High-throughput image analysis with deep learning captures heterogeneity and spatial relationships after kidney injury

Abstract

Recovery from acute kidney injury can vary widely in patients and in animal models. Immunofluorescence staining can provide spatial information about heterogeneous injury responses, but often only a fraction of stained tissue is analyzed. Deep learning can expand analysis to larger areas and sample numbers by substituting for time-intensive manual or semi-automated quantification techniques. Here we report one approach to leverage deep learning tools to quantify heterogenous responses to kidney injury that can be deployed without specialized equipment or programming expertise. We first demonstrated that deep learning models generated from small training sets accurately identified a range of stains and structures with performance similar to that of trained human observers. We then showed this approach accurately tracks the evolution of folic acid induced kidney injury in mice and highlights spatially clustered tubules that fail to repair. We then demonstrated that this approach captures the variation in recovery across a robust sample of kidneys after ischemic injury. Finally, we showed markers of failed repair after ischemic injury were correlated both spatially within and between animals and that failed repair was inversely correlated with peritubular capillary density. Combined, we demonstrate the utility and versatility of our approach to capture spatially heterogenous responses to kidney injury.

Pax Protein Depletion in Proximal Tubules Triggers Conserved Mechanisms of Resistance to Acute Ischemic Kidney Injury and Prevents Transition to Chronic Kidney Disease

“…For clarity only one 400 μm square region is shown (yellow outline on full size image) but analysis was performed over the entire cortex and OSOM (white outline on full size images) for one section from each animal, typically ~2 x 10 5 nuclei per section. Single channel images for markers of interest were segmented using a U-Net convolutional neural network specifically trained to identify positive cells, including their nuclei 27,52…”

“…We next subjected these animals to severe unilateral IRI, with preservation of the contralateral kidney, to model the AKI-to-CKD transition (Figure 1b). We used deep-learning-assisted image segmentation 27 to classify nuclei of GFP+ proximal tubule cells according to co-staining for Vcam1, KIM-1, Pax2, or Pax8 across the entire cortex and outer stripe of the outer medulla (OSOM) 14 d after injury using immunostaining (example analysis is shown in Supplementary Figure S2). This time corresponds to a peak population of cells with a persistent injury phenotype marked by Vcam1 and KIM-1 6 .…”