multi-channel composite images

[Knecht, David]

The sample data has a 5 channel composite image.  I am unsure how one creates such an image and how the colors are assigned.  If you had 5 grayscale images you wanted to merge to a composite and you want to designate the color of each channel, how would you do that?  Thanks- Dave  

Dr. David Knecht    
Department of Molecular and Cell Biology
Co-head Flow Cytometry and Confocal Microscopy Facility
U-3125
91 N. Eagleville Rd.
University of Connecticut
Storrs, CT 06269
860-486-2200
860-486-4331 (fax)

[ctrueden]

Hi David,

The sample data has a 5 channel composite image.  I am unsure how one
> creates such an image and how the colors are assigned.  If you had 5
> grayscale images you wanted to merge to a composite and you want to
> designate the color of each channel, how would you do that?  Thanks- Dave
>

First, make sure the five grayscale images are part of a single stack (Image
> Stacks > Images to Stack). Then convert it to a Composite image with:
Image > Color > Make Composite. By default, it will assign Red to channel
#1, Green to #2, Blue to #3, Cyan to #4 and Magenta to #5. (For #6 it's
Yellow and #7 is just black-to-white, i.e. Gray.)

To change the LUT for a particular channel, just change it using a command
such as Image > Lookup Tables > Fire. It will only affect the LUT for the
currently selected channel. There are shortcuts built in to the Channels
window that pops up when running "Make Composite"—just click the More button
to choose a LUT preset.

-Curtis

On Fri, May 27, 2011 at 3:03 PM, David Knecht <[hidden email]>wrote:

> The sample data has a 5 channel composite image.  I am unsure how one
> creates such an image and how the colors are assigned.  If you had 5
> grayscale images you wanted to merge to a composite and you want to
> designate the color of each channel, how would you do that?  Thanks- Dave
>
> Dr. David Knecht
> Department of Molecular and Cell Biology
> Co-head Flow Cytometry and Confocal Microscopy Facility
> U-3125
> 91 N. Eagleville Rd.
> University of Connecticut
> Storrs, CT 06269
> 860-486-2200
> 860-486-4331 (fax)
>

[Michael P Ellis]

Dear Dr. David Knecht,

You may want to try my SmartCaptureLite extensions. After having turned your 5 greyscale images into a stack (as Curtis suggests) you can then use the CompositeAdjuster plugin to conveniently assign colours and adjust levels all from a single interactive palette. (Use the MakeComposite button on the CompositeAdjuster palette to convert your stack to a Composite image)

See http://www.dsuk.biz/DSUK/SmartCaptureLite.html and watch the movies to see how to assign colours to the individual planes.

The SmartCapturePlugin is free of charge and I would be interested receive any feedback.

Regards -- Michael Ellis

On 27 May 2011, at 21:44, Curtis Rueden wrote:

> Hi David,
>
> The sample data has a 5 channel composite image.  I am unsure how one
>> creates such an image and how the colors are assigned.  If you had 5
>> grayscale images you wanted to merge to a composite and you want to
>> designate the color of each channel, how would you do that?  Thanks- Dave
>>
>
> First, make sure the five grayscale images are part of a single stack (Image
>> Stacks > Images to Stack). Then convert it to a Composite image with:
> Image > Color > Make Composite. By default, it will assign Red to channel
> #1, Green to #2, Blue to #3, Cyan to #4 and Magenta to #5. (For #6 it's
> Yellow and #7 is just black-to-white, i.e. Gray.)
>
> To change the LUT for a particular channel, just change it using a command
> such as Image > Lookup Tables > Fire. It will only affect the LUT for the
> currently selected channel. There are shortcuts built in to the Channels
> window that pops up when running "Make Composite"—just click the More button
> to choose a LUT preset.
>
> -Curtis
>
> On Fri, May 27, 2011 at 3:03 PM, David Knecht <[hidden email]>wrote:
>
>> The sample data has a 5 channel composite image.  I am unsure how one
>> creates such an image and how the colors are assigned.  If you had 5
>> grayscale images you wanted to merge to a composite and you want to
>> designate the color of each channel, how would you do that?  Thanks- Dave
>>
>> Dr. David Knecht
>> Department of Molecular and Cell Biology
>> Co-head Flow Cytometry and Confocal Microscopy Facility
>> U-3125
>> 91 N. Eagleville Rd.
>> University of Connecticut
>> Storrs, CT 06269
>> 860-486-2200
>> 860-486-4331 (fax)
>>

[Daniel James White]

Hi Dave,

On May 28, 2011, at 6:00 AM, IMAGEJ automatic digest system wrote:

>
> Date:    Fri, 27 May 2011 16:03:46 -0400
> From:    David Knecht <[hidden email]>
> Subject: multi-channel composite images
>
> The sample data has a 5 channel composite image.  I am unsure how one creates such an image and how the colors are assigned.  If you had 5 grayscale images you wanted to merge to a composite and you want to designate the color of each channel, how would you do that?  Thanks- Dave  

Are you really sure you want to make a colour merge of 5 channels?

