deconvolution problem

[Amy Cao]

Hi,

Has anyone tried the deconvolution lab with ImageJ, with the dataset of
fluorescent beads, which is provided by
http://bigwww.epfl.ch/deconvolution/?p=bars ? I have tried, but I cannot
get the result as they shows, I don't know why. Is there anyone can help?

Thanks,

Amy

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[bnorthan]

Hi Amy

My suggestion is to try DeconvolutionLab with the following settings:

Set the Algorithm to Richardson Lucy, set the iterations to 100, make
sure you set the PSF to the correct data set, then check "Flip PSF
quadrants" (if the PSF is centered).  I find you have to be careful
and make sure the image you want to deconvolve is focused before
opening the plugin.  When the deconvolution starts watch the output in
the log window carefully and verify it is loading the correct input
and psf and flipping the psf if needed.

With these settings you should be able to get an OK result, and at
least confirm it is working.   Then you could try Richardson-Lucy with
TV regularization and/or Thresholded Landweber.  Theoretically these
algorithms should give better results, at least with noisy images but
might require more adjusting of advanced paramaters.

Brian

On Mon, Jan 14, 2013 at 3:41 PM, Amy Cao <[hidden email]> wrote:

> Hi,
>
> Has anyone tried the deconvolution lab with ImageJ, with the dataset of
> fluorescent beads, which is provided by
> http://bigwww.epfl.ch/deconvolution/?p=bars ? I have tried, but I cannot
> get the result as they shows, I don't know why. Is there anyone can help?
>
> Thanks,
>
> Amy
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Amy Cao]

Hi Brian,

Thank you so much for your suggestion, I have made it according to your
suggestion. I think my problem is I did not flip PSF. But I don't
understand why I should flip PSF, do you have some idea on it?

Thanks,

Amy

On Mon, Jan 14, 2013 at 8:27 PM, Brian Northan <[hidden email]> wrote:

> Hi Amy
>
> My suggestion is to try DeconvolutionLab with the following settings:
>
> Set the Algorithm to Richardson Lucy, set the iterations to 100, make
> sure you set the PSF to the correct data set, then check "Flip PSF
> quadrants" (if the PSF is centered).  I find you have to be careful
> and make sure the image you want to deconvolve is focused before
> opening the plugin.  When the deconvolution starts watch the output in
> the log window carefully and verify it is loading the correct input
> and psf and flipping the psf if needed.
>
> With these settings you should be able to get an OK result, and at
> least confirm it is working.   Then you could try Richardson-Lucy with
> TV regularization and/or Thresholded Landweber.  Theoretically these
> algorithms should give better results, at least with noisy images but
> might require more adjusting of advanced paramaters.
>
> Brian
>
> On Mon, Jan 14, 2013 at 3:41 PM, Amy Cao <[hidden email]> wrote:
> > Hi,
> >
> > Has anyone tried the deconvolution lab with ImageJ, with the dataset of
> > fluorescent beads, which is provided by
> > http://bigwww.epfl.ch/deconvolution/?p=bars ? I have tried, but I cannot
> > get the result as they shows, I don't know why. Is there anyone can help?
> >
> > Thanks,
> >
> > Amy
> >
> > --
> > ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[bnorthan]

Hi Amy

Yes.  The equations in the deconvolution consider the center of the
PSF to be 0,0,0.  While on the other hand in an image the center is
ussually at width/2, height/2, depth/2.   So flip PSF moves the
coordinate system to where the algorithm needs it to be.

Brian

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Amy Cao]

Hi Brian,

Is that mean all the algorithms in Deconvolution Lab require the center of
the PSF to be 0,0,0?

Sorry about this simple question, because I am not doing the algorithms
things, and I always confused about that. Is it possible for you maybe to
recommend some papers about this deconvolution so that I can get some ideas
about it?

Thank you so much,

Jianwei Cao

On Thu, Jan 17, 2013 at 6:50 AM, Brian Northan <[hidden email]> wrote:

> Hi Amy
>
> Yes.  The equations in the deconvolution consider the center of the
> PSF to be 0,0,0.  While on the other hand in an image the center is
> ussually at width/2, height/2, depth/2.   So flip PSF moves the
> coordinate system to where the algorithm needs it to be.
>
> Brian
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[bnorthan]

Hi Amy

Yes as far as I know all the algorithms in Deconvolution Lab need the
PSF to be shifted.

