Stitches together any mode of serial section image data:
- TEM
- Block-face SEM
- Fluorescence Array Tomography
Fast and scalable:
- Runs on linux cluster using Sun Grid Engine API
- Align 2 to billions of images
- Approx. linear time/volume & mem/volume scaling
- Two million 4MB images align in about 8 man-hours
Handled pathologies:
- Missing tiles or whole sections
- Fragmented / small / irregular sections
- Arbitrarily rotated/translated sections
- Burns, scars, foreign matter
- Exposure inhomogeneity
Input:
- 8 or 16 bit TIF, PNG, MRC
- Simple meta-data as text or TrakEM2 XML
Output:
- Basically 1 affine or homographic transform per image tile
- Flexible output as text tables or TrakEM2 XML files
- One linear transform / tile; not an elastic aligner
- Unfinished handling of geometry-altering folds and tears
- All images in a data set must be of same fixed dimensions
Developed over several years at HHMI/Janelia Research Campus, originally by Louis Scheffer, and subsequently refined into current form by Bill Karsh. See reference "Automated Alignment of Imperfect EM Images for Neural Reconstruction".