ReproFlow: a scalable environment for automated MRI and behavioral data integration
Poster No:
2277
Submission Type:
Abstract Submission
Authors:
Horea-Ioan Ioanas1, Vadym Melnyk2, David Kennedy3, Yaroslav Halchenko1
Institutions:
1Center for Open Neuroscience, Dartmouth College, Hanover, NH, 2Center for Open Neuroscience, Kyiv, Ukraine, 3University of Massachusetts Chan Medical School, Worcester, MA
First Author:
Co-Author(s):
Introduction:
Reproducibility is a critical consideration for modern neuroscience and is greatly aided by automation of data acquisition and standardization of data records.
MRI and behavioral data are two of the foremost modalities in human neuroscience, making the seamless integration of these modalities a significant concern for numerous research centers.
The Brain Imaging Data Structure (BIDS) is a preeminent data standard, well-suited for both modalities, and which ensures interoperability of data analysis tools as well as transparency of data records.
The ReproNim project has made significant contributions in extending the BIDS standard, and creating tools for BIDS conversion, data sharing, quality assurance (QA).
ReproFlow is an environment which integrates numerous ReproNim tools - such as HeuDiConv, ReproIn, ReproStim, ReproEvents, ReproMon, con/noisseur, ///repronim/containers, DataLad, and the datalad-containers extension - in order to provide a scalable and automated solution for MRI and behavioral data acquisition and integration in a standardized form.
Here we present a pilot implementation of this environment, set up at the Dartmouth Brain Imaging Center, covering both software and open hardware solutions.
The adaptation of this environment can help other centers establish a robust, multi-modal, and BIDS-compliant data acquisition pipeline, and thus significantly advance the reliability of modern neuroscience.
MRI and behavioral data are two of the foremost modalities in human neuroscience, making the seamless integration of these modalities a significant concern for numerous research centers.
The Brain Imaging Data Structure (BIDS) is a preeminent data standard, well-suited for both modalities, and which ensures interoperability of data analysis tools as well as transparency of data records.
The ReproNim project has made significant contributions in extending the BIDS standard, and creating tools for BIDS conversion, data sharing, quality assurance (QA).
ReproFlow is an environment which integrates numerous ReproNim tools - such as HeuDiConv, ReproIn, ReproStim, ReproEvents, ReproMon, con/noisseur, ///repronim/containers, DataLad, and the datalad-containers extension - in order to provide a scalable and automated solution for MRI and behavioral data acquisition and integration in a standardized form.
Here we present a pilot implementation of this environment, set up at the Dartmouth Brain Imaging Center, covering both software and open hardware solutions.
The adaptation of this environment can help other centers establish a robust, multi-modal, and BIDS-compliant data acquisition pipeline, and thus significantly advance the reliability of modern neuroscience.
Methods:
We have developed a number of Free and Open Source Software (FOSS) solutions, and made extensive contributions to the BIDS standard, in order to ensure both standard support for multimodal metadata, and adequate tools to automatically populate the metadata space.
The ReproFlow environment consists of 8 core tools developed by the ReproNim project.
HeuDiConv provides configurable MRI conversion from DICOM to a desired layout.
ReproIn provides configuration for HeuDiConv via an extensive heuristic syntax, as well as a user assistance utility.
ReproEvents provides audio and video capture capabilites to integrate complex stimuli with MRI data.
ReproStim provides support for capturing behavioral events from participants.
Con/noisseur captures and performs QA on operator input at the scanner console.
ReproMon complements the QA capabilities by providing support for online operator feedback and alerts in case of incidents or anomalous metadata input.
///ReproNim//containers provides DataLad dataset with popular containers and assistance scripts to ensure reproducible execution.
DataLad and datalad-containers enable data and container manipulation, as well as provenance tracking.
The ReproFlow environment consists of 8 core tools developed by the ReproNim project.
HeuDiConv provides configurable MRI conversion from DICOM to a desired layout.
ReproIn provides configuration for HeuDiConv via an extensive heuristic syntax, as well as a user assistance utility.
ReproEvents provides audio and video capture capabilites to integrate complex stimuli with MRI data.
ReproStim provides support for capturing behavioral events from participants.
Con/noisseur captures and performs QA on operator input at the scanner console.
ReproMon complements the QA capabilities by providing support for online operator feedback and alerts in case of incidents or anomalous metadata input.
///ReproNim//containers provides DataLad dataset with popular containers and assistance scripts to ensure reproducible execution.
DataLad and datalad-containers enable data and container manipulation, as well as provenance tracking.
Results:
Over the course of its development, our HeuDiConv/ReproIn implementation at the Dartmouth Brain Imaging Center has been used to collect and standardize over 40 MRI datasets, which are now openly shareable in an understandable fashion for inspection and reuse by the broader research community.
We have additionally collected corresponding audio/video stimuli using ReproStim, which were successfully used to recover previously undocumented experimental aspects (such as randomization order) and to improve data quality by identifying the presence of lag between modalities.
ReproEvents, ReproMon, and Con/noisseur are currently in early deployment and provide incipient event time stamp synchronization between the various modalities.
///ReproNim/containers contains all BIDS-Apps, NeuroDesk applications, and other containers required to reliably execute the ReproFlow tools.
We have additionally collected corresponding audio/video stimuli using ReproStim, which were successfully used to recover previously undocumented experimental aspects (such as randomization order) and to improve data quality by identifying the presence of lag between modalities.
ReproEvents, ReproMon, and Con/noisseur are currently in early deployment and provide incipient event time stamp synchronization between the various modalities.
///ReproNim/containers contains all BIDS-Apps, NeuroDesk applications, and other containers required to reliably execute the ReproFlow tools.
·Schematic of elements in the workflow and their interconnections.
Conclusions:
We argue based on our results that data integration remains a non-trivial matter for multi-modal set-ups and that significant improvements in automation and transparency are necessary to ensure data reliability.
In particular, general-purpose open-source tools are needed in order to ensure sustainability of acquisition frameworks over time, and to ensure relevant know-how is shared across centers.
We propose ReproFlow as a solution for these requirements and encourage re-use of this environment.
In particular, general-purpose open-source tools are needed in order to ensure sustainability of acquisition frameworks over time, and to ensure relevant know-how is shared across centers.
We propose ReproFlow as a solution for these requirements and encourage re-use of this environment.
Neuroinformatics and Data Sharing:
Databasing and Data Sharing 2
Workflows 1
Keywords:
Acquisition
Data Organization
MRI
Open-Source Hardware
Open-Source Software
Workflows
Other - Automation; BIDS; Multimodal; Reproducibility
1|2Indicates the priority used for review
Provide references using author date format
Halchenko Y.O et al. (2021). DataLad: distributed system for joint management of code, data, and their relationship. Journal of Open Source Software, 6(63):3262, jul 2021. doi: 10.21105/joss.03262
Halchenko Y.O et al. (under review). HeuDiConv — flexible DICOM conversion into structured directory layouts. JOSS.
https://github.com/openjournals/joss-reviews/issues/5839