QSIPrep: A robust and unified workflow for preprocessing and reconstructing diffusion MRI
Poster No:
1559
Submission Type:
Abstract Submission
Authors:
Matthew Cieslak1, Philip Cook1, Thijs Dhollander2, Fang-Cheng Yeh3, Eleftherios Garyfallidis4, Mark Elliott5, Valerie Sydnor1, Ursula Tooley1, Josiane Bourque1, Xiaosong He1, Will Foran3, Laura Cabral3, Beatriz Luna3, Adam Pines1, David Roalf1, Allyson Mackey1, John Detre1, Max Kelz1, Jean Vettel6, Barry Giesbrecht7, Desmond Oathes1, Danielle Bassett1, Scott Grafton7, Theodore Satterthwaite1
Institutions:
1University of Pennsylvania, Philadelphia, PA, 2Florey Institute of Neuroscience, Melbourne, VIC, 3University of Pittsburgh, Pittsburgh, PA, 4University of Indiana, Bloomington, IN, 5University of Pennsylvania, Pennsylvania, PA, 6Army Research Labs, Aberdeen, MD, 7University of California Santa Barbara, Santa Barbara, CA
First Author:
Co-Author(s):
Introduction:
Although diffusion-weighted magnetic resonance imaging (dMRI) can take many forms, they all sample q-space in order to characterize water diffusion. Numerous pipelines and software platforms have been built for processing dMRI data, but most work on only a subset of sampling schemes, or implement only parts of the processing workflow. Comparisons across methods are hindered by incompatible software, diverse file formats, and inconsistent naming conventions, among others. Here we introduce QSIPrep, a new processing pipeline for diffusion images that is compatible with virtually all dMRI sampling schemes via a uniform, containerized application. Preprocessing includes denoising, distortion correction, head motion correction, coregistration, and spatial normalization. Individual algorithms from a diverse set of cutting-edge software suites are combined to capitalize upon their complementary strengths. Throughout, QSIPrep provides both visual and quantitative measures of data quality and "glass-box" methods reporting. Together, these features allow for easy implementation of best practices while simultaneously maximizing reproducibility.
Methods:
QSIPrep is a BIDS-app that builds preprocessing workflows based on the data and accompanying metadata in a BIDS dataset. Scans can be automatically grouped together by their acquisition parameters so that denoising operations (e.g. MP-PCA [1] and bias correction [2]) are performed on images that share similar susceptibility distortions. Distortion correction can be applied using any BIDS-supported fieldmap type or an experimental registration-based distortion correction. By allowing for both FSL's eddy [3] and the SHORELine method [4], head motion correction is available for virtually any dMRI sampling scheme. All spatial transformations utilize ANTs [5]. Visual reports and quality control (QC) measures are produced for each operation, and manuscript-ready text is automatically generated to describe the whole workflow using a similar framework to fMRIPrep [6].
After preprocessing, QSIPrep provides a set of curated reconstruction workflows that showcase recommended pipelines from MRtrix [7], MRtrix3Tissue [8], DSI Studio [9], and DIPY [10]. Inter-operability allows users to mix-and-match across software platforms, for example reconstructing FODs in MRtrix and running tractography in DSI Studio. Since inputs and outputs to each method are managed within QSIPrep, users can try new methods without needing to familiarize themselves with the technical details of multiple software packages. Structural connectivity measures and scalar images are produced in the same format from all reconstruction workflows, facilitating methods comparison at the group level. The preprocessing and reconstruction pipelines are shown in Figure 1.
After preprocessing, QSIPrep provides a set of curated reconstruction workflows that showcase recommended pipelines from MRtrix [7], MRtrix3Tissue [8], DSI Studio [9], and DIPY [10]. Inter-operability allows users to mix-and-match across software platforms, for example reconstructing FODs in MRtrix and running tractography in DSI Studio. Since inputs and outputs to each method are managed within QSIPrep, users can try new methods without needing to familiarize themselves with the technical details of multiple software packages. Structural connectivity measures and scalar images are produced in the same format from all reconstruction workflows, facilitating methods comparison at the group level. The preprocessing and reconstruction pipelines are shown in Figure 1.
·Figure 1: The QSIPrep workflow for preprocessing (left) and reconstructing (right) diffusion-weighted MRI.
