*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hello all,

We have a number of folks who routinely test the limits of our A1-RS system by collecting for example large and high-resolution atlases of clarified brains (at least the 300 um -1 mm depth that we can reach with an inverted scope).  Our coded stage and a software feature in Elements lets us create 3D stacks of very large images stitched together from 16 to 100 individual images per slice.

We have found that the alignment can go slightly off from one slice to the next, screwing with our deconvolution and analysis results.  This is not by itself shocking; these jobs are asking an awful lot of the system.  We have the system pretty well isolated from vibration and air currents and nobody goes near it while collecting.

Elements has an alignment feature built in but it does not produce great results from very large images (e.g. 4096 ^2 and up).  I have tried the registration plugins menu in Fiji and my current favorite, StackReg, can produce solid results.  Rigid body registration seems to do the job most of the time.  On the other hand its progress bar is not very helpful and some alignments can take overnight to all weekend, if they work.

So I thought it would help to throw this out to those of you who do this more often – do you have a favorite workflow for big-job registrations?  Do you have a favorite freeware, Matlab plugin or paid option for these big jobs?  Should I stop stitching the images and let the coded stage handle alignment on its own?

We use a workstation with a decent graphics card and 36 GB of RAM.  We can put in a lot more if that will help.

Any advice on or off list would be appreciated.

Thanks and all the best,


TF

Timothy Feinstein, Ph.D. | Confocal Manager
333 Bostwick Ave., N.E., Grand Rapids, Michigan 49503
Phone: 616-234-5819 | Email: [log in to unmask]<mailto:[log in to unmask]>