Analyze your first slide

Analyze your first slide#

This example uses a bundled sample, requires no external slide file, and demonstrates the core LazySlide workflow.

import lazyslide as zs

wsi = zs.datasets.sample(with_data=False)
print(wsi.properties)

For your own file, replace the second line with:

wsi = zs.open_wsi("path/to/slide.svs")

Check that the dimensions, MPP, and magnification are plausible before running a physical-resolution workflow.

zs.pp.find_tissues(wsi)
zs.pl.tissue(wsi)

This adds tissue polygons under wsi.shapes["tissues"]. If the stain is faint or unusual, see Tissue detection and tiling.

zs.pp.tile_tissues(wsi, tile_px=256, mpp=0.5)
zs.pl.tiles(wsi, linewidth=0.5)

This records tile coordinates under wsi.shapes["tiles"]; it does not duplicate the complete WSI in memory.

zs.tl.feature_extraction(wsi, model="resnet50")

The default output key combines the model and tile-set names, for example resnet50_tiles. Inspect available content with:

print(wsi.shapes.keys())
print(wsi.tables.keys())

zs.pl.tiles(wsi, feature_key="resnet50", color=["1", "99"])

For persistent analysis, open the slide with a backing store and then write it:

wsi = zs.open_wsi("path/to/slide.svs", backed_file="analysis.zarr")
# Run the processing steps above.
wsi.write(overwrite=True)

The original slide remains separate; the backing store contains the analysis data. Continue with the data model, preprocessing tutorial, or How-To guides.