Image processing in the Acute to Chronic Pain Signatures Project
Poster No:
2220
Submission Type:
Abstract Submission
Authors:
Patrick Sadil1, Konstantinos Arfanakis2, Enamul Hoque Bhuiyan3, Brian Caffo1, Vince Calhoun4,5,6, Mark DeLano7, James Ford8,9, Xiaodong Guo10, Micah Johnson1, Heejung Jung11, Ari Kahn12, Nondas Leloudas13, Qingfei Luo3, Martin Lindquist1, Todd Mulderink7, Scott Peltier14, Christopher Sica2, Pottumarthi Prasad13, Joshua Urrutia12, Carol Vance15, Tor Wager11, David Zhu16, Xiaohong Joe Zhou3, Yong Zhou7, The Acute to Chronic Pain Signatures Consortium A2CPS17
Institutions:
1Johns Hopkins University, Baltimore, MD, 2Rush University Medical Center, Chicago, IL, 3University of Illinois College of Medicine at Chicago, Chicago, IL, 4Emory, Atlanta, GA, 5Georgia State University, Atlanta, GA, 6Georgia Institute of Technology, Atlanta, GA, 7Corewell Health Radiology, Grand Rapids, MI, 8Geisel School of Medicine at Dartmouth, Hanover, NH, 9Dartmouth Hitchcock Medical Center, Lebanon, NH, 10University of Chicago, Chicago, IL, 11Dartmouth College, Hanover, NH, 12University of Texas at Austin, Austin, TX, 13NorthShore University HealthSystem, Evanston, IL, 14University of Michigan, Ann Arbor, MI, 15University of Iowa, Iowa City, IA, 16Michigan State University, East Lansing, MI, 17A2CPS, Baltimore, MD
First Author:
Co-Author(s):
Vince Calhoun
Emory|Georgia State University|Georgia Institute of Technology
Atlanta, GA|Atlanta, GA|Atlanta, GA
Emory|Georgia State University|Georgia Institute of Technology
Atlanta, GA|Atlanta, GA|Atlanta, GA
James C. Ford, PhD
Geisel School of Medicine at Dartmouth|Dartmouth Hitchcock Medical Center
Hanover, NH|Lebanon, NH
Geisel School of Medicine at Dartmouth|Dartmouth Hitchcock Medical Center
Hanover, NH|Lebanon, NH
Introduction:
Typically, the pain of an acute injury goes away after it heals. But sometimes, the pain of an injury, surgery, or disease can linger, eventually becoming chronic. Currently, a high proportion of people in the United States transition to chronic pain after an acute event, but the causes of this transition remain unknown. A new project aims to study the transition: the Acute to Chronic Pain Signatures (A2CPS) initiative (Berardi et al. 2022; Sluka et al. 2021). The study aims to collect neuroimaging data on over 1000 participants who will undergo an incident of acute pain–surgery for either total knee replacement or thoracic surgery. Information collected includes psychosocial, omics, quantitative sensory testing, and brain magnetic resonance imaging data. Participants are imaged both before and after their surgery. The multimodality and scale of these data will provide a unique opportunity to test candidate biomarkers for susceptibility to chronic pain and generate novel, putative, biomarkers and biosignatures. Here, we present the first release of the brain imaging data, quality control procedures, and analysis pipelines.
Methods:
The A2CPS imaging protocol was designed to allow collection of several candidate biomarkers (Figure 1a), including volumetric and gray matter density differences in individual regions of interest, structural connectivity, fractional anisotropy, evoked responses, functional connectivity measures, graph theoretic measures, and multivariate signature responses. The scan protocols are based on those from the Adolescent Brain Cognitive Development study (Casey et al. 2018), and were tailored to the study and to the scanner hardware of participating collection sites.
Analyses comprise three stages (Figure 1b). First, the data are indexed and organized according to the Brain Imaging Data Structure (Gorgolewski et al. 2016). Then, the data are sent through a series of pipelines, including both established (MRIQC, fMRIPrep, FreeSurfer, QSIprep, fsl_anat, CAT12; Esteban et al. 2017, Esteban et al. 2018, Fischl 2012, Cieslak et al. 2021, Jenkinson et al. 2012. Gaser et al. 2022) and bespoke pipelines. Finally, the outputs from these pipelines are aggregated, de-identified, and stored with data from other modalities.
