, fax: 650 725 8021

Sindy Kueh wrote:

> Hi all
>
> I am a newbie with using imageJ and am trying to count the number of
> clusters and also to measure the cluster size of GABAA receptor a1
> subunit (fluorescence labelled). I have tried looking up on the ImageJ
> homepage and could not find or do not know where to look for information
> on how to correctly set threshold etc on counting these
> clusters/particles. I have downloaded a 2 page pdf file which explains
> how to count particle etc, but have difficulty locating the file-embryos
> in which is needed to do the "tutorial".
> I read that in order to perform the analysis, I have to change the file
> type to either 16-bit or 8 bit-gray scale, which I did... however, I had
> difficulty trying to set the threshold and have returned results in the
> order of thousands of clusters where only a few hundreds are present. I
> have attached an example of a file of what I need to measure... the
> clusters are labelled green.
> I hope someone can help.
>
> Many thanks
>
> Best regards
> Sindy
>
>
>
> <<Example1.JPG>>
>
>
>

Sindy Kueh

Re: Counting particle (fluorescence labelled) using ImageJ

Hi Damon,

Would like to thanks you again for the clear instruction. I know it has
been a while ago, but I finally tried performing the analysis as you had
suggested. I am still learning how to use ImageJ and indeed it is a very
useful software if used to its capacity.
Anyway, yes, I am interested in the macro you mentioned in your earlier
email to capture individual neighbouring puncta which might otherwise
become a blob using a single treshold.

With best regards
Sindy

Sindy Kueh
Anatomy and Cell Biology
Room 30.2.09 University of Western Sydney
School of Medicine
Building 30 Goldsmith Ave
Campbelltown Campus NSW 2560
Locked Bag 1797 Penrith South DC NSW 1797
Tel (02) 9852 4722
Fax (02) 9852 4701

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of
Damon Poburko
Sent: Wednesday, February 03, 2010 4:26 AM
To: [hidden email]
Subject: Re: Counting particle (fluorescence labelled) using ImageJ

Hi Sindy,

It sounds like you have probably found the Analyze>Analyze Particles
function, which is a good place to start. Probably the most important
aspect that you will have to set, other than the threshold, will be your
minimum and maximum particle size. If you images are scaled (i.e. in
microns, etc), the units will be square microns, otherwise it will just
be in pixels (which is more intuitive). You can get an idea of the sizes
that make sense for you particles by tracing a few by hand and measuring
their area. (i.e. trace, add to the ROI manager ("T" is the shortcut),
repeat for other clusters, then set your measurements to include area).

As for the thresholding, there are a few things to consider.
1. Running a Gaussian filter (Process>Filters>Gaussian Blur...) will
help remove noise. This is a handy start to almost any process involving
thresholding. (I applied a Blur of Sigma=0.9 to your image and counted
~600 particles with a modest threshold) 2. Is you background uniform? If
not, you could consider using
Process>Subtract Background.. with a ball size a little large than your
clusters. This effectively tries to locally subtract the background
while preserving bright puncta smaller than the size of the rolling
ball.
3. Finally, you could consider a method that applied multiple thresholds
to your image and counts clusters/puncta that are restricted to certain
size or shape criteria. This becomes quite usefull in the case of
closely neighbouring puncta that will sometimes turn into one big glob
when using a single threshold that is low enough to let you capture
somewhat dim puncta. If you are interested I will send you just such a
macro.

Best of luck,
Damon

--

Damon Poburko, PhD
Postdoctoral Research Fellow
Stanford University School of Medicine
Dept. of Molecular & Cellular Physiology
279 Campus Dr., Beckman B103, Stanford, CA 94305
Ph: 650 725 7564, fax: 650 725 8021

Sindy Kueh wrote:
> Hi all
>
> I am a newbie with using imageJ and am trying to count the number of
> clusters and also to measure the cluster size of GABAA receptor a1
> subunit (fluorescence labelled). I have tried looking up on the ImageJ

