lazyslide.tl.virtual_stain

virtual_stain(wsi, model='rosie', image_key=None, tile_key='tiles', device=None, amp=None, autocast_dtype=None, batch_size=32, num_workers=0, pbar=True)

Translate the H&E images to multiplexed images.

A new multi-channel image will be created and stored in the WSIData object. The marker name is recorded in the image channel names.

Parameters:

wsiWSIData

The whole-slide image data to work on.

modelstr, default: “rosie”

The virtual staining model to use.

image_keystr, default: None

The key to store the new image.

tile_keystr, default: “tiles”

The key for the tile table.

devicestr, optional

Which device to use for inference. If None, the default device is used.

ampbool, optional

Whether to use automatic mixed precision.

autocast_dtypetorch.dtype, optional

The dtype for automatic mixed precision.

batch_sizeint, default: 32

The batch size for inference.

num_workersint, default: 0

The number of workers for data loading.

pbarbool, default: True

If the progress bar should be shown.

Returns:

None

The virtual stain image is added to the images slot of the WSIData object under the key '{model_name}_prediction'.

Examples

>>> import lazyslide as zs
>>> wsi = zs.datasets.sample()
>>> zs.tl.virtual_stain(wsi)
>>> wsi.images["rosie_prediction"]