Results:
As a demonstration, we preprocessed a 113-direction DSI dataset (n=209), a 258-direction DSI dataset (n=150) and a 62-direction multi-shell dataset (n=87) with QSIPrep. A model-free q-space nearest-neighbors QC measure (Neighboring DWI Correlation NDC, [11]) was calculated before and after preprocessing with QSIPrep. NDC reflects the similarity of images that are close to one another in q-space. QSIPrep improved the NDC measure for each of the datasets it was tested on. Results are shown in Figure 2.
·Figure 2: NDC scores from before (x-axis) and after (y-axis) preprocessing with QSIPrep. Identity is plotted as a solid line. The largest quality metric increases were observed for DSI data.
Conclusions:
QSIPrep provides reproducible, versatile, and scalable pipelines for processing nearly all kinds of diffusion MRI data. It is currently being used across a number of institutions to process DTI, DSI, DKI, CS-DSI and multi-shell schemes. In the month following its public release, QSIPrep was launched over 4,000 times. QSIPrep is both open-source and community-driven. We encourage new users to try the software; code contributions and bug reports are welcome. Documentation is available at qsiprep.readthedocs.io.
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
Diffusion MRI Modeling and Analysis 1
Image Registration and Computational Anatomy
Motion Correction and Preprocessing
Neuroinformatics and Data Sharing:
Workflows 2
Keywords:
Data analysis
MRI
Spatial Warping
Tractography
WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC
1|2Indicates the priority used for review
My abstract is being submitted as a Software Demonstration.
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Are you Internal Review Board (IRB) certified? Please note: Failure to have IRB, if applicable will lead to automatic rejection of abstract.
Was any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Was any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using author date format
[2] Tustison, N. J., Avants, B. B., Cook, P. A., Zheng, Y., Egan, A., Yushkevich, P. A., & Gee, J. C. (2010). N4ITK: improved N3 bias correction. IEEE transactions on medical imaging, 29(6), 1310.
[3] Andersson, J. L., & Sotiropoulos, S. N. (2016). An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging. Neuroimage, 125, 1063-1078.
[4] Matthew Cieslak, Philip A. Cook, Scott T. Grafton, Mark A. Elliott, David Roalf, Danielle S. Bassett, Desmond Oathes, Theodore Satterthwaite. (2019) A head motion correction algorithm for arbitrary q space sampling schemes with high b values. Poster presented at: OHBM Rome, Italy.
[5] Tustison, N. J., Cook, P. A., Klein, A., Song, G., Das, S. R., Duda, J. T., ... & Avants, B. B. (2014). Large-scale evaluation of ANTs and FreeSurfer cortical thickness measurements. Neuroimage, 99, 166-179.
[6] Esteban O, Markiewicz CJ, Blair RW, Moodie CA, Isik AI, Erramuzpe A, Kent JD, Goncalves M, DuPre E, Snyder M, Oya H, Ghosh SS, Wright J, Durnez J, Poldrack RA, Gorgolewski KJ. fMRIPrep: a robust preprocessing pipeline for functional MRI. (2018) Nature Methods
[7] J-Donald Tournier, Robert Smith, David Raffelt, Rami Tabbara, Thijs Dhollander, Maximilian Pietsch, Daan Christiaens, Ben Jeurissen, Chun-Hung Yeh, Alan Connelly,
MRtrix3: A fast, flexible and open software framework for medical image processing and visualisation (2019), NeuroImage, 116137
[8] https://3tissue.github.io
[9] Yeh, Fang-Cheng, Van Jay Wedeen, and Wen-Yih Isaac Tseng, Generalized-sampling imaging. (2010) Medical Imaging, IEEE Transactions on 29.9, 1626-1635
[10] Garyfallidis, E., Brett, M., Amirbekian, B., Rokem, A., Van Der Walt, S., Descoteaux, M., & Nimmo-Smith, I. (2014). Dipy, a library for the analysis of diffusion MRI data. Frontiers in neuroinformatics, 8, 8.
[11] Yeh, F. C., Zaydan, I. M., Suski, V. R., Lacomis, D., Richardson, R. M., Maroon, J., & Barrios-Martinez, J. (2019). Differential tractography as a track-based biomarker for neuronal injury. NeuroImage, 576025.