Puncta quantitation

[Zachary Freyberg]

Hi,
I have been visualizing nerve terminal dynamics using fluorophores  
concentrated at the terminals and visualized by microscopy.  In the  
process, I've been making movies where I capture several ROI's of  
puncta simulatneously where each puncta represents a nerve terminal.  
These puncta disappear or diminish in intensity over time at differing  
rates. I'd like to use ImageJ to quantify the change in punctal  
intensity from the initial frame (pre-treatment condition) compared to  
the last frame (where intensity has variably diminished depending on  
the ROI based on whatever treatment the nerve terminals have  
undergone).  What would be the optimal way of doing so? Given that I'm  
looking at several different populations of puncta with their own  
rates of intensity diminishment, ideally, it would be great to graph  
the intensity changes as a distribution. Also, what would be the  
optimal means of selecting thresholds to identify as many of the  
puncta within an ROI as possible?

Another problem is that, given all of the choices in ImageJ: do I  
deconvolve the frames and, if so, will this help the program to better  
identify puncta? Should this be helpful, what would be the best method  
of deconvolution in ImageJ?

Any and all help is much appreciated!

Thank you again,
Zachary Freyberg MD, PhD

[Gabriel Lapointe-2]

On Wed, 2010-03-31 at 16:08 -0400, Zachary Freyberg wrote:

> Hi,
> I have been visualizing nerve terminal dynamics using fluorophores  
> concentrated at the terminals and visualized by microscopy.  In the  
> process, I've been making movies where I capture several ROI's of  
> puncta simulatneously where each puncta represents a nerve terminal.  
> These puncta disappear or diminish in intensity over time at differing  
> rates. I'd like to use ImageJ to quantify the change in punctal  
> intensity from the initial frame (pre-treatment condition) compared to  
> the last frame (where intensity has variably diminished depending on  
> the ROI based on whatever treatment the nerve terminals have  
> undergone).  What would be the optimal way of doing so? Given that I'm  
> looking at several different populations of puncta with their own  
> rates of intensity diminishment, ideally, it would be great to graph  
> the intensity changes as a distribution.

It depends on what your looking at but, assuming that your puncta don't
move, you could threshold them to create a mask and then use it with the
Analyse Particles... function on all your images. All puncta will keep
the same number so it should be straight forward to graph the intensity
in a spreadsheet.

> Also, what would be the  
> optimal means of selecting thresholds to identify as many of the  
> puncta within an ROI as possible?

I study RNA transport granules and to identify as many puncta as
possible I really like the sliding threshold
You can find a plugin here:
http://gabriellapointe.ca/imagej/plugins/sliding.php 

>
> Another problem is that, given all of the choices in ImageJ: do I  
> deconvolve the frames and, if so, will this help the program to better  
> identify puncta? Should this be helpful, what would be the best method  
> of deconvolution in ImageJ?


It depends on your images but deconvolution can always help.
Unfortunately I haven't try extensively ImageJ's deconvolution plugins
since I'm working with a LSCM and I can't figure out how to compute good
confocal theorical PSF.

good luck,

Gabriel Lapointe, MSc.
Laboratoire de Luc DesGroseillers, PhD.
Pavillon Roger-Gaudry Local A-538
Département de biochimie
Faculté de Médecine de l'Université de Montréal
2900 boul. Édouard-Montpetit,
Montréal, Qc, H3T 1J4
Tel : [phone] postes 5187, 5152, 5162 ou 1048
Fax : [phone]
[hidden email]
http://gabriellapointe.ca 

[Robert Dougherty]

Gabriel,

>>
>> Another problem is that, given all of the choices in ImageJ: do I  
>> deconvolve the frames and, if so, will this help the program to better  
>> identify puncta? Should this be helpful, what would be the best method  
>> of deconvolution in ImageJ?
>
>
> It depends on your images but deconvolution can always help.
> Unfortunately I haven't try extensively ImageJ's deconvolution plugins
> since I'm working with a LSCM and I can't figure out how to compute good
> confocal theorical PSF.

Have you tried squaring a regular PSF?

