Measuring and reducing the carbon emissions of fMRI research computing

[NickESouter]

By SEA-SIG representatives: Nick Souter (University of Sussex), Niall Duncan (Taipei Medical University), Nikhil Bhagwat (McGill University), Polona Kalc (Jena University Hospital)

Recording and streaming volunteers Polona Kalc & Niall Duncan

Emergent sessions

Short description and the goals for the session

We represent the OHBM Sustainability and Environmental Action Special Interest Group (SEA-SIG). During this session, we will present recent empirical work we have conducted on measuring and reducing the compute power and therefore the carbon emissions of preprocessing and statistical analysis of fMRI data. In particular, we will discuss how to use multiple carbon tracking tools and real-time carbon intensity task schedulers, including live demonstration of their use to attendees.

Green computing is an increasingly important aspect of socially responsible science, and intersects with a number of open science practices including preregistration of data processing parameters, development of clear data management plans, and reflection on how and where to share data publicly. We will reflect on these intersections and invite discussion on how we should best approach tensions between open science and sustainability.

Speaker Nick Souter will join the session virtually, and Niall Duncan and Nikhil Bhagwat will provide live presenation and demonstration in person in the OSR. Polona Kalc, incoming SEA-SIG chair, will also be present in person to introduce the session.

Goals:

• Learn the source of carbon emissions arise in neuroimaging computing and data storage
• Learn how to measure and reduce carbon emissions in their own neuroimaging data processing
• Reflect on best practice for environmentally sustainable and open neuroimaging research

Useful Links

OHBM SEA-SIG [https://ohbm-environment.org/]
10 recommendations for sustainable computing in neuroimaging [https://direct.mit.edu/imag/article/doi/10.1162/imag_a_00043/118246/Ten-recommendations-for-reducing-the-carbon]
Measuring the carbon footprint of fMRIPrep [https://osf.io/preprints/osf/wmzcq]
Climate Aware Task Scheduler [https://github.com/GreenScheduler/cats]
Green Algorithms 4 HPC [https://www.green-algorithms.org/GA4HPC/]
CodeCarbon [https://codecarbon.io/]

@NickESouter
@nw-duncan
@nikhil153

[slugtmeijer]

https://ohbm.github.io/osr2024/schedule/
There you find the crowdcast link for the session to join from online and share if you want.