Plot all masks types in napari

[berombau]

Currently, a lot of different cell annotation masks are generated and visualized in the notebooks using matplotlib or sq.pl.spatial_segment:

1. instance mask from the cellpose segmentation step (every cell a unique ID)

• from which we calculate GeoPandas polygons, do some filtering and can again rasterize to a filtered instance mask

2. filtered instance mask (every cell a unique ID)
3. filtered semantic mask annotated by cell cluster after allocation step and clustering (every cell a cluster ID)
4. filtered semantic mask annotated by a cell type after annotation step (every cell a cell type annotation)

The colors for instance masks and cell clusters should be random using the default colors.
The colors for the cell annotation should come from an upfront defined colormap, so each cell type has a stable color.

napari-spatialdata currently supports visualization of a categorical observation for each cell, but using a colored Points layer.

A napari Labels layer should be created for each mask type and with the correct face_color:

• instance mask
• filtered instance mask
• filtered semantic cluster mask
• filtered semantic annotation mask

[berombau]

fyi, current Points layer in napari-spatialdata, which does show a 'legend' on top using the Points layer. The Labels layer should show a name of the cell cluster or annotation type when hovering using the features or property functionality.

[lopollar]

To switch fro mask to shapes that you can plot in napari:

coords_for_napari=[]
for i in adatap.obsm['polygons'].geometry.exterior:
if i:
i=i.simplify(tolerance=2)
coords_for_napari.append(np.array(i.coords)[:,[1,0]])

[lopollar]

viewer = napari.view_image(I, rgb=False)
viewer.add_shapes(coords_for_napari,name='voronoi',shape_type='polygon')

[lopollar]

This issue is resolved for non-spatial data plotting, and can be asked for by Lotte, it will still need some adaptation to work with spatialData