natverse / hemibrainr

Code for working with data from Janelia FlyEM's hemibrain project
https://natverse.github.io/hemibrainr/
7 stars 4 forks source link

hemibrain_reroot doesn't work with hemibrain.surf #4

Open jefferis opened 4 years ago

jefferis commented 4 years ago

Unfortunately hemibrain_reroot doesn't work with the new hemibrain surface mesh. I'm not sure if it is expanded with respect to the underlying meshes. For FAFB I have used the signed distance to help with soma identification rather than assuming that they have to be outside the mesh. Not sure if that could help here.

bad.soma = c("5813015982","885788485","5813013913")
# Read in these neurons
neurons.bs = neuprintr::neuprint_read_neurons(bad.soma)
# Re-root
neurons = hemibrain_reroot(neurons.bs, meshes = hemibrain.surf)
alexanderbates commented 4 years ago

I can just deflate the mesh ... @schlegelp do you remember how much you inflated its constituent ROIs?

schlegelp commented 4 years ago

In theory I reversed all dilations I did. I'm guessing the increase in size is the result of the initial downsampling (voxelsize = 1 micron) and the final remeshing to reduce the complexity.

I'm afraid the increase in size is also not uniform - otherwise I would have tried to compensate straight away.

jefferis commented 4 years ago

A note that this is now improved with the updated hemibrain.surf in 8ba96b7 but one of the three neurons above still fails (for reasons that aren't completely clear to me).