Machine learning in renal pathology

Matthew Nicholas Basso; Moumita Barua; Julien Meyer; Rohan John; April Khademi

doi:10.3389/fneph.2022.1007002

Machine learning in renal pathology

Front Nephrol. 2022 Nov 29:2:1007002. doi: 10.3389/fneph.2022.1007002. eCollection 2022.

Authors

Matthew Nicholas Basso¹, Moumita Barua^{2

3

4

5}, Julien Meyer⁶, Rohan John⁷, April Khademi^{1

8

9}

Affiliations

¹ Image Analysis in Medicine Lab (IAMLAB), Department of Electrical, Computer, and Biomedical Engineering, Ryerson University, Toronto, ON, Canada.
² Division of Nephrology, University Health Network, Toronto, ON, Canada.
³ Toronto General Hospital Research Institute, Toronto General Hospital, Toronto, ON, Canada.
⁴ Department of Medicine, University of Toronto, Toronto, ON, Canada.
⁵ Institute of Medical Sciences, University of Toronto, Toronto, ON, Canada.
⁶ School of Health Services Management, Ryerson University, Toronto, ON, Canada.
⁷ Department of Pathology, University Health Network, Toronto, ON, Canada.
⁸ Keenan Research Center for Biomedical Science, St. Michael's Hospital, Unity Health Network, Toronto, ON, Canada.
⁹ Institute for Biomedical Engineering, Science, and Technology (iBEST), a partnership between St. Michael's Hospital and Ryerson University, Toronto, ON, Canada.

Abstract

Introduction: When assessing kidney biopsies, pathologists use light microscopy, immunofluorescence, and electron microscopy to describe and diagnose glomerular lesions and diseases. These methods can be laborious, costly, fraught with inter-observer variability, and can have delays in turn-around time. Thus, computational approaches can be designed as screening and/or diagnostic tools, potentially relieving pathologist time, healthcare resources, while also having the ability to identify novel biomarkers, including subvisual features.

Methods: Here, we implement our recently published biomarker feature extraction (BFE) model along with 3 pre-trained deep learning models (VGG16, VGG19, and InceptionV3) to diagnose 3 glomerular diseases using PAS-stained digital pathology images alone. The BFE model extracts a panel of 233 explainable features related to underlying pathology, which are subsequently narrowed down to 10 morphological and microstructural texture features for classification with a linear discriminant analysis machine learning classifier. 45 patient renal biopsies (371 glomeruli) from minimal change disease (MCD), membranous nephropathy (MN), and thin-basement membrane nephropathy (TBMN) were split into training/validation and held out sets. For the 3 deep learningmodels, data augmentation and Grad-CAM were used for better performance and interpretability.

Results: The BFE model showed glomerular validation accuracy of 67.6% and testing accuracy of 76.8%. All deep learning approaches had higher validation accuracies (most for VGG16 at 78.5%) but lower testing accuracies. The highest testing accuracy at the glomerular level was VGG16 at 71.9%, while at the patient-level was InceptionV3 at 73.3%.

Discussion: The results highlight the potential of both traditional machine learning and deep learning-based approaches for kidney biopsy evaluation.

Keywords: computational pathology; deep learning; digital image analysis; glomerulus; machine learning; membranous nephropathy; minimal change disease; thin-basement membrane nephropathy.