Jun 20, 2023

Kidney Functional Tissue Unit (FTU) Segmentation

DOI
https://dx.doi.org/10.17504/protocols.io.kxygx3mkog8j/v1
Kidney Functional Tissue Unit (FTU) Segmentation
  • 1Vanderbilt University;
  • 2Delft University of Technology
  • VU Biomolecular Multimodal Imaging Center / Spraggins Research Group
  • KPMP
Katerina V Djambazova
Icon indicating open access to content
QR code linking to this content
DOI: https://dx.doi.org/10.17504/protocols.io.kxygx3mkog8j/v1
Protocol Citation: Nathan Heath Patterson, ellie.l.pingry , Katerina Djambazova, Melissa Farrow, Raf Van De Plas, Jeff Spraggins 2023. Kidney Functional Tissue Unit (FTU) Segmentation. protocols.io https://dx.doi.org/10.17504/protocols.io.kxygx3mkog8j/v1
Manuscript citation:
Patterson, N. H.; Neumann, E. K.; Sharman, K.; Allen, J.; Harris, R.; Fogo, A. B.; De Caestecker, M.; Caprioli, R. M.; Van De Plas, R.; Spraggins, J. M. Autofluorescence Microscopy as a Label-Free Tool for Renal Histology and Glomerular Segmentation. https://doi.org/10.1101/2021.07.16.452703.
License: This is an open access  protocol  distributed under the terms of the  Creative Commons Attribution License,  which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: June 16, 2023
Last Modified: October 18, 2023
Protocol  Integer ID: 83532
Keywords: kidney functional tissue unit, functional tissue unit, segmentation model to autofluorescence, segmentation of glomeruli, microscopy image, distal tubule, ftus, kidney, ftu, autofluorescence, proximal tubule, glomeruli, tissue, segmentation, segmentation model
Abstract
This protocol explains how to apply a segmentation model to autofluorescence microscopy images to find kidney functional tissue units (FTUs). Currently the model allows for segmentation of glomeruli, proximal tubules, thick ascending limb, distal tubules, and collecting ducts.

Materials
For performance, a workstation with at least 32 GB of memory is advised and a GPU with 12 GB memory is required for any GPU-based prediction.
Apply kidney FTU segmentation to autofluorescence images
Configure the segmentation prediction by creating the dataset .yaml file. Add the path to the saved model and images in the configuration file.
Create the python environment needed to run wsimap. See instructions at the wsimap GitHub.
Run the model using the command line from the wsimap directory like so:
python scripts/config/instance-predict-from-config-newmodels-multi.py "/path/to/configuration-file-from-step2.yaml


Collect .geojson segmentations in output folder specified in .yaml file from Step 2. These can be visualized in QuPath.
Protocol references
Patterson, N. H.; Neumann, E. K.; Sharman, K.; Allen, J.; Harris, R.; Fogo, A. B.; De Caestecker, M.; Caprioli, R. M.; Van De Plas, R.; Spraggins, J. M. Autofluorescence Microscopy as a Label-Free Tool for Renal Histology and Glomerular Segmentation. https://doi.org/10.1101/2021.07.16.452703.