Coordinate warping of microscope movement

[simomounir]

So i have a use-case and I tried to used the warp_points.py example.

use case is: we image the sample in tiles and then the standard is to stitch the adjacent tiles. the tiles are numbered from 1 to N and each tile has an X,Y coordinate pair. I have a stitched Ref image and another image to be aligned to it.

I do the alignment and get /pickle as result. The tile coordinates (in millimeter) are associated to the Ref image.

This is how i proceed:

#########################################################

Load a registrar that has been saved

registrar_f = "registrar.pickle"
registrar = registration.load_registrar(registrar_f)
COORD_LEVEL = 0 # pyramid level from which the ROI coordinates originated. Usually 0.

Get slide from which the ROI coordinates originated

pt_source_img_f = "Ref.tiff"
pt_source_slide = registrar.get_slide(pt_source_img_f)

Convert coordinates to pixel units

um_per_px = pt_source_slide.reader.scale_physical_size(COORD_LEVEL)[0:2]
new_xy_px = bbox_xy_um/np.array(um_per_px)

Warp coordinates to position in registered slides

xy_in_registered_img = pt_source_slide.warp_xy(new_xy_px,
slide_level=COORD_LEVEL,
pt_level=COORD_LEVEL)

####################

I want to get the transformed coordinates that I input to the microscope and get tiles aligned with Ref.

-should i then put the Other/moving image as reference/fixed image? or should i inverse the transform and keep the Ref.tiff as fixed/reference?

• is the transformation to stage physical coordinate done right?
  -i assume also only rigid should be done.