CRL case study: automated image analysis of lung fibrosis with AI

In this case study Charles River Laboratories pathologists are enhancing image analysis in pre-clinical toxicology, lung studies with Aiforia's AI.
Written by Aiforia

The Speciality Pathology Services unit located at the Charles River Laboratories UK site is increasing the speed and accuracy of image analysis in pre-clinical toxicology studies with Aiforia's AI software and services. Veterinary pathologists and scientists at the CRO are harnessing the power of artificial intelligence to automate the analysis of lung fibrosis; producing data at high speed. 

Automated image analysis of lung fibrosis

Screenshot 2020-11-05 at 16.16.41

Overview of case study

  • Sirius red stained rat lung sections
  • AI model trained with 42 images
  • AI model includes 7 feature classes:

Screenshot 2020-11-05 at 16.18.13Screenshot 2020-11-05 at 16.18.24


Screenshot 2020-11-05 at 16.19.46

Automated image analysis results

Screenshot 2020-11-05 at 16.20.39
  • Analysis of individual whole-slide images or in batches
  • Results per individual image
  • Result exported as csv/excel files or via API
  • Area, count, percentages, septal thickness

Lung tissue detection

Screenshot 2020-11-05 at 16.21.48

Lung parenchyma and structural collagen

Screenshot 2020-11-05 at 16.22.38

Lung parenchyma and structural collagen

Screenshot 2020-11-05 at 16.23.47

Lung parenchyma without alveolar space

Screenshot 2020-11-05 at 16.24.37

Fibrotic lesion

Screenshot 2020-11-05 at 16.27.09

Fibrotic lesion

Screenshot 2020-11-05 at 16.28.20

Poorly inflated lung tissue and fibrosis

Screenshot 2020-11-05 at 16.28.56

Poorly inflated lung tissue and fibrosis

Screenshot 2020-11-05 at 16.29.54

Dense collagen within the fibrotic lesion

Screenshot 2020-11-05 at 16.30.30

AI model modification and re-training

  • Easy to add training data and re-train the AI model
  • Possible to modify the AI model classes
    • Addition or removal of classes
    • Re-arrangement of classes
Screenshot 2020-11-05 at 16.32.21

 

Want to find out more about our cloud-based platform for pathology and histology education?

Contact us