Evaluation of 14 nonlinear deformation algorithms applied to human brain MRI registration

By Arno Klein, Jesper L. R. Andersson, B. Ardekani, John Ashburner, Brian B. Avants, M. Chiang, Gary E. Christensen, D. Louis Collins, James C. Gee, P. Hellier and 9 more. NeuroImage, 2009

2,380 citations recorded by the source corpus.

All fields of neuroscience that employ brain imaging need to communicate their results with reference to anatomical regions. In particular, comparative morphometry and group analysis of functional and physiological data require coregistration of brains to establish correspondences across brain structures. It is well established that linear registration of one brain to another is inadequate for aligning brain structures, so numerous algorithms have emerged to nonlinearly register brains to one another. This study is the largest evaluation of nonlinear deformation algorithms applied to brain image registration ever conducted. Fourteen algorithms from laboratories around the world are evaluated using 8 different error measures. More than 45,000 registrations between 80 manually labeled brains were performed by algorithms including: AIR, ANIMAL, ART, Diffeomorphic Demons, FNIRT, IRTK, JRD-fluid, ROMEO, SICLE, SyN, and four different SPM5 algorithms (“SPM2-type” and regular Normalization, Unified Segmentation, and the DARTEL Toolbox). All of these registrations were preceded by linear registration between the same image pairs using FLIRT. One of the most significant findings of this study is that the relative performances of the registration methods under comparison appear to be little affected by the choice of subject population, labeling protocol, and type of overlap measure. This is important because it suggests that the findings are generalizable to new subject populations that are labeled or evaluated using different labeling protocols. Furthermore, we ranked the 14 methods according to three completely independent analyses (permutation tests, one-way ANOVA tests, and indifference-zone ranking) and derived three almost identical top rankings of the methods. ART, SyN, IRTK, and SPM's DARTEL Toolbox gave the best results according to overlap and distance measures, with ART and SyN delivering the most consistently high accuracy across subjects and label sets. Updates will be published on the http://www.mindboggle.info/papers/ website.

Subfields: Computer Vision and Pattern Recognition, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/6d2669ebc27f776d940ca95db0ad4ae455316cf2

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

Medical image registration: a review (2014, methodology): “Regarding non-rigid registration, 14 algorithms were compared in the registration of brains in Klein et al. (2009), namely: AIR, ANIMAL, ART, Diffeomorphic Demons, FNIRT, IRTK, JRD-ﬂuid, ROMEO, SICLE, SyN and four different SPM5 algorithms (‘SPM2-type’ and regular Normalization, Uniﬁed…”
Benchmarking of participant‐level confound regression strategies for the control of motion artifact in studies of functional connectivity (2017, methodology): “, 2010), each structural image was diffeomorphically registered to the study-specific PNC template using the top-performing SyN deformation (Klein et al., 2009).”
An Unsupervised Learning Model for Deformable Medical Image Registration (2018, result): “We compare our approach to Symmetric Normalization (SyN) [4], the top-performing registration algorithm in a comparative study [25].”
elastix: A Toolbox for Intensity-Based Medical Image Registration (2010, background): “More recently, a fairly complete overview of image registration software is given in [64], which compares the performance of 14 nonrigid registration programs on MR brain images.”
Scalable High Performance Image Registration Framework by Unsupervised Deep Feature Representations Learning (2015, methodology): “57%, which is slightly better than the top-ranked registration method in [3].”
A bi-ventricular cardiac atlas built from 1000+ high resolution MR images of healthy subjects and an analysis of shape and motion (2015, positive): “The reason to use B-spline non-rigid registration is that it has demonstrated good performance in previous studies (Klein et al., 2009), but other registration tools can also be used in this step.”
VoxelMorph: A Learning Framework for Deformable Medical Image Registration (2018, methodology): “3) Baseline Methods: We use Symmetric Normalization (SyN) [11], the top-performing registration algorithm in a comparative study [2] as a first baseline.”
Lead-DBS: A toolbox for deep brain stimulation electrode localizations and visualizations (2015, positive): “the co-registration to a subcortical atlas (Klein et al., 2009).”
Survey of Non-Rigid Registration Tools in Medicine (2017, methodology): “A tool developed by the The Oxford Centre for Functional Magnetic Resonance Imaging of the Brain (FMRIB) was excluded as it does not provide non-rigid registration methods (FMRIB’s Linear Image Registration Tool).”
Visualization of image data from cells to organisms (2010, neutral): “Many tools have been developed for image registratio”
Nipype: A Flexible, Lightweight and Extensible Neuroimaging Data Processing Framework in Python (2011, background): “Increased diversity of neuroimaging data processing software has made systematic comparison of performance and accuracy of underlying algorithms essential (for examples, see Klein et al., 2009, 2010).”
Deformable Medical Image Registration: A Survey (2013, methodology): “…Internet Brain Segmentation Repository - IBSR [448], the CUMC12 dataset 2 acquired at the Columbia University Medical Center, the MGH10 dataset 2 scanned at the MGH/MIT/HMS Athinoula A. Martinos Center for Biomedical Imaging) has made possible evaluation studies like the one by Klein et al. [449].”
Large Deformation Diffeomorphic Image Registration with Laplacian Pyramid Networks (2020, methodology): “s Demons [27] and SyN [2]) and two cutting edge DLIR methods (denoted as VM[4] and DIF-VM [5]). Demons and SyN are the top-performing registration among 14 classical non-linear deformation algorithms [12]. Both Demons and SyN utilize a 3-level multi-resolution strategy to capture large deformation. ”