wsidata.IterAccessor

class IterAccessor(obj)

Bases: object

An accessor to iterate over the WSI data.

Usage: wsidata.iter

The iter accessor will return a data container, you can access the data within container using attributes. All container has .plot() method to visualize its content.

tile_images(key, color_norm=None, format='yxc', image_size=None, annot_key=None, annot_names=None, annot_labels=None, mask_dtype=None, shuffle=False, sample_n=None, seed=0)

Extract tile images from the WSI.

Parameters:

keystr

The tile key.

color_normstr, {“macenko”, “reinhard”}, default: None

Color normalization method.

formatstr, {“yxc”, “cyx”}, default: “yxc”

The channel format of the image.

image_sizeint or tuple of (int, int), optional

The desired output image size. If tile_spec exists, this overrides the output size. If tile_spec does not exist, this is used when deriving tiles from shapes on the fly.

annot_keystr, default: None

The key to the annotation table in shapes slot.

annot_namesstr or array of str, default: None

The name of the annotation column in the annotation table or a list of names.

annot_labelsstr or dict, default: None

The name of the label column in the annotation table or a dictionary of label mappings.

mask_dtypenp.dtype, default: np.uint8

The data type of the annotation mask, if you have annotation labels more than 255, consider using data type with higher bound.

shufflebool, default: False

If True, return tile images in random order.

sample_nint, default: None

The number of samples to return.

seedint, default: 0

The random seed.

Returns:

TileImage

tissue_contours(key, shuffle=False, seed=0)

A generator to extract tissue contours from the WSI.

Parameters:

keystr

The tissue key.

shufflebool, default: False

If True, return tissue contour in random order.

seedint, default: 0

Returns:

TissueContour

tissue_images(key, level=-1, mask_bg=False, color_norm=None, format='yxc', shuffle=False, seed=0, image_size=None)

Extract tissue images from the WSI.

Parameters:

keystr

The tissue key.

levelint, default: 0

The level to extract the tissue images.

mask_bgbool | int, default: False

Mask the background with the given value.

If False, the background is not masked.

If True, the background is masked with 0.

If an integer, the background is masked with the given value.

color_normstr, {“macenko”, “reinhard”}, default: None

Color normalization method.

formatstr, {“yxc”, “cyx”}, default: “yxc”

The channel format of the image.

Returns:

TissueImage