Export Webknossos annotations to OME-Zarr or NIfTI-Zarr

[kabilar]

Description

Currently Webknossos annotations are in the *.wkw format. We should create scripts to export and convert these annotations to either OME-Zarr or NIfTI-Zarr. We have decided to select the output format to match the base imaging layer(s) so that the orientation is preserved. We can use either the wkcuber or webknossos Python libraries.

Goals

• Determine additional metadata in the Webknossos annotation files (e.g. annotator name, mapping of segmentation names)
• Create a script to perform the following operations:
  • Determine the base imaging layer(s) format of the corresponding Webknossos dataset (i.e. OME-Zarr or NIfTI-Zarr)
  • Export Webknossos annotation as a OME-Zarr file
  • Export Webknossos annotation as a NIfTI-Zarr file
  • Add the above Webknossos metadata to the exported file
  • Note: There is a Zarr webknossos/neuroglancer incompatibility. Webknossos compresses colors in layers and Neuroglancer compresses color with chunks.

Thank you.

[jingjingwu1225]

Hi @kabilar, I noticed a zarr link option avaliable for annotations on webknossos, it generates a url link for selected layer( layer volume for annotations), the zarr file has a ome.zarr structure but follows a cxyz direction. I calculated the border distance for each slice at a low-resolution level and get corresponding offset at other resolution levels. And for metadata, we could just download the skeleton annotations as nml format and save essential information into metadata of mask.ome.zarr file.