In terms of papers if you want something general I would recomend the
following.

Biggs, S. C., “A practical guide to deconvolution of fluorescence
microscope imagery,” Microscopy Today 18(1), 10-14 (2010).

In terms of specific details for DeconvolutionLab... they implement a
few algorithms.  Richardson Lucy is the basic one.   Richardson Lucy
with TV Regularization and Thresholded Landweber are more advanced.
They should produce better results but have more complicated input
parameters.  I tried them a couple of times before but didn't get the
result I expected.  Though I personally haven't had a chance to work
with those algorithms much and figure out the right settings.  Maybe
someone else has more experience with them??  You should keep asking
around.    The references for those algorithms are below....

Dey, N., Bland-Feraud L., Zimmer, C., Roux, P., Kam, Z., Olivo-Marin,
J.-C., and Zerubia, J., “Richardson-Lucy algorithm with total
variation regularization for 3D confocal microscope deconvolution,”
Microscopy Research and Technique, 69, 260-266 (2006).

Vonesch, C. and Unser, M., “A fast thresholded Landweber algorithm for
wavelet-regularized multidimensional deconvolution,” IEEE transactions
on image processing, 17(4), 539-549 (2008).

On Fri, Jan 18, 2013 at 3:43 PM, Amy Cao <[hidden email]> wrote:

> Hi Brian,
>
> Is that mean all the algorithms in Deconvolution Lab require the center of
> the PSF to be 0,0,0?
>
> Sorry about this simple question, because I am not doing the algorithms
> things, and I always confused about that. Is it possible for you maybe to
> recommend some papers about this deconvolution so that I can get some ideas
> about it?
>
> Thank you so much,
>
> Jianwei Cao
>
> On Thu, Jan 17, 2013 at 6:50 AM, Brian Northan <[hidden email]> wrote:
>
>> Hi Amy
>>
>> Yes.  The equations in the deconvolution consider the center of the
>> PSF to be 0,0,0.  While on the other hand in an image the center is
>> ussually at width/2, height/2, depth/2.   So flip PSF moves the
>> coordinate system to where the algorithm needs it to be.
>>
>> Brian
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Amy Cao]

Hi Brian,

Thanks for the recommendation, it really helps! I would keep on learning.

Best wishes,

Amy Cao

On Fri, Jan 18, 2013 at 4:01 PM, Brian Northan <[hidden email]> wrote:

> Hi Amy
>
> Yes as far as I know all the algorithms in Deconvolution Lab need the
> PSF to be shifted.
>
> In terms of papers if you want something general I would recomend the
> following.
>
> Biggs, S. C., “A practical guide to deconvolution of fluorescence
> microscope imagery,” Microscopy Today 18(1), 10-14 (2010).
>
> In terms of specific details for DeconvolutionLab... they implement a
> few algorithms.  Richardson Lucy is the basic one.   Richardson Lucy
> with TV Regularization and Thresholded Landweber are more advanced.
> They should produce better results but have more complicated input
> parameters.  I tried them a couple of times before but didn't get the
> result I expected.  Though I personally haven't had a chance to work
> with those algorithms much and figure out the right settings.  Maybe
> someone else has more experience with them??  You should keep asking
> around.    The references for those algorithms are below....
>
> Dey, N., Bland-Feraud L., Zimmer, C., Roux, P., Kam, Z., Olivo-Marin,
> J.-C., and Zerubia, J., “Richardson-Lucy algorithm with total
> variation regularization for 3D confocal microscope deconvolution,”
> Microscopy Research and Technique, 69, 260-266 (2006).
>
> Vonesch, C. and Unser, M., “A fast thresholded Landweber algorithm for
> wavelet-regularized multidimensional deconvolution,” IEEE transactions
> on image processing, 17(4), 539-549 (2008).
>
> On Fri, Jan 18, 2013 at 3:43 PM, Amy Cao <[hidden email]> wrote:
> > Hi Brian,
> >
> > Is that mean all the algorithms in Deconvolution Lab require the center
> of
> > the PSF to be 0,0,0?
> >
> > Sorry about this simple question, because I am not doing the algorithms
> > things, and I always confused about that. Is it possible for you maybe to
> > recommend some papers about this deconvolution so that I can get some
> ideas
> > about it?
> >
> > Thank you so much,
> >
> > Jianwei Cao
> >
> > On Thu, Jan 17, 2013 at 6:50 AM, Brian Northan <[hidden email]>
> wrote:
> >
> >> Hi Amy
> >>
> >> Yes.  The equations in the deconvolution consider the center of the
> >> PSF to be 0,0,0.  While on the other hand in an image the center is
> >> ussually at width/2, height/2, depth/2.   So flip PSF moves the
> >> coordinate system to where the algorithm needs it to be.
> >>
> >> Brian
> >>
> >> --
> >> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
> >>
> >
> > --
> > ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[mirrownight]