We have only 3 colour receptors , red , green and blue.... the primary colours,
so if you have 4 or 5 or more channels, their colour must consist of a mixture of at least 2 primary colours...
so you must confuse it with either being itself, or 2 of the other channels having signal at that point.

If the structures are very well morphologically/spatially separated, then it might not confuse the brain too much.... but....
in the usual case of diffuse staining in biological samples... its a problem.

To make things worse, a large number of men (approaching 10%) and some women are to some degree colour blind.
That means that 3 colours are already problematic.... So Red / Green is a really bad choice for 2 channel colour merges...
more folks can see Green/Magenta, or red/cyan.... but that still doesnt make it a good idea...

The basic point is that there is usually far too much information in in colour merge images for our brains to get to grips with.
We might get the most obvious features, but even these can be masked by our clever psychovisual system auto adjusting the colour
to get better colour contrast, so the same yellow does not look the same depending on what its next to.
This is super useful for finding fruit in the jungle, but gets in the way of quantitative science.

We miss most of the subtle information that is there and faithfully captured by our digital imaging devices,
but we need to visualise it in a way that our brains can comprehend and not be misled by.

(see the example script spirals in Fiji:
File - open sample - spirals

and http://www.archimedes-lab.org/color_optical_illusions.html )

If its a question or art: making a pretty cover image, then all bets are off...

To make quantitative  assessments... colour merge images could not really be any worse.
ImageJ helps us measure stuff, and so we dont need to be over reliant on our instinctive
tendency to believe what we "see". Much of what we think we see turnes out to be less than true
once you start actually measuring things.... expecially the case in colocalization analysis.

(see http://pacific.mpi-cbg.de/wiki/index.php/Colocalization_Analysis )

One might like to ask one's self:
WHY do I want to make a 5 channel colour merge image?
WHAT is it that I am "looking" for to assess?

....just my 2 cents for the benefit of the listers....
(tell me to shut up if I sound like a broken record)

cheers

Dan

>
> Dr. David Knecht    
> Department of Molecular and Cell Biology
> Co-head Flow Cytometry and Confocal Microscopy Facility
> U-3125
> 91 N. Eagleville Rd.
> University of Connecticut
> Storrs, CT 06269
> 860-486-2200
> 860-486-4331 (fax)

Dr. Daniel James White BSc. (Hons.) PhD
Senior Microscopist / Image Visualisation, Processing and Analysis
Light Microscopy and Image Processing Facilities
Max Planck Institute of Molecular Cell Biology and Genetics
Pfotenhauerstrasse 108
01307 DRESDEN
Germany

[phone] (German Mobile)
[phone] (Work phone at MPI-CBG)
[phone] (Fax MPI-CBG LMF)

http://www.bioimagexd.net  BioImageXD
http://pacific.mpi-cbg.de                Fiji -  is just ImageJ (Batteries Included)
http://www.chalkie.org.uk                Dan's Homepages
https://ifn.mpi-cbg.de  Dresden Imaging Facility Network
dan (at) chalkie.org.uk
( white (at) mpi-cbg.de )

[Michael P Ellis]

Dan, hi,

Indeed merging more than three channels results in visually ambiguous images when statically rendered (both due to the rendering technology and the ability for the viewer to resolve colours). However, we routinely use composite images of more than three colours but using an interactive ability to conveniently to adjust the channels (enable/disable/recolour etc), you can still usefully explore such the images.

