Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
As for the images taken on the same scope, this is not so difficult. 'All'
you need is a sample with (preferrably subresolution) items which can be
seen in the different imaging modes. Then you need to identify the position
(e.g. the centre of mass) of the items in x and y, which usually can be done
with subpixel accuracy. Once you have the coordinates of the same features
in the different images, you can determine the affine transformation needed
for the image registration. You will have to take one image as a reference
image and determine the affine transformation for the other two. To
calculate the transformation matrix, in principle you only need three
points. The accuracy will increase, though, by using more points and
determine the 'best fit' matrix by regression.
The advantage here is that you can determine the registration
transformations from a separate sample with your alignment objects. Then you
can just apply the transforms to your actual images without the need to
embed any alingment objects in your sample.
Alternatively, you can use the location of identifiable features in your
sample to calculate the transformation matrices, provided that you can
determine their position with good precision. Subpixel accuracy is necessary
if you are planning on doing any sort of colocalisation analysis.
As for registering the image taken on the other scope, the only thing you
can do there is to either embed the alignment objects in the sample or to
use identifiable features from the sample, if there are any that are visible
in all imaging modes. Once you have corresponding locations, the mathematics
are the same. You will need to calculate the transformation for each image
though since the translation and/or rotation will be different each time.
PS: An affine transformation should be sufficient since you don't have to
deal with perspective issues. I am also assuming that your optics are of
good quality so that you don't have non-linear distortions (e.g. pincushon
or barrel distortions).
PPS: You can contact me off list if you are unsure how to calculate the
transformations.
PPPS: I am using Matlab from Mathworks for these kind of things. If you
happen to have the Matlab Image Analysis Toolbox, there is even a software
tool to register images by point-and-clicking. Of course, the alignment
accuracy is limited here by your clicking abilities.
Rereading your question I realize that you were asking what kind of sample
to use for the image registration, rather than how to perform the
registration ... Sorry, can't help you there.
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]
> >Search the CONFOCAL archive at
> >http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
> >
> >Hello All,
> >
> >We're taking some multimodal images of skin (SHG, 2P (auto),
> >reflectance, and Raman), 20-60x.
> >
> >The SHG, 2P, and reflectance are on one scope and the Raman is on
> >another (pretty cool images in all modes).
> >
> >We've reached a quandary on how to register these images (even get
> >near similar fields of view, which are between 50-250 um).
> >
> >A pen mark would be way too big. Too many hair follicles to narrow
> >down to. Latex beads will probably float away (the skin needs to
> >stay hydrated). A grid coverslip? Maybe a gene gun?
> >
> >The scopes are in different buildings, so stability of the
> "mark" IS an issue.
> >
> >Any and all input is appreciated, again.
> >Best,
> >
> >Gary
> >
> >
> >
> >Gary Laevsky, Ph.D.
> >Keck Facility Manager, CenSSIS
> >Northeastern University
> >302 Stearns
> >360 Huntington Ave.
> >Boston, MA 02115
> >Office(617) 373 - 2589
> >Lab(617) 373 - 7756
> >Fax(617) 373 - 7783
> >
> >http://www.censsis.neu.edu
> >
> >http://www.ece.neu.edu/groups/osl
> >
> >http://www.keck3dfm.neu.edu
>
> ______________________________________________________________
> ______________
> Michael Cammer Analytical Imaging Facility Albert
> Einstein Coll. of Med.
> URL: http://www.aecom.yu.edu/aif/
>
|
