A high-performance computing voxel-based analysis pipeline for the rodent brain with a formal validation framework
Robert J. Anderson, James J. Cook, Natalie Delpratt, John C. Nouls, Bin Gu, James O. McNamara, G. Allan Johnson, Alexandra Badea
Duke University Medical Center, Durham, NC
Neuroinformatics volume 17, pages451–472(2019). Open Access
Voxel-based analysis (VBA) of preclinical magnetic resonance images is a powerful research tool for neuroscientists, but its use is limited by the prohibitively high computational demands in this domain. We have developed an automated VBA processing pipeline running on a high-performance computing cluster to mitigate this impediment, and can now routinely produce VBA results in 1-3 days for studies comprising large multidimensional arrays--a task that previously took upward of a month. Attempts to rigorously validate the pipeline have revealed a need for more complete and quantitative VBA evaluation methods. To address this, we propose a validation framework consisting of Jacobian calculation testing, morphological phantom creation, and three metrics derived from phantom VBA. The adoption of such a framework should facilitate the creation and communication of VBA results with increased confidence and integrity. We have used this framework to guide the selection of spatial registration parameters in a VBA study involving a mouse model of epilepsy. Due to significantly shortened processing times we have been able to explore multiple parameter sets and examine how alternative choices can impact VBA results. Additionally, testing the Jacobian calculation has revealed greater reliability when using the geometric rather than the finite differences method, and removed a potential source of confusion in this critical VBA processing step in regards to the direction of the warp used. Verifying the accuracy of VBA has so far not received the attention it rightfully deserves, and should be the focus of a broader effort within the community. We hope that by addressing the serious challenges posed by processing times and ensuring the reliability of results, this work will precipitate the ubiquitous adoption of high-quality VBA techniques among preclinical neuroimaging researchers.
CIVM makes many types of data acquired for published and yet unpublished studies available through our CIVM VoxPort application. Use of VoxPort is free. Registration is required. Register for VoxPort access now. A new browser window or tab will open.
CIVM makes all data from published studies available for research. We ask that you provide contact information, and agree to give credit to the Duke Center for In Vivo Microscopy for any written or oral presentation using data from this site. Please use the following acknowledgement: Imaging data provided by the Duke Center for In Vivo Microscopy NIH/NIBIB (P41 EB015897).
Copyright (c) 2015, May-Anh Vu
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
While the code is shared under a BSD license and is open for research purposes it is not guaranteed to be fit for any particular, and especially clinical purpose.
Instructions: Click on a link below. A new browser window or tab will open where you will be prompted to login to CIVMVoxPort. If you do not have login credentials, follow the instructions to register for access. After you login, come back to this page and re-click on a link below to go directly to the desired page.
View the Supplementary Data
The authors would like to thank James Cook, Center for In Vivo Microscopy, Duke University. We are grateful for support from Duke University Exploratory Research Fund, NIA K01-AG041211, Duke Center for In Vivo Microscopy (NIH/NIBIB P41 EB015897)