napari segmentation example

[will-moore]

This uses code from https://napari.org/stable/tutorials/segmentation/annotate_segmentation.html

I have tested this workflow with an RGB image from IDR, exported with Batch_Image_Export, but you can also simply save https://idr.openmicroscopy.org/webclient/render_image/179764/0/0/ and import it...

Then, using napari-omero open it in napari with $ omero napari view Image:ID

Copy the code from the link above - you just need the imports (maybe not all needed) and the segment() method.
Specifically I copied this code and pasted it into napari terminal...

import numpy as np
from skimage import data
from skimage.filters import threshold_otsu
from skimage.segmentation import clear_border
from skimage.measure import label, regionprops_table
from skimage.morphology import closing, square, remove_small_objects
import napari


def segment(image):
    """Segment an image using an intensity threshold determined via
    Otsu's method.

    Parameters
    ----------
    image : np.ndarray
        The image to be segmented

    Returns
    -------
    label_image : np.ndarray
        The resulting image where each detected object labeled with a unique integer.
    """
    # apply threshold
    thresh = threshold_otsu(image)
    bw = closing(image > thresh, square(4))

    # remove artifacts connected to image border
    cleared = remove_small_objects(clear_border(bw), 20)

    # label image regions
    label_image = label(cleared)

    return label_image

Then you can use the first "Red" layer from the RGB, slice it to 2D (required by code above) and compute() to turn it from a dask array into a numpy array, which also seems to be needed.

image = viewer.layers[0].data
image.shape
>>> (1, 1, 1024, 1344)
plane = image[0, 0]
plane = plane.compute()

Now we can use the segment() method we created above to give us labels...
Then we add back 2 dimensions to get back to 4D (shape is (1, 1, 1024, 1344) again) which seems to be needed for saving to OMERO.

labels = segment(plane)
lab3d = np.expand_dims(labels, axis=0)
lab4d = np.expand_dims(lab3d, axis=0)

Finally, we can add the labels to the viewer to see them in napari...

viewer.add_labels(lab4d)

Then you can hit Save ROIs to OMERO and you should see the labels saved as new Masks in OMERO: