Tractography dissection variability: What happens when 42 groups dissect 14 white matter bundles on the same dataset?

• Type of Material: Article
• Authors: Kurt G. Schilling, Vanderbilt University; François Rheault, University of Sherbrooke; Laurent Petit, University of Bordeaux; Colin B. Hansen, Vanderbilt University; Vishwesh Nath, Vanderbilt University; Fang-Cheng Yeh, University of Pittsburgh; Gabriel Girard, Center for BioMedical Imaging; Muhamed Barakovic, University of Basel; Jonathan Rafael-Patino, Ecole Polytech Federale de Lausanne; Thomas Yu, Ecole Polytech Federale de Lausanne; Elda Fischi-Gomez, Ecole Polytech Federale de Lausanne; Marco Pizzolato, Technical University of Denmark; Mario Ocampo-Pineda, University of Verona; Simona Schiavi, University of Verona; Vince D. Calhoun, Emory University
• Language: English
• Date: 2021-08-25
• Publisher: Academic Press Inc. Elsevier Science
• Publication Version: Author Accepted Manuscript (After Peer Review)
• Copyright Statement: © 2021 Published by Elsevier Inc.
• License: Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International
• Final Published Version (URL): http://dx.doi.org/10.1016/j.neuroimage.2021.118502
• Title of Journal or Parent Work: NeuroImage
• Volume: 243
• Start Page: 118502
• End Page: 118502
• Grant/Funding Information: LF acknowledges the support of the Cluster of Excellence Matters of Activity. Image Space Material funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germanýs Excellence Strategy – EXC 2025.; MP received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 754462; AJG was supported by P41EB015898. NM was supported by R01MH119222, K24MH116366, and R01MH111917.; LC acknowledges support from CONACYT and UNAM.; KS, BL, CH were supported by the National Institutes of Health under award numbers R01EB017230, and T32EB001628, and in part by ViSE/VICTR VR3029 and the National Center for Research Resources, Grant UL1 RR024975-01.; SW is supported by a Medical Research Council PhD Studentship UK [MR/N013913/1].; JPA, MA and SMS acknowledges the support of FCT - Fundação para a Ciência e a Tecnologia within CINTESIS, R&D Unit (reference UID/IC/4255/2013).; ARM acknowledges support from CONACYT. LJO, YR, and FZ were supported by NIH P41EB015902 and R01MH119222.; CL is partly supported by NIH grants P41 EB027061 and P30 NS076408 “Institutional Center Cores for Advanced Neuroimaging.; C-HY is grateful to the Ministry of Science and Technology of Taiwan (MOST 109-2222-E-182-001-MY3) for the support.; JYMY received positional funding from the Royal Children’s Hospital Foundation (RCH 1000).; MM was funded by the Wellcome Trust through a Sir Henry Wellcome Postdoctoral Fellowship [213722/Z/18/Z].; NSF OAC-1916518, NSF IIS-1912270, NSF IIS-1636893, NSF BCS-1734853, Microsoft Faculty Fellowship to F.P. D.B. was partially supported by NIH NIMH T32-MH103213 to William Hetrick (Indiana University); NSF OAC-1916518, NSF IIS-1912270, NSF IIS-1636893, NSF BCS-1734853, NIH NIBIB 1R01EB029272-01, and a Microsoft Faculty Fellowship to F.P.; EJC-R is supported by the Swiss National Science Foundation (SNSF, Ambizione grant PZ00P2 185814/1).; This work was also possible thanks to the support of the Institutional Research Chair in NeuroInformatics of Université de Sherbrooke, NSERC and Compute Canada (MD, FR).; The Wisconsin group acknowledges the support from a core grant to the Waisman Center from the National Institute of Child Health and Human Development (IDDRC U54 HD090256).; CMWT is supported by a Sir Henry Wellcome Fellowship (215944/Z/19/Z) and a Veni grant from the Dutch Research Council (NWO) (17331).; JYMY, JC, and CEK acknowledge the support of the Royal Children’s Hospital Foundation, Murdoch Children’s Research Institute, The University of Melbourne Department of Paediatrics, and the Victorian Government’s Operational Infrastructure Support Program.; FC acknowledges the support of the National Health and Medical Research Council of Australia (APP1091593 and APP1117724) and the Australian Research Council (DP170101815).; PG, CR, NL and AV were partially supported by ANID-Basal FB0008 and ANID-FONDECYT 1190701 grants.; This project has received funding from the European Union’s Horizon 2020 Research and Innovation Programme under Grant Agreement No. 785907 & 945539 (HBP SGA2 & SGA3), and from the ANR IFOPASUBA- 19-CE45-0022-01.
• Keywords: Mean diffusivity; Neurosciences & Neurology; Dissection; Diffusion MRI; Bundle segmentation; White matter; Arcuate fasciculus; Tensor imaging tractography; In-vivo; Science & Technology; Neurosciences; Fractional anisotropy; Fiber pathways; Anatomical accuracy; Tractography; Principal eigenvector measurements; Neuroimaging; Corpus callosum; Life Sciences & Biomedicine; Radiology, Nuclear Medicine & Medical Imaging
• Research Categories: Health Sciences, Radiology; Engineering, Biomedical; Computer Science; Biology, Neuroscience

Abstract

White matter bundle segmentation using diffusion MRI fiber tractography has become the method of choice to identify white matter fiber pathways in vivo in human brains. However, like other analyses of complex data, there is considerable variability in segmentation protocols and techniques. This can result in different reconstructions of the same intended white matter pathways, which directly affects tractography results, quantification, and interpretation. In this study, we aim to evaluate and quantify the variability that arises from different protocols for bundle segmentation. Through an open call to users of fiber tractography, including anatomists, clinicians, and algorithm developers, 42 independent teams were given processed sets of human whole-brain streamlines and asked to segment 14 white matter fascicles on six subjects. In total, we received 57 different bundle segmentation protocols, which enabled detailed volume-based and streamline-based analyses of agreement and disagreement among protocols for each fiber pathway. Results show that even when given the exact same sets of underlying streamlines, the variability across protocols for bundle segmentation is greater than all other sources of variability in the virtual dissection process, including variability within protocols and variability across subjects. In order to foster the use of tractography bundle dissection in routine clinical settings, and as a fundamental analytical tool, future endeavors must aim to resolve and reduce this heterogeneity. Although external validation is needed to verify the anatomical accuracy of bundle dissections, reducing heterogeneity is a step towards reproducible research and may be achieved through the use of standard nomenclature and definitions of white matter bundles and well-chosen constraints and decisions in the dissection process.