ASN's Mission

ASN leads the fight to prevent, treat, and cure kidney diseases throughout the world by educating health professionals and scientists, advancing research and innovation, communicating new knowledge, and advocating for the highest quality care for patients.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on Twitter

Kidney Week

Abstract: PO0491

A Computational Pipeline for Estimating Renal Histologic Primitives

Session Information

Category: Bioengineering

  • 300 Bioengineering

Authors

  • Sarder, Pinaki, University at Buffalo, Buffalo, New York, United States
  • Chen, Yijiang, Case Western Reserve University, Cleveland, Ohio, United States
  • Ginley, Brandon, University at Buffalo, Buffalo, New York, United States
  • Rosenberg, Avi Z., Johns Hopkins University, Baltimore, Maryland, United States
  • Janowczyk, Andrew, Case Western Reserve University, Cleveland, Ohio, United States
  • Lutnick, Brendon, University at Buffalo, Buffalo, New York, United States
  • Lucarelli, Nicholas, University at Buffalo, Buffalo, New York, United States
  • Alpers, Charles E., University of Washington, Seattle, Washington, United States
  • Jain, Sanjay, Washington University in St Louis, St Louis, Missouri, United States
  • Grewenow, Stephanie, University of Washington, Seattle, Washington, United States
  • Steck, Becky, University of Michigan, Ann Arbor, Michigan, United States
  • Barisoni, Laura, Duke University, Durham, North Carolina, United States
  • Madabhushi, Anant, Case Western Reserve University, Cleveland, Ohio, United States
  • Balis, Ulysses G. J., University of Michigan, Ann Arbor, Michigan, United States

Group or Team Name

  • Kidney Precision Medicine Project
Background

While a number of image segmentation approaches exist for digital pathology, in this study, we aimed to combine three complementary tools for enhanced quality assessment and segmentation of histologic primitives on kidney biopsy images. Specifically, we evaluated the integrated image analysis approach on two chronic kidney disease (CKD) renal tissue biopsy WSI cases from the curated Kidney Precision Medicine Project (KPMP) database.

Methods

HistoQC, an established tool for automated quality control (QC) of digital pathology images, eliminated staining and image artifacts from the biopsies. Next, two complementary computational tools were interfaced. HAIL (Human-AI-Loop), a supervised convolutional neural network, was used to isolate renal compartments (i.e., glomeruli, vessels, interstitium, and tubules). HAIL output was used to gate the training process of the second algorithm, VIPR (Vectorizing spatially-Invariant Pattern Recognition), a kernel-based, high-dimensional textural classifier capable of extracting distinct histopathological features as distinct sub-regions, ultimately allowing for the generation of a resulting pixel-level classification of all tissue compartments.

Results

The integrated pipeline precisely classified tubular basement membrane, brush border, nuclei, nucleoli, and tubular epithelial cytoplasm from CKD WSIs in a manner consistent with subject matter expert opinion for region fractionation, when resultant segmentation images were manually reviewed (Fig. 1).

Conclusion

We have shown the potential of our pipeline for image curation, and segmentation and sub-characterization of renal histologic primitives.

Funding

  • NIDDK Support