Registering 3D image Sequence - Stabilizing a point of interest

Hello - this is my first post so let me know if I’m not doing this right!

I have collected 4D image sequences of neuronal process growth in developing C. elegans embryos. I would like to register the sequence so that I can visualize growth of the axon - ie have a specific neuron be in a fixed position while the rest of the image is rotated and translated to keep that neuron in position. The embryo moves a lot, and changes as it grows, so I’m not sure if standard 3d image registration would work. I think I can do this with manual landmark selection in imagej, by selecting landmarks that are in or near my neuron of interest, but I haven’t found a plugin that makes it streamlined enough to be feasible for large datasets. Even better would be something somewhat automated that can register the neuron of interest despite the significant displacement and deformation of the rest of the volume. I imagine fully automated would be difficult, but maybe something where the user specifies a 3D ROI that is used for registration of the volumes?

Here is a link containing 2 such volumes to give you an idea of the task:

https://drive.google.com/drive/folders/1cMYeVowHNuaNfDStNlPPZvCuGZ7Pk3ie?usp=sharing

Any advice would be appreciated! Thank you,
Raffi

2 Likes

Hi @Rcohn

Have you tried the Correct 3D Drift PlugIn for ImageJ? The plugin allows for drift calculations to be restricted to an ROI that moves with the structure of interest. Possibly this could function for your limited registration automation. I quickly took a stab by doing a local background subtraction (~50 pixel) and an roi limited 3d registration correction (~40 pixel bubble around middle most cell). Attached are the kinda rough results as a temporal color coded 3d projection.

BckSub_NoReg_TempColorCodespectrum.tif (6.9 MB)

BckSub_ReReg_TempColorCodespectrum.tif (7.1 MB)

Hopefully you can see in the reregistered example there’s a central cell that is white (present in all/most time points).

Hi Mcarvin, thanks for the suggestion. Unfortunately I just saw it now - but I will look into it! Thanks!