ExploreASL: An image processing pipeline for multi-center ASL perfusion MRI studies
Mutsaerts, Henk J M M; Petr, Jan; Groot, Paul; Vandemaele, Pieter; Ingala, Silvia; Robertson, Andrew D; Václavů, Lena; Groote, Inge; Kuijf, Hugo; Zelaya, Fernando; O'Daly, Owen; Hilal, Saima; Wink, Alle Meije; Kant, Ilse; Caan, Matthan W A; Morgan, Catherine; de Bresser, Jeroen; Lysvik, Elisabeth; Schrantee, Anouk; Bjørnebekk, Astrid; Clement, Patricia; Shirzadi, Zahra; Kuijer, Joost P A; Wottschel, Viktor; Anazodo, Udunna C; Pajkrt, Dasja; Richard, Edo; Bokkers, Reinoud P H; Reneman, Liesbeth; Masellis, Mario; Günther, Matthias; MacIntosh, Bradley J; Achten, Eric; Chappell, Michael A; van Osch, Matthias J P; Golay, Xavier; Thomas, David L; De Vita, Enrico; Bjørnerud, Atle; Nederveen, Aart; Hendrikse, Jeroen; Asllani, Iris; Barkhof, Frederik
(2020) NeuroImage, volume 219
(Article)
Abstract
Arterial spin labeling (ASL) has undergone significant development since its inception, with a focus on improving standardization and reproducibility of its acquisition and quantification. In a community-wide effort towards robust and reproducible clinical ASL image processing, we developed the software package ExploreASL, allowing standardized analyses across centers and scanners. The
... read more
procedures used in ExploreASL capitalize on published image processing advancements and address the challenges of multi-center datasets with scanner-specific processing and artifact reduction to limit patient exclusion. ExploreASL is self-contained, written in MATLAB and based on Statistical Parameter Mapping (SPM) and runs on multiple operating systems. To facilitate collaboration and data-exchange, the toolbox follows several standards and recommendations for data structure, provenance, and best analysis practice. ExploreASL was iteratively refined and tested in the analysis of >10,000 ASL scans using different pulse-sequences in a variety of clinical populations, resulting in four processing modules: Import, Structural, ASL, and Population that perform tasks, respectively, for data curation, structural and ASL image processing and quality control, and finally preparing the results for statistical analyses on both single-subject and group level. We illustrate ExploreASL processing results from three cohorts: perinatally HIV-infected children, healthy adults, and elderly at risk for neurodegenerative disease. We show the reproducibility for each cohort when processed at different centers with different operating systems and MATLAB versions, and its effects on the quantification of gray matter cerebral blood flow. ExploreASL facilitates the standardization of image processing and quality control, allowing the pooling of cohorts which may increase statistical power and discover between-group perfusion differences. Ultimately, this workflow may advance ASL for wider adoption in clinical studies, trials, and practice.
show less
Download/Full Text
Keywords: Arterial spin labeling, Cerebral perfusion, Image processing, Multi-center, Quality control, Neurology, Cognitive Neuroscience, Journal Article
ISSN: 1053-8119
Publisher: Academic Press Inc.
Note: Funding Information: Several additional features are scheduled for future releases, including full BIDS support (Gorgolewski et al., 2016); support for Hitachi and Canon datasets; unit testing to ensure stability of the pipeline through the continuous development; inclusion of WM atlases for extended WM analysis; a GUI for easier configuration and execution; quantification of advanced ASL schemes such as velocity- and acceleration-selective ASL (Schmid et al., 2015) and integration of the BASIL toolbox to allow multi-PLD and time-encoded sequence quantification (Chappell et al., 2009); and support for individual-center calibration, e.g. using the recently introduced Quantitative ASL Perfusion Reference (QASPER) (Oliver-Taylor et al., 2017) phantom (Gold Standard Phantoms, London, UK). Although ExploreASL's computation times are moderate for research purposes, a clinical scanner implementation would benefit from parallelization on graphical processing units (GPUs) to provide robust automatic QC within clinical scanning time (e.g. <5 min). Another improvement would be the investigation of the effect of image processing choices, as well as the availability of physiological and quantification parameters for different populations (Fazlollahi et al., 2015). This would allow for the incorporation of quantification confidence intervals in the output of ExploreASL. For anonymization purposes, the face can be removed from the structural scans (Nichols et al., 2017; Leung et al., 2015) using a defacing algorithm such as the one implemented in SPM12, but further testing is required to verify that the analysis is not affected (de Sitter et al., 2017b). Statistical analyses can be biased for populations with large inter-subject differences in their deformations, e.g., developing brains or a wide range of atrophy. Dedicated templates are typically used for infants to ensure proper segmentation and normalization (Shi et al., 2011). For older children, the use of a dedicated template is still advised (Sanchez et al., 2012), although adult templates are often sufficient. Further errors in deformations and volume changes can be encountered when stretching pediatric brains to an adult standard space. Either, such deformations need to be accounted in both the volumes and CBF maps, or the analysis has to be performed in standard space, knowing that the pGM thresholds can be different with different total brain volumes