Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal

You are right, it is also our experience that you can increase the overlay
accuracy by a more complex transformation than a simple translation. In our
case it is in a non-confocal application, but the principle stays the same.
By using a general affine transformation instead of a simple translation we
can increase the average overlay accuracy from approx. 0.5 micron to approx.
10 nm (this is in 2D). You could even use more complex transformations
(projective or even non-linear ones), but 10 nm is around the accuracy with
which we can determine the (Gaussian) centres of fluorescent beads.

(An affine transformation is a general combination of translation, scaling
along x, scaling along y, rotation, shear (i.e. transformation of angles),
and reflection along x and/or y.)

Before the actual experiments we take a dual channel image of Tetra Speck
beads. An image processing program we wrote in Matlab then calculates the
affine transformation parameters based on the (Gaussian) centres of the
beads. Finally, the dual channel movies are automatically overlayed using
those transformation parameters.

If you like you can send me a dual channel image of the Tetra Speck beads
(at least 4 beads in the image as widely spread as possible) and I could
check by how much you can improve the accuracy by using the more general
afiine transformation (maybe your confocal is already good as it is and a
translation is all you need).

Also, I think about 2 weeks ago someone on the list asked about the
registration accuracy of other confocal users. Maybe this person meanwhile
can share the information obtained from other confocalists what is the
overlay accuracy that is generally obtained on confocal microscopes?

Best regards,

Kevin


Kevin Braeckmans, Ph.D.
Lab. General Biochemistry & Physical Pharmacy
Ghent University
Harelbekestraat 72
9000 Ghent
Belgium
Tel: +32 (0)9 264.80.78
Fax: +32 (0)9 264.81.89
E-mail: [log in to unmask]


> -----Oorspronkelijk bericht-----
> Van: Confocal Microscopy List 
> [mailto:[log in to unmask]] Namens Stanislav Vitha
> Verzonden: vrijdag 22 september 2006 21:34
> Aan: [log in to unmask]
> Onderwerp: 3D registration of two channels
> 
> Search the CONFOCAL archive at
> http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
> 
> Dear confocalists,
> one of our confocal users is attempting colocalization 
> analysis in double immunostained fly brains. The subcellular 
> features of interest are very small (~1 um), so good 
> registration of the two image channels (Alexa Fluor
> 488 and Alexa Fluor 633 fluorochromes) will be crucial. 
> Besides optimizing the imaging condition, the plan is to 
> perform deconvolution of the z- stacks for each channel, and 
> also to incorporate sub-resolution fluorescent beads (such as 
> the Tetraspeck beads from Invitrogen) as a positive 
> colocalization control and a fiducial marker. My concern is 
> that according to my tests, even with this hand-picked 
> PlanApo optics, there is not only a residual chromatic 
> aberration (longitudinal), but also a slight chromatic 
> difference of magnification and a difference in the curvature 
> of the field. So simply shifting the two z-stacks relative to 
> each other in X, Y, and Z will not result in perfect 
> registration. Is there any software that could use the bead 
> signal as fiducial markers and perform 3D warping 
> of one stack to bring it into registration with the second 
> z-stack?     
> 
> Thaks in advance.
> 
> Stan Vitha
> Microscopy and Imaging Center
> Texas A & M University
>