> homepage and could not find or do not know where to look for
> information on how to correctly set threshold etc on counting these
> clusters/particles. I have downloaded a 2 page pdf file which explains

> how to count particle etc, but have difficulty locating the
> file-embryos in which is needed to do the "tutorial".
> I read that in order to perform the analysis, I have to change the
> file type to either 16-bit or 8 bit-gray scale, which I did...
> however, I had difficulty trying to set the threshold and have
> returned results in the order of thousands of clusters where only a
> few hundreds are present. I have attached an example of a file of what

> I need to measure... the clusters are labelled green.
> I hope someone can help.
>
> Many thanks
>
> Best regards
> Sindy
>
>
>
> <<Example1.JPG>>
>
>
>

Du, Guanghua

Re: Counting particle (fluorescence labelled) using ImageJ

Hi, Sindy,
you can have a try using the focipicker plugin which was written
originally for counting immunofluorescence particles.
http://rsbweb.nih.gov/ij/plugins/foci-picker3d/

best regards

guanghua

Sindy Kueh wrote:

> Hi Damon,
>
> Would like to thanks you again for the clear instruction. I know it has
> been a while ago, but I finally tried performing the analysis as you had
> suggested. I am still learning how to use ImageJ and indeed it is a very
> useful software if used to its capacity.
> Anyway, yes, I am interested in the macro you mentioned in your earlier
> email to capture individual neighbouring puncta which might otherwise
> become a blob using a single treshold.
>
> With best regards
> Sindy
>
>
> Sindy Kueh
> Anatomy and Cell Biology
> Room 30.2.09 University of Western Sydney
> School of Medicine
> Building 30 Goldsmith Ave
> Campbelltown Campus NSW 2560
> Locked Bag 1797 Penrith South DC NSW 1797
> Tel (02) 9852 4722
> Fax (02) 9852 4701
>
> -----Original Message-----
> From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of
> Damon Poburko
> Sent: Wednesday, February 03, 2010 4:26 AM
> To: [hidden email]
> Subject: Re: Counting particle (fluorescence labelled) using ImageJ
>
> Hi Sindy,
>
> It sounds like you have probably found the Analyze>Analyze Particles
> function, which is a good place to start. Probably the most important
> aspect that you will have to set, other than the threshold, will be your
> minimum and maximum particle size. If you images are scaled (i.e. in
> microns, etc), the units will be square microns, otherwise it will just
> be in pixels (which is more intuitive). You can get an idea of the sizes
> that make sense for you particles by tracing a few by hand and measuring
> their area. (i.e. trace, add to the ROI manager ("T" is the shortcut),
> repeat for other clusters, then set your measurements to include area).
>
> As for the thresholding, there are a few things to consider.
> 1. Running a Gaussian filter (Process>Filters>Gaussian Blur...) will
> help remove noise. This is a handy start to almost any process involving
> thresholding. (I applied a Blur of Sigma=0.9 to your image and counted
> ~600 particles with a modest threshold) 2. Is you background uniform? If
> not, you could consider using
> Process>Subtract Background.. with a ball size a little large than your
> clusters. This effectively tries to locally subtract the background
> while preserving bright puncta smaller than the size of the rolling
> ball.
> 3. Finally, you could consider a method that applied multiple thresholds
> to your image and counts clusters/puncta that are restricted to certain
> size or shape criteria. This becomes quite usefull in the case of
> closely neighbouring puncta that will sometimes turn into one big glob
> when using a single threshold that is low enough to let you capture
> somewhat dim puncta. If you are interested I will send you just such a
> macro.
>
> Best of luck,
> Damon
>

--
best wishes!
**************************************
Guanghua Du, PhD
James-Franck Str. 1
Physik Dept, E12, TUM
85748, Garching b. Muenchen
Germany

Tel:+49-89-28914286
E-mail: [hidden email]
**************************************

dpoburko

Re: Counting particle (fluorescence labelled) using ImageJ

In reply to this post by Sindy Kueh

Hi Sindy,

Here is the link to the macro.
http://rsb.info.nih.gov/ij/macros/Mulitple_Thresholds.txt It will
probably take a bit of playing with to get the parameters appropriately
set for your images. Have a gov with it, and it you're having problems,
feel free to drop me a line.