Bob


Robert Dougherty, Ph.D.
President, OptiNav, Inc.
4176 148th Ave. NE
Redmond, WA 98052
[phone]
FAX [phone]
www.optinav.com
[hidden email]

[Francis Burton]

At 04:46 01/04/2010, Robert Dougherty <[hidden email]> wrote:

>>> Another problem is that, given all of the choices in ImageJ: do I  
>>> deconvolve the frames and, if so, will this help the program to better  
>>> identify puncta? Should this be helpful, what would be the best method  
>>> of deconvolution in ImageJ?
>>
>>
>> It depends on your images but deconvolution can always help.
>> Unfortunately I haven't try extensively ImageJ's deconvolution plugins
>> since I'm working with a LSCM and I can't figure out how to compute good
>> confocal theorical PSF.
>
>Have you tried squaring a regular PSF?

Bob,

Do you mean raising to the power of 2 or making square? What is the
rationale for doing that?

Francis

[Robert Dougherty]

On Mar 31, 2010, at 11:08 PM, Francis Burton  
<[hidden email]> wrote:

> At 04:46 01/04/2010, Robert Dougherty <[hidden email]> wrote:
>>>> Another problem is that, given all of the choices in ImageJ: do I
>>>> deconvolve the frames and, if so, will this help the program to  
>>>> better
>>>> identify puncta? Should this be helpful, what would be the best  
>>>> method
>>>> of deconvolution in ImageJ?
>>>
>>>
>>> It depends on your images but deconvolution can always help.
>>> Unfortunately I haven't try extensively ImageJ's deconvolution  
>>> plugins
>>> since I'm working with a LSCM and I can't figure out how to  
>>> compute good
>>> confocal theorical PSF.
>>
>> Have you tried squaring a regular PSF?
>
> Bob,
>
> Do you mean raising to the power of 2 or making square? What is the
> rationale for doing that?
>
> Francis

Francis,
Raise to the power 2. The rational is that the light passes through  
the optics of a confocal microscope twice. The regular computed PSF  
(there must be a better name) represents the light from the scanned  
source point reaching the object. Multiplying this function by the  
regular computed PSF accounts for the return trip through the lens to  
the detector. This idea was suggested to me by a user of my PSF and  
deconvolution plugins some time ago, and I have not found a reason to  
dispute it.
Bob

[dpoburko]

Hi Zachary,

    In what is probably a very similar approach as Gabriel LaPointe's
Slidethreshold we use a macro we call Multiple_thresholds (
http://rsb.info.nih.gov/ij/macros/Mulitple_Thresholds.txt ) to select
axonal boutons presenting evoked  vessicle fusion reported by pHluorins.
This also works pretty well for non-dynamic probes like VAMP2-mCherry. I
haven't tried Gabriel's method, but I am assuming that these two methods
will give pretty similar results.
   In any case, once you have a method for unbiased selection of ROIs
around boutons, then it's easier just to extract the temporal plots to
Excel or some other software that's better suited for plotting traces.

Cheers,
Damon





On 3/31/2010 1:08 PM, Zachary Freyberg wrote:

> Hi,
> I have been visualizing nerve terminal dynamics using fluorophores
> concentrated at the terminals and visualized by microscopy.  In the
> process, I've been making movies where I capture several ROI's of
> puncta simulatneously where each puncta represents a nerve terminal.  
> These puncta disappear or diminish in intensity over time at differing
> rates. I'd like to use ImageJ to quantify the change in punctal
> intensity from the initial frame (pre-treatment condition) compared to
> the last frame (where intensity has variably diminished depending on
> the ROI based on whatever treatment the nerve terminals have
> undergone).  What would be the optimal way of doing so? Given that I'm
> looking at several different populations of puncta with their own
> rates of intensity diminishment, ideally, it would be great to graph
> the intensity changes as a distribution. Also, what would be the
> optimal means of selecting thresholds to identify as many of the
> puncta within an ROI as possible?
>
> Another problem is that, given all of the choices in ImageJ: do I
> deconvolve the frames and, if so, will this help the program to better
> identify puncta? Should this be helpful, what would be the best method
> of deconvolution in ImageJ?
>
> Any and all help is much appreciated!
>
> Thank you again,
> Zachary Freyberg MD, PhD

[Joachim Wesner]

Hi Robert and Francis,

almost exactly. The single (unsquared) intensity PSF would apply for a
"wide-field" illuminated point object, not for a confocal microscope.
It´s not really that the same "rays" pass twice through the same regions of
the lens (as is for ex. the case in interferometric testing of an
objective lens) where you will need to square of the *COMPLEX* (phase and
intensity) PSF, but the detected intensity in a confocal microscope
for off-axis points is reduced for two reasons:

1) The point in question is not as strongly illuminated as if centered, as
determined by the objective NA and illumination wavelength

2) Any light scattered back (or fluorecent light) will be detected less
intense for off axis sources determined by NA, tube lens focal length and
pinhole diameter as also detection wavelength

For the most simplest case this is equivalent to the "one-pass" intensity
PSF squared.

