Complex exercise

In this complex exercise we will setup a workflow for multiple processing images in a folder. We will measure properties of the segmented nuclei and stored them in one big table.

If you have any questions and we are not around, feel free to open a thread on image.sc

import napari
import pandas as pd
from skimage.io import imread
from napari_simpleitk_image_processing import label_statistics
import os
import seaborn
import matplotlib.pyplot as plt

data_path = "../../data/BBBC007_batch/"

Load an image from the specified folder and open it in napari.

viewer = napari.Viewer()

image = imread("../../data/BBBC007_batch/17P1_POS0013_D_1UL.tif")

viewer.add_image(image)

Image segmentation

Set up an image segmentation workflow for this image to segment nuclei. Do this using the napari-assistant and export a notebook. Copy paste the code from this notebook into this function:

def segment_image(image):

    # enter code here
    
    return label_image

test_label_image = segment_image(image)
test_label_image

Feature extraction

Use the following function body to program a routine that takes the image, a corresponding label image and retrieve measurements. You may for example use the function label_statistics from napari_simpleitk_image_processing. Make sure your function creates a pandas DataFrame.

def analyze_image(image, label_image):
    
    # enter code here
    
    return pd.DataFrame(statistics)

analyze_image(image, test_label_image)

Batch processing

Use the following for-loop to iterate through the folder, segment all images, analyse the images, create data frames and concatenate all results in one table.

results = pd.DataFrame()

for file_name in os.listdir(data_path):
    if file_name.endswith('.tif'):
        file_path = os.path.join(data_path, file_name)
        
        
        # enter code here
        
        
        # save from which image the measurements were derived from
        table['file_name'] = file_name
        
        # collect results
        results = pd.concat([results, table])

results

Save the resulting measurements to disk.

Plotting

Extract area measurements from two different images from the table above and draw a box plot.

fig, ax = plt.subplots(figsize = (10, 5))

# enter code here