Cheers,
Damon

On 4/27/2010 11:39 PM, Sindy Kueh wrote:

Sindy Kueh

Re: Counting particle (fluorescence labelled) using ImageJ

Hi Damon,

I had a look at the macro.. Thanks. I haven't setup a new macro before
and am at a lost on where to begin... This may sound absolutely silly,
but do I input the whole macro as one, or as different macros....

Hope to hear from you soon..

Sindy
-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of
Damon Poburko
Sent: Thursday, April 29, 2010 6:42 AM
To: [hidden email]
Subject: Re: Counting particle (fluorescence labelled) using ImageJ

Hi Sindy,

Here is the link to the macro.
http://rsb.info.nih.gov/ij/macros/Mulitple_Thresholds.txt It will
probably take a bit of playing with to get the parameters appropriately
set for your images. Have a gov with it, and it you're having problems,
feel free to drop me a line.

Cheers,
Damon

On 4/27/2010 11:39 PM, Sindy Kueh wrote:
> Hi Damon,
>
> Would like to thanks you again for the clear instruction. I know it
> has been a while ago, but I finally tried performing the analysis as
> you had suggested. I am still learning how to use ImageJ and indeed it

> is a very useful software if used to its capacity.
> Anyway, yes, I am interested in the macro you mentioned in your
> earlier email to capture individual neighbouring puncta which might
> otherwise become a blob using a single treshold.
>
> With best regards
> Sindy
>
>
> Sindy Kueh
> Anatomy and Cell Biology
> Room 30.2.09 University of Western Sydney
> School of Medicine
> Building 30 Goldsmith Ave
> Campbelltown Campus NSW 2560
> Locked Bag 1797 Penrith South DC NSW 1797 Tel (02) 9852 4722 Fax (02)
> 9852 4701
>
> -----Original Message-----
> From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of
> Damon Poburko
> Sent: Wednesday, February 03, 2010 4:26 AM
> To: [hidden email]
> Subject: Re: Counting particle (fluorescence labelled) using ImageJ
>
> Hi Sindy,
>
> It sounds like you have probably found the Analyze>Analyze Particles
> function, which is a good place to start. Probably the most important
> aspect that you will have to set, other than the threshold, will be
> your minimum and maximum particle size. If you images are scaled (i.e.

> in microns, etc), the units will be square microns, otherwise it will
> just be in pixels (which is more intuitive). You can get an idea of
> the sizes that make sense for you particles by tracing a few by hand
> and measuring their area. (i.e. trace, add to the ROI manager ("T" is
> the shortcut), repeat for other clusters, then set your measurements
to include area).

>
> As for the thresholding, there are a few things to consider.
> 1. Running a Gaussian filter (Process>Filters>Gaussian Blur...) will
> help remove noise. This is a handy start to almost any process
> involving thresholding. (I applied a Blur of Sigma=0.9 to your image
> and counted ~600 particles with a modest threshold) 2. Is you
> background uniform? If not, you could consider using
> Process>Subtract Background.. with a ball size a little large than
> Process>your
> clusters. This effectively tries to locally subtract the background
> while preserving bright puncta smaller than the size of the rolling
> ball.
> 3. Finally, you could consider a method that applied multiple
> thresholds to your image and counts clusters/puncta that are
> restricted to certain size or shape criteria. This becomes quite
> usefull in the case of closely neighbouring puncta that will sometimes

> turn into one big glob when using a single threshold that is low
> enough to let you capture somewhat dim puncta. If you are interested I

> will send you just such a macro.
>
> Best of luck,
> Damon
>
>

dpoburko

Re: Counting particle (fluorescence labelled) using ImageJ

Hi Sindy,

Here's what you need to do. Copy the macro txt file into your
Image>Plugins>Macros folder in your file explorer. The next time you
open image J, you should see the macro in the Plugins>Macros sub-menu.
Simply click in the macro, and away you go. If that's not making sense,
it's all pretty well explained in the documentation on the image J website.

