THE MNI Ecosystem

Database management and processing ecosystem for data-sharing initiatives

NOVEMBER 2017

Samir Das
Director of Technology

McGill Centre for Integrative Neuroscience
Montreal Neurological Institute

THE MNI Ecosystem

Database management and processing ecosystem for data-sharing initiatives

NOVEMBER 2017

Samir Das
Director of Technology

McGill Centre for Integrative Neuroscience
Montreal Neurological Institute

THE MNI Ecosystem

Database management and processing ecosystem for data-sharing initiatives

NOVEMBER 2017

Samir Das
Director of Technology

McGill Centre for Integrative Neuroscience
Montreal Neurological Institute

What’s involved?

Longitudinal Acquisition, Storage and Curation, Interoperability, Reproducibility, Transfer, Anonymization, Security, Privacy, Ethics, APIs, Validation, Quality Control, Protocol Checking, Preprocessing, Analysis, HPC, Provenance, Ontological Standarization,
Data Harmonization, Upgrades, Maintenance, Bug Fixes, User Interface, Javascript, Bootstrap, Tracking, Extensibility, Data Management, Summary Statistics, Workflows,
Development, Tool Integration, Data Sharing, Download, Multi-Modal Linking, Querying, Image Processing, Visualization, Networking, System Administration, Partnerships, Funding, HR ...No big deal!

LORIS-CBRAIN INTEGRATION

WHAT IS LORIS?

“..is a modular and extensible web-based data and project management software that integrates all aspects of a multi-center neuroimaging research study. It is an OPEN SOURCE framework and database suitable for managing large datasets.”

What is LORIS?


Heterogenous data acquisition

Storage, processing & dissemination

LORIS Globally

What does LORIS do?

What does LORIS do?

Extensible and modular design

What does LORIS actually do?

DATA ENTRY
DATA MANAGEMENT
DATA VISUALISATION
DATA QUERYING

TRY ME!

Heterogenous Data

Candidate Profile, Instrument Builder, Feedback Module, Double Data Entry, Conflict Resolver

Imaging Data

1. Acquisition & Storage

2. Visualisation: Imaging Browser, BrainBrowser

3. Quality Control: Radiological Review Module, DCC MRI

Data Querying

  • Imaging Statistiscs

  • Data Querying Tool (DQT)

Cross-Modal Querying

Real-Time Query Results

Statistical Analysis

Genomics Browser

WHAT IS CBRAIN?


“..is an OPEN SOURCE web-based collaborative High-Performance Computing (HPC) platform for neuroimaging research, connecting researchers to the tools required to handle large datasets.”

What does CBRAIN do?

..enables distributed execution of software pipelines

..aggregates multiple distributed file systems into uniform view

NOW

611 users (199 international)
191 sites
29 countries

CBRAIN Computing Resources

51 tools, 1M+ files
6 Compute Canada clusters
24M CPU core hours

CBRAIN portal

TRY ME!

CARMIN

Common API for Research Medical Imaging Network

A common web API for remote pipeline execution

NIAK

http://niak.simexp-lab.org/

From Data

To

Visualization

BrainBrowser

..a set of web-based 3D visualization tools primarily used for viewing neurological data i.e. MRI scans.

It allows for real-time manipulation and analysis of 3D neuroimaging data through any modern web browser. TRY ME!

Volume Viewer on LORIS

BigBrain

The BigBrain is the digitized reconstruction of 7404 hi-res coronal histological sections (20 microns isotropic).

It is the brain of a 65 year-old man with no neurological or psychiatric diseases in clinical records at time of death.

BigBrain - high resolution reference brain

Multimodal integration into an anatomically realistic standard space

Years of development

20 micron resolution

7404 histological slices

1 Terabyte dataset

Click to launch the Neuroglancer Viewer!

What is BigBrain?

The BigBrain dataset is the result of a collaborative effort between the teams of Dr. Katrin Amunts and Dr. Karl Zilles (Forschungszentrum Jülich), and Dr. Alan Evans (Montreal Neurological Institute)

Available freely at https://bigbrain.loris.ca

Visualization with Atelier 3D

PIVT (New)

Display

Moving Neurogimaging to the web

Easy integration into CBRAIN and LORIS

Pixpipe

Big Brain and Neuroglancer

Feedback

  • 8 year study
  • Longitudinal data
  • 532 subjects
  • 8000 distinct variables
  • 37000 individual assessments
  • T1, T2, PD, DTI
  • Spectroscopy
  • ∼3TB of imaging data
  • 2000 MRI acquisitions

  • Infant brain development in autism
  • 3000+ scans
  • 700+ subjects
  • 7000 distinct variables
  • 150,000+ individual assessments
  • T1, T2, DTI, BOLD
  • ∼5 TBs of imaging data
  • Genetic/biospecimen data

Quebec Parkinson Network

OMEGA

  • Open MEG Archive

  • First open MEG repository

  • 180 user accounts created

  • Associated structural data

  • 500+ reads on academia.edu

ADNI

  • Cited 343 times

  • Data-use agreements

  • Used for countless analyses

  • Restrictive

Healthy Brains for Healthy Lives (HBHL)

An interdisciplinary program at McGill University leveraging neuroinformatic technologies to transform terminal or life-long afflications to treatable or even curable conditions

What is Data Sharing?