The native tools in ImageJ are allow you to do this (though the UI is a little cumbersome), hence we wrote CompositeAdjuster just for this sort of situation.

http://www.youtube.com/embed/g0hoRLxBlWE?rel=0&amp;hd=1

http://www.dsuk.biz/DSUK/SmartCaptureLite.html

Regards -- Michael Ellis
Digital Scientific UK Ltd


On 30 May 2011, at 09:17, Daniel James White wrote:

> Hi Dave,
>
> On May 28, 2011, at 6:00 AM, IMAGEJ automatic digest system wrote:
>
>>
>> Date:    Fri, 27 May 2011 16:03:46 -0400
>> From:    David Knecht <[hidden email]>
>> Subject: multi-channel composite images
>>
>> The sample data has a 5 channel composite image.  I am unsure how one creates such an image and how the colors are assigned.  If you had 5 grayscale images you wanted to merge to a composite and you want to designate the color of each channel, how would you do that?  Thanks- Dave  
>
> Are you really sure you want to make a colour merge of 5 channels?
>
> We have only 3 colour receptors , red , green and blue.... the primary colours,
> so if you have 4 or 5 or more channels, their colour must consist of a mixture of at least 2 primary colours...
> so you must confuse it with either being itself, or 2 of the other channels having signal at that point.
>
> If the structures are very well morphologically/spatially separated, then it might not confuse the brain too much.... but....
> in the usual case of diffuse staining in biological samples... its a problem.
>
> To make things worse, a large number of men (approaching 10%) and some women are to some degree colour blind.
> That means that 3 colours are already problematic.... So Red / Green is a really bad choice for 2 channel colour merges...
> more folks can see Green/Magenta, or red/cyan.... but that still doesnt make it a good idea...
>
> The basic point is that there is usually far too much information in in colour merge images for our brains to get to grips with.
> We might get the most obvious features, but even these can be masked by our clever psychovisual system auto adjusting the colour
> to get better colour contrast, so the same yellow does not look the same depending on what its next to.
> This is super useful for finding fruit in the jungle, but gets in the way of quantitative science.
>
> We miss most of the subtle information that is there and faithfully captured by our digital imaging devices,
> but we need to visualise it in a way that our brains can comprehend and not be misled by.
>
> (see the example script spirals in Fiji:
> File - open sample - spirals
>
> and http://www.archimedes-lab.org/color_optical_illusions.html )
>
> If its a question or art: making a pretty cover image, then all bets are off...
>
> To make quantitative  assessments... colour merge images could not really be any worse.
> ImageJ helps us measure stuff, and so we dont need to be over reliant on our instinctive
> tendency to believe what we "see". Much of what we think we see turnes out to be less than true
> once you start actually measuring things.... expecially the case in colocalization analysis.
>
> (see http://pacific.mpi-cbg.de/wiki/index.php/Colocalization_Analysis )
>
> One might like to ask one's self:
> WHY do I want to make a 5 channel colour merge image?
> WHAT is it that I am "looking" for to assess?
>
> ....just my 2 cents for the benefit of the listers....
> (tell me to shut up if I sound like a broken record)
>
> cheers
>
> Dan
>
>
>>
>> Dr. David Knecht    
>> Department of Molecular and Cell Biology
>> Co-head Flow Cytometry and Confocal Microscopy Facility
>> U-3125
>> 91 N. Eagleville Rd.
>> University of Connecticut
>> Storrs, CT 06269
>> 860-486-2200
>> 860-486-4331 (fax)
>
> Dr. Daniel James White BSc. (Hons.) PhD
> Senior Microscopist / Image Visualisation, Processing and Analysis
> Light Microscopy and Image Processing Facilities
> Max Planck Institute of Molecular Cell Biology and Genetics
> Pfotenhauerstrasse 108
> 01307 DRESDEN
> Germany
>
> [phone] (German Mobile)
> [phone] (Work phone at MPI-CBG)
> [phone] (Fax MPI-CBG LMF)
>
> http://www.bioimagexd.net  BioImageXD
> http://pacific.mpi-cbg.de                Fiji -  is just ImageJ (Batteries Included)
> http://www.chalkie.org.uk                Dan's Homepages
> https://ifn.mpi-cbg.de  Dresden Imaging Facility Network
> dan (at) chalkie.org.uk
> ( white (at) mpi-cbg.de )

[Gabriel Landini]

On Monday 30 May 2011, Daniel James White wrote:
> To make things worse, a large number of men (approaching 10%) and some
> women are to some degree colour blind. That means that 3 colours are
> already problematic.... So Red / Green is a really bad choice for 2
> channel colour merges... more folks can see Green/Magenta, or red/cyan....
> but that still doesnt make it a good idea...

Now that you mentioned it... We think that blue-yellow is the best pair to
merge in 2 channel data because not only is discriminated by red-green colour
blinds and red-green anomalous individuals but also because their perceived
colour labels are the same as those with trichromatic vision (unlike the case
of green-magenta images, see fig 1 here):
http://www.dentistry.bham.ac.uk/cb

Regards,

Gabriel