RFE: alignment for medical images (DICOM etc.)

[Elaine Meng]

Are there quick methods that can be applied to overlay two medical image stacks? I tried our fitmap command on a couple of CTs but had to force quit eventually. Admittedly I only tried that one test case, without making an effort to start at a nearly correct position.

In documentation for DICOM visualization/analysis software, I see terms like "image fusion" (although used more to indicate synchronization of slice views rather than 3D) and "registration". There appear to be a number of plugins for registration in 3D Slicer, but unclear how many of them apply to 3D. Yet, if planes are in register, it seems like that is the same thing as proper registration for the whole stack of planes. See these pages for Slicer:
<​https://www.slicer.org/wiki/Slicer3:Registration>
...and one that I found for Osirix (the non-open-source cousin of Horos):
<​https://en.wikibooks.org/wiki/Online_OsiriX_Documentation/Making_fusion_images>

This is mainly to keep in the backs of our brains until a collaborator requests, feel free to reclassify/reassign.

[goddard@…]

Aligning medical scans could be a key feature.


If the 3D maps cover the same body region bounds then some quick matching of the center of mass of the density could give a good initial fit that then might refine quickly with fitmap.  Segger uses this technique.  The problem is if the two scans don't cover the same region of the person -- any extra in one image would cause a misalignment using the simple center of mass.

A more robust way might be to fitmap using a very low resolution version of the binned map, then fitmap at the full resolution, ideally combining these steps in a single toolbar button or GUI.