[STORY] Analysis the .svs file using QuPath.

[anzhao]

Metadata

[anzhao]

The patient ID and other information is under the label of associated images under the Image tab in QuPath. We can use Optical Character Recognition (OCR) library such as Keras-OCR to extract the info we needed.

[anzhao]

The complete workflow of auto-extract the Patient ID and other info from the label of associated images under the Image tab within a svs file:

1. Extract the Slide Label Image:

import openslide

# Open the .svs file
slide = openslide.OpenSlide('an.svs')

# Extract the slide label image
label_image = slide.associated_images['label']

# Save the label image for the next step Optical Character Recognition (OCR) processing
label_image.save('patient_id.png')

2. Perform OCR on the Extracted Image

#!/usr/bin/env python3
import keras_ocr

# Create the pipeline
pipeline = keras_ocr.pipeline.Pipeline()

# Read the image
images = [keras_ocr.tools.read('patient_id.png')]

# Perform OCR and Recognize text in images
prediction_groups = pipeline.recognize(images)

# Print the recognized text
for predictions in prediction_groups:
    for text, box in predictions:
        print(text)

[anzhao]

The single slide image produce six DICOMs, when I open them using QuPath, it all display correctly.