Load ome.zarr Image with labels from a public S3 repository, analyze using StarDist and compare results

public 1yr ago Version: Version 1 0 bookmarks

View Workflow

load-ome-zarr-image-with-labels-from-a-public-s3-r — View Workflow

Help improve this workflow!

This workflow has been published but could be further improved with some additional meta data:

Keyword(s) in categories input, output, operation

You can help improve this workflow by suggesting the addition or removal of keywords, suggest changes and report issues, or request to become a maintainer of the Workflow .

The image is referenced in the paper "NesSys: a novel method for accurate nuclear segmentation in 3D" published August 2019 in PLOS Biology: https://doi.org/10.1371/journal.pbio.3000388 and can be viewed online in the Image Data Resource . This original image was converted into the Zarr format. The analysis results produced by the authors of the paper were converted into labels and linked to the Zarr file which was placed into a public S3 repository. In this notebook, the Zarr file is then loaded together with the labels from the S3 storage and analyzed using StarDist . The StarDist analysis produces a segmentation, which is then viewed side-by-side with the original segmentations produced by the authors of the paper obtained via the loaded labels. Launch This notebook uses the environment_stardist.yml file. See Setup .

Code Snippets

# Package to access data on S3
%pip install aiohttp==3.8.4 zarr==2.14.2

%pip install stardist==0.8.2

Jupyter Notebook From line 2 of Day_5/stardist.ipynb

import dask
import dask.array as da

import matplotlib.pyplot as plt
%matplotlib inline
from dask.diagnostics import ProgressBar
import numpy
import zarr

Jupyter Notebook numpy matplotlib dask zarr From line 9 of Day_5/stardist.ipynb

20	image_id = 6001247

Jupyter Notebook From line 20 of Day_5/stardist.ipynb

ENDPOINT_URL = 'https://uk1s3.embassy.ebi.ac.uk/'

Jupyter Notebook From line 24 of Day_5/stardist.ipynb

def load_binary_from_s3(name, resolution='0'):
    root = '%s/%s/' % (name, resolution)
    with ProgressBar():
        return numpy.asarray(da.from_zarr(ENDPOINT_URL + root))

Jupyter Notebook From line 28 of Day_5/stardist.ipynb

name = 'idr/zarr/v0.1/%s.zarr' % image_id
data = load_binary_from_s3(name)
print(data.shape)

Jupyter Notebook From line 35 of Day_5/stardist.ipynb

name = 'idr/zarr/v0.1/%s.zarr/labels' % image_id
labels = load_binary_from_s3(name)
print(labels.shape)

Jupyter Notebook From line 41 of Day_5/stardist.ipynb

from stardist.models import StarDist2D
model_versatile = StarDist2D.from_pretrained('2D_demo')

Jupyter Notebook From line 47 of Day_5/stardist.ipynb

from csbdeep.utils import normalize
axis_norm = (0,1)
img = normalize(data[0, 1, :, :, :], 1,99.8, axis=axis_norm)
results = []
for im in img:
    new_labels, details = model_versatile.predict_instances(im)
    results.append(new_labels)

label_slices = numpy.array(results)

Jupyter Notebook From line 52 of Day_5/stardist.ipynb

from ipywidgets import *

def update(z=0):
    fig = plt.figure(figsize=(10, 10))
    plt.subplot(121)
    c = 1
    plt.imshow(data[0, c, z, :, :], cmap='gray')
    try:
        plt.imshow(labels[0, c, z, :, :], cmap='jet', alpha=0.5)
    except Exception:
        print(z)
    plt.subplot(122)
    plt.imshow(data[0, c, z, :, :], cmap='gray')
    plt.imshow(label_slices[z, :, :], cmap='jet', alpha=0.5)
    plt.tight_layout()
    fig.canvas.flush_events()

interact(update, z= widgets.IntSlider(value=0, min=0, max=data.shape[2]-1, step=1, description="Select Z", continuous_update=False))