Auto Fluorescence + 3D Deconvolution - The Beauties (part 1)

[RogelioMoreno]

I looked at AutoQuant X3 and was very impressed. Do you know how much it costs?

Ecki

I got a quotation for around $7K.

Rogelio

[arturoag75]

I looked at AutoQuant X3 and was very impressed. Do you know how much it costs?

Ecki

I got a quotation for around $7K.

Rogelio

Arturo

[Ecki]

I got a quotation for around $7K.

Ups - I guess there is no hobbyist version available. Thank you Rogelio.

[discomorphella]

Unfortunately none of the Media Cybernetics (Autoquant) nor the other commercially available programs are priced for hobbyists. I use several at work, but not at home. However, although Maximum Likelihood or Max Entropy algorithms are complex, they are also not state secrets (anymore), and there are some implementations in MATLAB (which looks very similar to C but you have to have MATLAB, also not cheap) and also in JAVA for ImageJ and some in open-source GNU licensed versions. There may be other implementations in the public domain as well. I have not tested either version, but given the mutli-thousand dollar price its probably worth trying the ImageJ plugins (free) at least.
Here are 2 references that will be of use to those with a good optics and signal processing background. I will look for more introductory materials about blind or Max Likelihood deconvolution but I am afraid that most of the basic material out there assumes that one has a decent background in both linear algebra and statistics (and optics).
This is a survey from an astronomers perspective

http://www.google.com/url?sa=t&rct=j&q= ... 7864,d.cGE

and here is a good description of an open-source 3D deconvolution package (with background maths and flowcharts)

http://onlinelibrary.wiley.com/doi/10.1 ... 3205.x/pdf

Rogelio,
You can try to find PPD maybe from a vendor such as ChemSavers or maybe even from a vendor on ebay. I don't know any chemical suppliers in Central America, but there may be someone who could order you a small quantity. There are other radical scavengers you can try as well, such as mercaptoethanol (stinky!!). I'll look up some more antifade recipes for you. We usually use media with DABCO or PPD in it.

Someday I'll have to make an introductory talk on how this works for the forum, but, well, I have to run to my SEM time now...

David

edit:

Ok, here's one using n-propyl gallate

http://www.jacksonimmuno.com/technical/anti-fade.asp

and there are some reports of Ascorbic acid, aka Vitamin C.
You can certainly obtain some pure vitamin C, and use that at millimolar concentrations.

[emsiem]

I did some web research about available deconvolution software and, true, there is no advanced packages available for home budget (except ImageJ plugins). There is Photiosity available for 4$ (Mac only), but it is not designed for microscopy and lacks advanced feature other professional packages offer. Still, it is a cheap solution and worth 3.99. Not so bad results.
I tried different trial and "trial" versions and can say that (at least for me) Huygens gives best results (compared to Autoquant and Imaris which I tested).
There is also free app for Mac at www.deconvolve.net worth a look (very fast, but almost not any parameter settings at all).

Having deconvolution option in Zerene Stacker or Helicon Focus would be dream come true ;-)

[RogelioMoreno]

David, thank you for the details.

emsiem, I will try Huygens.

Rogelio

[xavosp]

Has anybody tried Piccure (http://intelligentimagingsolutions.com/index.php)?

It is a simple deconvolution sofware advertised to get rid of camera shaking, but maybe it's worth trying it.

Have no idea if it works well for microscopy, though.

[g4lab]

I had the pleasure of taking a digital imaging course from Dr. John C. Russ
whom I have tried to lure onto this web forum.

He has suite of imaging tools which are plug ins for photoshop including deconvolution.

http://www.reindeergraphics.com/foveaprotutorial.html


http://www.reindeergraphics.com/index.html

http://www.reindeergraphics.com/science.html

The software is Fovea Pro which is a suite of photoshop plug ins
iirc the price was about $750

Their website and webstore are currently down but I have made a telephone inquiry and will update this when they reply.

http://www.drjohnruss.com/

Dr. Russ' course was so interesting, that I took it twice, and would have taken it three times if I would have been able to, the third year. He may have retired from giving courses.


If you are a skilled with software like Rik or Rogelio Fovea Pro's source code is available in this book:The Image Processing Handbook (5th edition, CRC Press, 2006, ISBN 0-8493-7254-2). as a disk. You can complile only the plug ins you might need to use (such as deconvolution which is in section seven of the first link above.

[rjlittlefield]

To clarify... Most deconvolution packages are 2D only. Their goal is to back out motion blur or defocus blur from a single image. That's much different from 3D deconvolution, which (for the purposes of this thread) essentially works by using focused information in each depth plane to remove out-of-focus haze in other planes.

It may be tempting to think that you could apply 2D deconvolution on each source frame, then run a conventional stack. But that won't actually work because the required information is missing at each step. It would be like trying to recover x from x+blur(y), without knowing what y is. 3D deconvolution is more like trying to recover x and y simultaneously from x+blur(y) [in one source frame] and blur(x)+y [in a different source frame]. If you know or assume something about the blur function, then at least you have a chance with the latter problem.

--Rik

[Charles Krebs]

Simply gorgeous images!

[g4lab]

To clarify... Most deconvolution packages are 2D only. Their goal is to back out motion blur or defocus blur from a single image. That's much different from 3D deconvolution, which (for the purposes of this thread) essentially works by using focused information in each depth plane to remove out-of-focus haze in other planes.

It may be tempting to think that you could apply 2D deconvolution on each source frame, then run a conventional stack. But that won't actually work because the required information is missing at each step. It would be like trying to recover x from x+blur(y), without knowing what y is. 3D deconvolution is more like trying to recover x and y simultaneously from x+blur(y) [in one source frame] and blur(x)+y [in a different source frame]. If you know or assume something about the blur function, then at least you have a chance with the latter problem.

I didn't even think of the 3D aspect of it when I posted that. All of Dr. Russ plug ins are of course 2D. He does have a Z stacking plug in but it is not the deconvolution plug in which is indeed 2D. So it would help on a single fluorescence frame but not on a stack.

Thank you for pointing that distinction out.

[descall]

Absolutely stunning! Very nice indeed.

If you get hold of any other filter cubes it would be nice to see some comparisons with different cubes.

Do you know if the non-fluorescent black dots scattered over the chlorophyll regions of the desmids are where the pyrenoids are located? I wonder what the greenish dots are on the 4th image?

Wonderful stuff.

[RogelioMoreno]

Thank you for your comments.

Des, I wish I have more cubes to try. I also wonder about the black and greenish dots are.

Rogelio