Mit freundlichen Grüßen / Best regards

Joachim Wesner
Projektleiter Optik Technologiesysteme

Leica Microsystems CMS GmbH | GmbH mit Sitz in Wetzlar | Amtsgericht
Wetzlar  HRB 2432
Geschäftsführer:  Dr. Stefan Traeger | Dr. Wolf-Otto Reuter | Dr. David Roy
Martyr | Colin Davis
www.leica-microsystems.com



                                                                           
             Robert Dougherty                                              
             <[hidden email]>                                            
             Gesendet von:                                              An
             ImageJ Interest            [hidden email]                
             Group                                                   Kopie
             <[hidden email].                                            
             GOV>                                                    Thema
                                        Confocal PSF (was Puncta          
                                        quantitation)                      
             01.04.2010 16:32                                              
                                                                           
                                                                           
              Bitte antworten                                              
                    an                                                    
              ImageJ Interest                                              
                   Group                                                  
             <[hidden email].                                            
                   GOV>                                                    
                                                                           
                                                                           




On Mar 31, 2010, at 11:08 PM, Francis Burton
<[hidden email]> wrote:

> At 04:46 01/04/2010, Robert Dougherty <[hidden email]> wrote:
>>>> Another problem is that, given all of the choices in ImageJ: do I
>>>> deconvolve the frames and, if so, will this help the program to
>>>> better
>>>> identify puncta? Should this be helpful, what would be the best
>>>> method
>>>> of deconvolution in ImageJ?
>>>
>>>
>>> It depends on your images but deconvolution can always help.
>>> Unfortunately I haven't try extensively ImageJ's deconvolution
>>> plugins
>>> since I'm working with a LSCM and I can't figure out how to
>>> compute good
>>> confocal theorical PSF.
>>
>> Have you tried squaring a regular PSF?
>
> Bob,
>
> Do you mean raising to the power of 2 or making square? What is the
> rationale for doing that?
>
> Francis

Francis,
Raise to the power 2. The rational is that the light passes through
the optics of a confocal microscope twice. The regular computed PSF
(there must be a better name) represents the light from the scanned
source point reaching the object. Multiplying this function by the
regular computed PSF accounts for the return trip through the lens to
the detector. This idea was suggested to me by a user of my PSF and
deconvolution plugins some time ago, and I have not found a reason to
dispute it.
Bob



______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________

[Robert Dougherty]

Joachim,

I took the opportunity to think about this some more.  My PSF plugin works by analyzing the optical propagation between aperture of the objective lens and the object point.  it actually was developed for the reciprocal propagation case.  It assumes the aperture is uniformly illuminated and phased up to focus on the central point. Using this source distribution, it computes the complex field that would occur at each location in the image stack surrounding the point, and takes the magnitude-squared to produce the PSF of a diffraction limited wide-field microscope in the Fraunhofer approximation.  In FFT-based deconvolution, this PSF is assumed to be translationally invariant; it is assumed that if the source point were moved, then the entire PSF would move with it without changing shape.  This is a good approximation, at least in the transverse directions, when the FOV subtends a small angle.  Now consider a confocal microscope.  As I understand it, the confocal microscope images a point by first creating the  situation in my reciprocal simulation: it distributes light across the aperture so as to focus on the point.  The computed PSF therefore describes the illumination, assuming the microscope is diffraction limited.  This light then scatters incoherently from the object and is imaged, so the PSF applies a second time.  The claim is that the idea of squaring the wide-field PSF to simulate a confocal microscope is exact, assuming the microscope is diffraction limited (in both directions) and the FOV is small enough for the shift-invariant assumption.  In this view, your points 1) and 2) are shortcomings of the microscope, not the analytical procedure.  Am I missing something?
I see what you mean about squaring the complex field to account for a specular-reflecting object.  It is probably not a good to idea image an electron-microscope sample with a light microscope.  Would the very bright reflection damage something?
For the fluorescent imaging, maybe the right thing to do is compute two PSFs, one for the wavelength of the illumination and one for the wavelength of the emitted light, and take the product of these two functions.

