Sep 21, 2024

Public workspaceDeepSlice Automated Alignment

DOI
https://dx.doi.org/10.17504/protocols.io.3byl4946ogo5/v1
  • Michael Henderson1
  • 1Van Andel Institute
Michael Henderson
Icon indicating open access to content
QR code linking to this content
DOI: https://dx.doi.org/10.17504/protocols.io.3byl4946ogo5/v1
Protocol Citation: Michael Henderson 2024. DeepSlice Automated Alignment. protocols.io https://dx.doi.org/10.17504/protocols.io.3byl4946ogo5/v1
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: September 21, 2024
Last Modified: September 25, 2024
Protocol Integer ID: 108117
Keywords: deepslice automated alignment this protocol, deepslice in the quint workflow, deepslice automated alignment, quint workflow, deepslice, alignment, workflow, quint
Funders Acknowledgements:
Aligning Science Across Parkinson's
Abstract
This protocol explains how to use DeepSlice in the Quint Workflow
Troubleshooting
Before start
DeepSlice is a deep neural network that automatically aligns mouse histology images through the Allen Brain Atlas coordinate framework. Alignments are viewable and refinable in QuickNII and set up sections to give a good starting point near and around each section’s correct plane. DeepSlice’s alignment is not completely accurate and further fine tuning in QuickNII is necessary. This is a newly developed tool and not imperative for the workflow but helps speed the process of registration and substitutes the FileBuilder step.
DeepSlice Automated Alignment
Open DeepSlice in your web browser.



Select Choose Files.
a. Upload all images for registration from the QVN folder
Ensure Mouse is selected for species, and Model Ensemble is checked (uses two DeepSlice versions to optimize alignment ).
a. Avoid checking Angle Integration (this aligns all your brain sections to the same angle, which is inaccurate when blocks have different cutting angles)
b. Avoid checking Using Cutting Index (this suggests your sections numbers (_s###) correspond to serial section numbers spaced equally apart)  
c. Optional to allow DeepSlice to use your data to improve the neural network and its predictive accuracy.
Select Submit.
After all sections/slices are processed, press Download XML.
Insert the downloaded XML (titled by default ‘results.XML’) into your QVN folder with all your images for registration.