Abstract: FR-PO035
Tracking the 3D Architecture of Hundreds of Nephrons and Peritubular Capillaries in Health and Disease Using Light Sheet Microscopy and Deep Learning
Session Information
- AI, Digital Health, Data Science - II
November 03, 2023 | Location: Exhibit Hall, Pennsylvania Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Augmented Intelligence, Digital Health, and Data Science
- 300 Augmented Intelligence, Digital Health, and Data Science
Authors
- Poudel, Chetan, University of Washington, Seattle, Washington, United States
- Sandoval, Ruben M., Indiana University School of Medicine, Indianapolis, Indiana, United States
- Wong, Madeline K., University of Washington, Seattle, Washington, United States
- Liu, Jonathan T. C., University of Washington, Seattle, Washington, United States
- Vaughan, Joshua C., University of Washington, Seattle, Washington, United States
Group or Team Name
- Vaughan Lab.
Background
Until recently, it has been practically impossible to study whole kidney nephrons and peritubular capillaries throughout the many millimeters of tissue necessary to examine them in their entirety. This technical limitation has obscured correlations and causal relationships between different parts of the same nephron, and between nearby nephrons and vasculature in cases of injury. Our study will use recent technical breakthroughs in 3D imaging, tissue clearing, and artificial intelligence to study thick specimens and process large datasets. We aim to establish a 3D connectivity map of hundreds to thousands of nephrons and study the coupling of injuries between glomeruli, tubules, adjacent nephrons, and peritubular capillaries at different stages of chronic kidney disease.
Methods
We used in vivo antibody labeling and solvent-based clearing to uniformly stain and clear whole mouse kidneys, which were then imaged on a custom-built lightsheet microscope with sub-cellular spatial resolution. We processed the 3D fluorescence data using a human-in-the-loop AI method, a deep-learning-based supervised image segmentation that allows for scaling up to large volumes and segmenting many nephrons at a time.
Results
From a single fluorescence label, our deep learning segmentation model traced hundreds of whole nephrons in 3D. We also used vasculature labeling to visualize and quantify the characteristics of the peritubular capillary network around the nephrons. So far, we have generated 3D renderings of nephrons and connectivity maps, and performed associated morphometrics (length, volume, tortuosity) for nephrons in healthy kidneys.
Conclusion
We have developed a pipeline using 3D imaging and artificial intelligence to extract relevant 3D features in hundreds of whole nephrons and capillaries in the same tissue. We will apply this framework to study the spatial correlations of injuries at various timepoints in models of chronic kidney diseases, such as focal segmental glomerulosclerosis and ischemia-reperfusion injury.
Funding
- Private Foundation Support