Bob


On Apr 6, 2010, at 9:26 AM, Joachim Wesner wrote:

> Hi Robert and Francis,
>
> almost exactly. The single (unsquared) intensity PSF would apply for a
> "wide-field" illuminated point object, not for a confocal microscope.
> It´s not really that the same "rays" pass twice through the same regions of
> the lens (as is for ex. the case in interferometric testing of an
> objective lens) where you will need to square of the *COMPLEX* (phase and
> intensity) PSF, but the detected intensity in a confocal microscope
> for off-axis points is reduced for two reasons:
>
> 1) The point in question is not as strongly illuminated as if centered, as
> determined by the objective NA and illumination wavelength
>
> 2) Any light scattered back (or fluorecent light) will be detected less
> intense for off axis sources determined by NA, tube lens focal length and
> pinhole diameter as also detection wavelength
>
> For the most simplest case this is equivalent to the "one-pass" intensity
> PSF squared.
>
> Mit freundlichen Grüßen / Best regards
>
> Joachim Wesner
> Projektleiter Optik Technologiesysteme
>
> Leica Microsystems CMS GmbH | GmbH mit Sitz in Wetzlar | Amtsgericht
> Wetzlar  HRB 2432
> Geschäftsführer:  Dr. Stefan Traeger | Dr. Wolf-Otto Reuter | Dr. David Roy
> Martyr | Colin Davis
> www.leica-microsystems.com
>
>
>
>
>             Robert Dougherty                                              
>             <[hidden email]>                                            
>             Gesendet von:                                              An
>             ImageJ Interest            [hidden email]                
>             Group                                                   Kopie
>             <[hidden email].                                            
>             GOV>                                                    Thema
>                                        Confocal PSF (was Puncta          
>                                        quantitation)                      
>             01.04.2010 16:32                                              
>
>
>              Bitte antworten                                              
>                    an                                                    
>              ImageJ Interest                                              
>                   Group                                                  
>             <[hidden email].                                            
>                   GOV>                                                    
>
>
>
>
>
>
> On Mar 31, 2010, at 11:08 PM, Francis Burton
> <[hidden email]> wrote:
>
>> At 04:46 01/04/2010, Robert Dougherty <[hidden email]> wrote:
>>>>> Another problem is that, given all of the choices in ImageJ: do I
>>>>> deconvolve the frames and, if so, will this help the program to
>>>>> better
>>>>> identify puncta? Should this be helpful, what would be the best
>>>>> method
>>>>> of deconvolution in ImageJ?
>>>>
>>>>
>>>> It depends on your images but deconvolution can always help.
>>>> Unfortunately I haven't try extensively ImageJ's deconvolution
>>>> plugins
>>>> since I'm working with a LSCM and I can't figure out how to
>>>> compute good
>>>> confocal theorical PSF.
>>>
>>> Have you tried squaring a regular PSF?
>>
>> Bob,
>>
>> Do you mean raising to the power of 2 or making square? What is the
>> rationale for doing that?
>>
>> Francis
>
> Francis,
> Raise to the power 2. The rational is that the light passes through
> the optics of a confocal microscope twice. The regular computed PSF
> (there must be a better name) represents the light from the scanned
> source point reaching the object. Multiplying this function by the
> regular computed PSF accounts for the return trip through the lens to
> the detector. This idea was suggested to me by a user of my PSF and
> deconvolution plugins some time ago, and I have not found a reason to
> dispute it.
> Bob
>
>
>
> ______________________________________________________________________
> This email has been scanned by the MessageLabs Email Security System.
> For more information please visit http://www.messagelabs.com/email 
> ______________________________________________________________________

Robert Dougherty, Ph.D.
President, OptiNav, Inc.
4176 148th Ave. NE
Redmond, WA 98052
Tel. [phone]
FAX [phone]
www.optinav.com
[hidden email]