To ensure usability, the raw data have undergone quality control and assurance. The aim of this was a straightforward, three-tier quality rating assessing comparability: "no known defects", "minor defects/issues, correctable", and "major defects/issues, not expected to be comparable". The quality rating is derived from a combination of automated metadata checks (e.g., adherence to acquisition parameters), standardized visual review (e.g., checks for eye spillover), and automated thresholds on measures extracted from the data (e.g., average framewise displacement).
Analyses comprise three stages (Figure 1b). First, the data are indexed and organized according to the Brain Imaging Data Structure (Gorgolewski et al. 2016). Then, the data are sent through a series of pipelines, including both established (MRIQC, fMRIPrep, FreeSurfer, QSIprep, fsl_anat, CAT12; Esteban et al. 2017, Esteban et al. 2018, Fischl 2012, Cieslak et al. 2021, Jenkinson et al. 2012. Gaser et al. 2022) and bespoke pipelines. Finally, the outputs from these pipelines are aggregated, de-identified, and stored with data from other modalities.
To ensure usability, the raw data have undergone quality control and assurance. The aim of this was a straightforward, three-tier quality rating assessing comparability: "no known defects", "minor defects/issues, correctable", and "major defects/issues, not expected to be comparable". The quality rating is derived from a combination of automated metadata checks (e.g., adherence to acquisition parameters), standardized visual review (e.g., checks for eye spillover), and automated thresholds on measures extracted from the data (e.g., average framewise displacement).
Results:
The initial data release comprises 595 participants with imaging data, with all images collected before the incident of acute pain. To facilitate analyses, the (deidentified) raw data are provided along with phenotypes derived from the images. This initial release includes outputs from structural, resting, and task MRI (Figure 2). These image-derived phenotypes provide one way to begin machine-learning studies immediately. Users wishing to run their own analysis pipelines have access to the raw data.
Conclusions:
The A2CPS research initiative provides a unique opportunity to study the multimodality of the transition to chronic pain in a large dataset. This is the first of several planned releases. The dataset itself, analysis pipelines, and quality control procedures will be made available to researchers outside of the consortium. Future releases will include scans post-surgery, as well as refinements to the imaging derivatives. For neuroimagers, the data and associated pipelines provide a rich medium to study predictive biomarkers, and to contribute efforts to address the pressing health issue of chronic pain.
Neuroinformatics and Data Sharing:
Databasing and Data Sharing 1
Perception, Attention and Motor Behavior:
Perception: Pain and Visceral 2
Keywords:
Data analysis
Data Organization
Data Registration
Informatics
Machine Learning
Open Data
Pain
Workflows
1|2Indicates the priority used for review
Provide references using author date format
Casey, B. J., et al. 2018. “The Adolescent Brain Cognitive Development (ABCD) Study: Imaging Acquisition across 21 Sites.” Developmental Cognitive Neuroscience 32:43–54. doi: 10.1016/j.dcn.2018.03.001.
Cieslak, M., et al. 2021. “QSIPrep: An Integrative Platform for Preprocessing and Reconstructing Diffusion MRI Data.” Nature Methods. doi: 10.1038/s41592-021-01185-5.
Esteban, O., et al. 2017. “MRIQC: Advancing the Automatic Prediction of Image Quality in MRI from Unseen Sites” edited by B. C. Bernhardt. PLOS ONE 12(9):e0184661. doi: 10.1371/journal.pone.0184661.
Esteban, O., et al. 2018. “fMRIPrep: A Robust Preprocessing Pipeline for Functional MRI.” Nature Methods. doi: 10.1038/s41592-018-0235-4.
Fischl, B. 2012. “FreeSurfer.” NeuroImage 62(2):774–81. doi: 10.1016/j.neuroimage.2012.01.021.
Gaser, G., et al. 2022. CAT – A Computational Anatomy Toolbox for the Analysis of Structural MRI Data. preprint. Neuroscience. doi: 10.1101/2022.06.11.495736.
Gorgolewski, K. J., et al. 2016. “The Brain Imaging Data Structure, a Format for Organizing and Describing Outputs of Neuroimaging Experiments.” Scientific Data 3:160044. doi: 10.1038/sdata.2016.44.
Jenkinson, M., et al. 2012. “FSL.” NeuroImage 62(2):782–90. doi: 10.1016/j.neuroimage.2011.09.015.
Sluka, K., et al. 2021. “Conceptual Design and Protocol for the Acute to Chronic Pain Signatures Program (A2CPS).” The Journal of Pain 22(5):586. doi: 10.1016/j.jpain.2021.03.036.