Exchange of information

Datasets

Tools

Standardization

Databases

Collaborations

Conferences, Hackathons

Facebook, Google, Twitter, etc.


Image source: http://blog.veritythink.com/post/87880448269/creative-data-sharing-and-open-humanitarianism

Data Sharing Issues

Benefits Hurdles
More citations Fear
Reduces waste/duplication Technical challenges
Increased exposure Privacy concerns
Access to larger datasets Data Harmonization
Access to rare data Interoperability
Less attrition Reproducibility
Increased validation Obtaining ethics
Saving Tax $$/more funding Public dataset not identical

Global Data Sharing Initiatives

Committee on Best Practices in Data Analysis and Sharing (COBIDAS)

Collaboration with global data sharing groups

INCF, Open Science Framework, NITRC, Allen Institute, NDAR, HAWG (Atlas Building Group), Organization Human Brain Mapping, Human Brain Project, Compute Canada, Maelstrom, Edinburgh BRAINS ImageBank, COINS, XNAT, VIP, BrainCode, BIDS, NIDM, NeuroDevNet, GUSTO, QPN, Boutiques, VIP

Hackathons

Important value of hackathons to data sharing initiatives

Open Science

Cyberinfrastructure

Cybersecurity

Open Science Functionality

Acquisition & Storage Dissemination & Analysis
Organized/Accessible data Centralized repository
Long term storage Provenance Capture
Quality Control mechanisms API for interoperability
Web visualization Consent is factored in
Tablet/Mobile Friendly User account access control
Anonymized automatically Completely de-identified
Cross-modal linking Cross-study correlation
Online Data Querying Access to high performance computing

LORIS Biobank instance

LORIS Imaging instance

Benefits of Open Science

Organized/Accessible data Greater exposure
More collaborations Data validation
Access to Quality Control results Access to larger datasets
More citations Greater funding
Cross-modal linking Cross-study correlation
More robust findings Greater reproducibility

MNI History

1999: MNI ecosystem implemented for NIH multisite MRI Study of Normal Brain Development

2015

LORIS: 130 sites worldwide, 30,000+ data collection time points, 500+ behavioral instruments, 30+ TBs imaging datasets, 200,000 acquisitions

CBRAIN: 300+ users, 60 cities, 20 countries,
diverse projects: Prevent AD, K-ADNI, MAVAN, NeuroDevNet, 130 virtual sites, 600+ TBs storage grid across 25 servers!

National Network

Including "Data Publishing"

Future Direction

  • Phase 2 LORIS/CBRAIN connectivity
    •
  • Enhanced API
    •
  • Additional modalities (eg. MEG/EEG)
    •
  • Institutional requirements
    •
  • Enhanced visualization
    •
  • Numerous new modules
    •
  • Biobanking Tracking
    •
  • Interoperability!
    •
  • Pipeline reproducibility
    •
  • Cloud technology
    •
  • Flexibility in storage, sharing, and remote processing!

Thank you!Acknowledgements: Alan Evans, Alex Zijdenbos, Dario Vins, Jonathan Harlap, Matt Charlet, Andrew Corderey, Sebastian Muehlboeck, Reza Adalat, Louis Collins, Vladimir Fonov, Marc Rousseau, Mia Petkova, Rathi Gnanasekaran, David Brownlee, Tarek Sherif, Pierre Rioux, Nic Kassis, Leigh MacIntyre, Claude Lepage, Ilana Leppert, Natasha Beck, Tristan Glatard, Bert Vincent, Lindsay Lewis, Najma Mahani, Elodie Portales-Casamar, Alden Woodward, Sylvain Milot, Jean Francois Malouin, Sylvain Baillet, Daniel Kroetz, Martin Weiss, Mathieu Desrosier, Jason Karamchandani, Amit Bar-Or, Ted Fon, John Brietner, Derek Lo, Patrick Bermudez, Chris Steele, Pamela Patterson and one of my favourites: Pierre Bellec!

LORIS team on left