NWB-ify individual neurons: image stack, swc, others?

reconstrue / brightfield

Reconstruction of biocytin-stained neurons detected in brightfield microscopy image stacks

http://reconstrue.com

Apache License 2.0

2 stars 1 forks source link

NWB-ify individual neurons: image stack, swc, others? #81

Open JohnTigue opened 4 years ago

JohnTigue commented 4 years ago

As I play with various tools and sample data, there seems to not be a standard for representing/manifesting brighfield image stacks. M

[ ] See also #67
[ ] Probe the cache to see what they do with the SWCs.
[ ] Find out where that would belong in the HDF5 file.

JohnTigue commented 4 years ago

How about HFD5 for the core/manifest but keep the files in the VMs FS a la Exdir (Exdir docs).

Maybe the Allenauts has a goal of shoving everything in the HDF5 file, but still this would be NWB:N 2.0 even if the files are external to the HDF5 file, right? Yes, I think: hdf5-exdir-converter.

JohnTigue commented 4 years ago

Try this: a HDF5 file just for metadata, pointing to external image files, structured according to Exdir.

Features:

NWB compliant
Exdir for structure of dataset
Folks who don't care about NWB can still get at the data

Need to work out the image stack standard and labels for them s.t. the NWB/HDF5 contains metadata saying, "the stack is over there, and is structured/typed as per standard Foo."