write_probability_heatmap_as_ome_tiff¶

write_probability_heatmap_as_ome_tiff(image_path, probability, tile_size=(64, 64), levels=2, mpp=(0.25, 0.25), colormap=2)[source]¶

Saves output probability maps from segmentation models as heatmaps.

This function converts the probability maps from individual classes to heatmaps and saves them as pyramidal ome tiffs.

Parameters:

  • image_path (Path) – File path (including extension) to save image to.

  • probability (np.ndarray or zarr.core.Array) – The input image data in YXC (Height, Width, Channels) format.

  • tile_size (tuple) – Tile/Chunk size (YX/HW) for writing the tiff file. Only allows tile shapes allowed by tifffile. Default is (64, 64).

  • levels (int) – Number of levels for saving pyramidal ome tiffs. Default is 2.

  • mpp (tuple[float, float]) – Tuple of mpp values in y and x (YX/HW). Default is (0.25, 0.25).

  • colormap (int) – Colormap to save the heatmaps. Default is 2 (cv2.COLORMAP_JET).

Raises:

  • TypeError – If the input img is not a NumPy or Zarr array or does not have 3 dimensions.

  • ValueError – If input dimensions is not 3 (HWC) dimensions.

Return type:

None

Examples

>>> probability_map = imread("path/to/probability_map")
>>> write_probability_heatmap_as_ome_tiff(
... image_path=image_path,
... probability=probability_map,
... tile_size=(64, 64),
... class_name="tumor",
... levels=2,
... mpp=(0.5, 0.5),
... colormap=cv2.COLORMAP_JET,
... )