Hi Brian

I had the same question with Amy, too. Thank you for the suggestion about
flipping the PSF.
But I have other problems using the "Batch" module. I tried to load the
image stacks like showed in the photo:

<http://imagej.1557.x6.nabble.com/file/n5015763/Forum.png>

But failed to run deconvolution even all the images were 2D tiff.
And the dimension of the PSF was smaller than the image stacks.

Is there any file format limitation for the "Batch" module?

Thanks,


Mirrownight


bnorthan wrote

> Hi Amy
>
> Yes as far as I know all the algorithms in Deconvolution Lab need the
> PSF to be shifted.
>
> In terms of papers if you want something general I would recomend the
> following.
>
> Biggs, S. C., “A practical guide to deconvolution of fluorescence
> microscope imagery,” Microscopy Today 18(1), 10-14 (2010).
>
> In terms of specific details for DeconvolutionLab... they implement a
> few algorithms.  Richardson Lucy is the basic one.   Richardson Lucy
> with TV Regularization and Thresholded Landweber are more advanced.
> They should produce better results but have more complicated input
> parameters.  I tried them a couple of times before but didn't get the
> result I expected.  Though I personally haven't had a chance to work
> with those algorithms much and figure out the right settings.  Maybe
> someone else has more experience with them??  You should keep asking
> around.    The references for those algorithms are below....
>
> Dey, N., Bland-Feraud L., Zimmer, C., Roux, P., Kam, Z., Olivo-Marin,
> J.-C., and Zerubia, J., “Richardson-Lucy algorithm with total
> variation regularization for 3D confocal microscope deconvolution,”
> Microscopy Research and Technique, 69, 260-266 (2006).
>
> Vonesch, C. and Unser, M., “A fast thresholded Landweber algorithm for
> wavelet-regularized multidimensional deconvolution,” IEEE transactions
> on image processing, 17(4), 539-549 (2008).
>
> On Fri, Jan 18, 2013 at 3:43 PM, Amy Cao &lt;

> amycao723@

> &gt; wrote:
>> Hi Brian,
>>
>> Is that mean all the algorithms in Deconvolution Lab require the center
>> of
>> the PSF to be 0,0,0?
>>
>> Sorry about this simple question, because I am not doing the algorithms
>> things, and I always confused about that. Is it possible for you maybe to
>> recommend some papers about this deconvolution so that I can get some
>> ideas
>> about it?
>>
>> Thank you so much,
>>
>> Jianwei Cao
>>
>> On Thu, Jan 17, 2013 at 6:50 AM, Brian Northan &lt;

> bnorthan@

> &gt; wrote:
>>
>>> Hi Amy
>>>
>>> Yes.  The equations in the deconvolution consider the center of the
>>> PSF to be 0,0,0.  While on the other hand in an image the center is
>>> ussually at width/2, height/2, depth/2.   So flip PSF moves the
>>> coordinate system to where the algorithm needs it to be.
>>>
>>> Brian
>>>
>>> --
>>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>>
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
View this message in context: http://imagej.1557.x6.nabble.com/deconvolution-problem-tp5001407p5015763.html
Sent from the ImageJ mailing list archive at Nabble.com.

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[mirrownight]

Hi Brian

I had the same question with Amy, too. Thank you for the suggestion about flipping the PSF.
But I have other problems using the "Batch" module. I tried to load the image stacks like showed in the photo:




I failed to run deconvolution even all the images were 2D tiff.
And the dimension of the PSF was smaller than the image stacks.

Is there any file format limitation for the "Batch" module?

Thanks,


Mirrownight