Good luck!,
Damon

On 4/28/2010 9:09 PM, Sindy Kueh wrote:

> Hi Damon,
>
> I had a look at the macro.. Thanks. I haven't setup a new macro before
> and am at a lost on where to begin... This may sound absolutely silly,
> but do I input the whole macro as one, or as different macros....
>
> Hope to hear from you soon..
>
>
> Sindy
> -----Original Message-----
> From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of
> Damon Poburko
> Sent: Thursday, April 29, 2010 6:42 AM
> To: [hidden email]
> Subject: Re: Counting particle (fluorescence labelled) using ImageJ
>
> Hi Sindy,
>
> Here is the link to the macro.
> http://rsb.info.nih.gov/ij/macros/Mulitple_Thresholds.txt It will
> probably take a bit of playing with to get the parameters appropriately
> set for your images. Have a gov with it, and it you're having problems,
> feel free to drop me a line.
>
> Cheers,
> Damon
>
>
> On 4/27/2010 11:39 PM, Sindy Kueh wrote:
>
>> Hi Damon,
>>
>> Would like to thanks you again for the clear instruction. I know it
>> has been a while ago, but I finally tried performing the analysis as
>> you had suggested. I am still learning how to use ImageJ and indeed it
>>
>
>> is a very useful software if used to its capacity.
>> Anyway, yes, I am interested in the macro you mentioned in your
>> earlier email to capture individual neighbouring puncta which might
>> otherwise become a blob using a single treshold.
>>
>> With best regards
>> Sindy
>>
>>
>> Sindy Kueh
>> Anatomy and Cell Biology
>> Room 30.2.09 University of Western Sydney
>> School of Medicine
>> Building 30 Goldsmith Ave
>> Campbelltown Campus NSW 2560
>> Locked Bag 1797 Penrith South DC NSW 1797 Tel (02) 9852 4722 Fax (02)
>> 9852 4701
>>
>> -----Original Message-----
>> From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of
>> Damon Poburko
>> Sent: Wednesday, February 03, 2010 4:26 AM
>> To: [hidden email]
>> Subject: Re: Counting particle (fluorescence labelled) using ImageJ
>>
>> Hi Sindy,
>>
>> It sounds like you have probably found the Analyze>Analyze Particles
>> function, which is a good place to start. Probably the most important
>> aspect that you will have to set, other than the threshold, will be
>> your minimum and maximum particle size. If you images are scaled (i.e.
>>
>
>> in microns, etc), the units will be square microns, otherwise it will
>> just be in pixels (which is more intuitive). You can get an idea of
>> the sizes that make sense for you particles by tracing a few by hand
>> and measuring their area. (i.e. trace, add to the ROI manager ("T" is
>> the shortcut), repeat for other clusters, then set your measurements
>>
> to include area).
>
>> As for the thresholding, there are a few things to consider.
>> 1. Running a Gaussian filter (Process>Filters>Gaussian Blur...) will
>> help remove noise. This is a handy start to almost any process
>> involving thresholding. (I applied a Blur of Sigma=0.9 to your image
>> and counted ~600 particles with a modest threshold) 2. Is you
>> background uniform? If not, you could consider using
>> Process>Subtract Background.. with a ball size a little large than
>> Process>your
>> clusters. This effectively tries to locally subtract the background
>> while preserving bright puncta smaller than the size of the rolling
>> ball.
>> 3. Finally, you could consider a method that applied multiple
>> thresholds to your image and counts clusters/puncta that are
>> restricted to certain size or shape criteria. This becomes quite
>> usefull in the case of closely neighbouring puncta that will sometimes
>>
>
>> turn into one big glob when using a single threshold that is low
>> enough to let you capture somewhat dim puncta. If you are interested I
>>
>
>> will send you just such a macro.
>>
>> Best of luck,
>> Damon
>>
>>
>>