as discussed in Section 3.6. A more optimal solution is to use the CerebroMatic toolbox (Wilke et al., 2017) is a tool that accounts for this bias and will be incorporated in future releases of ExploreASL. Finally, we intend to implement ExploreASL as a cloud solution and a plugin for scanner workstation to allow seamless ExploreASL image processing in clinical routine.This project has received support from the following EU/EFPIA Innovative Medicines Initiatives (1 and 2) Joint Undertakings: EPAD grant no. 115736, AMYPAD grant no. 115952. Additionally, this work received support from the EU-EFPIA Innovative Medicines Initiatives Joint Undertaking (grant No 115952). HM is supported by Amsterdam Neuroscience funding. FB and XG are supported by NIHR funding through the UCLH Biomedical Research Centre. DLT is supported by the UCL Leonard Wolfson Experimental Neurology Centre (PR/ylr/18575). EDV is supported by the Wellcome/EPSRC Centre for Medical Engineering [WT 203148/Z/16/Z]. IA is supported by The Gleason Foundation. MJPvO receives research support from Philips, the EU under the Horizon 2020 program (project: CDS-QUAMRI, project number 634541), and the research program Innovational Research Incentives Scheme Vici with project number 016.160.351, which is financed by the Netherlands Organization for Scientific Research (NWO). MC received funding from the Engineering and Physical Sciences Research Council UK (EP/P012361/1), and is a shareholder of Nico.lab BV, Amsterdam, The Netherlands. The Wellcome Centre for Integrative Neuroimaging is supported by core funding from the Wellcome Trust (203139/Z/16/Z). The authors wish to thank the COST-AID (European Cooperation in Science and Technology - Arterial spin labeling Initiative in Dementia) Action BM1103 and the Open Source Initiative for Perfusion Imaging (OSIPI) and the ISMRM Perfusion Study groups for facilitating meetings for researchers to discuss the implementation of ExploreASL. The authors acknowledge Guillaume Flandin, Robert Dahnke, and Paul Schmidt for reviewing the structural module for its implementation of SPM12, CAT12, and LST, respectively; Krzysztof Gorgolewksi for his advice on the BIDS implementation; Jens Maus for help with MEX compilation; Cyril Pernet for providing the SPM Univariate Plus scripts; and Koen Baas for curating the Philips 3D GRASE data. The authors acknowledge the following researchers and teams: Yannis Paloyelis from King's College London, for providing the data of the INtranasal OxyTocin trial, Torbjørn Elvsåshagen from Oslo University Hospital for providing the Sleep study dataset; the EPAD investigators for providing the Amsterdam site elderly dataset; Kim van de Ven from Philips Healthcare for providing the 3D GRASE dataset; Philip de Witt Hamer from Amsterdam UMC for providing the PICTURE dataset, and Chris Chen from the National University of Singapore for providing the Singapore Memory Clinical dataset. Funding Information: This project has received support from the following EU / EFPIA Innovative Medicines Initiatives (1 and 2) Joint Undertakings: EPAD grant no. 115736 , AMYPAD grant no. 115952 . Additionally, this work received support from the EU- EFPIA Innovative Medicines Initiatives Joint Undertaking (grant No 115952 ). HM is supported by Amsterdam Neuroscience funding . FB and XG are supported by NIHR funding through the UCLH Biomedical Research Centre . DLT is supported by the UCL Leonard Wolfson Experimental Neurology Centre ( PR/ylr/18575 ). EDV is supported by the Wellcome/EPSRC Centre for Medical Engineering [ WT 203148/Z/16/Z ]. IA is supported by The Gleason Foundation . MJPvO receives research support from Philips , the EU under the Horizon 2020 program (project: CDS-QUAMRI, project number 634541 ), and the research program Innovational Research Incentives Scheme Vici with project number 016.160.351 , which is financed by the Netherlands Organization for Scientific Research (NWO) . MC received funding from the Engineering and Physical Sciences Research Council UK ( EP/P012361/1 ), and is a shareholder of Nico.lab BV, Amsterdam, The Netherlands. The Wellcome Centre for Integrative Neuroimaging is supported by core funding from the Wellcome Trust ( 203139/Z/16/Z ). The authors wish to thank the COST-AID (European Cooperation in Science and Technology - Arterial spin labeling Initiative in Dementia) Action BM1103 and the Open Source Initiative for Perfusion Imaging (OSIPI) and the ISMRM Perfusion Study groups for facilitating meetings for researchers to discuss the implementation of ExploreASL. The authors acknowledge Guillaume Flandin, Robert Dahnke, and Paul Schmidt for reviewing the structural module for its implementation of SPM12, CAT12, and LST, respectively; Krzysztof Gorgolewksi for his advice on the BIDS implementation; Jens Maus for help with MEX compilation; Cyril Pernet for providing the SPM Univariate Plus scripts; and Koen Baas for curating the Philips 3D GRASE data. The authors acknowledge the following researchers and teams: Yannis Paloyelis from King’s College London, for providing the data of the INtranasal OxyTocin trial, Torbjørn Elvsåshagen from Oslo University Hospital for providing the Sleep study dataset; the EPAD investigators for providing the Amsterdam site elderly dataset; Kim van de Ven from Philips Healthcare for providing the 3D GRASE dataset; Philip de Witt Hamer from Amsterdam UMC for providing the PICTURE dataset, and Chris Chen from the National University of Singapore for providing the Singapore Memory Clinical dataset. Publisher Copyright: © 2020
(Peer reviewed)
