Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Aug 15;45(12):e70003.
doi: 10.1002/hbm.70003.

Measuring and reducing the carbon footprint of fMRI preprocessing in fMRIPrep

Nicholas E Souter  1 , Nikhil Bhagwat  2 , Chris Racey  1   3 , Reese Wilkinson  4 , Niall W Duncan  5 , Gabrielle Samuel  6 , Loïc Lannelongue  7   8   9   10 , Raghavendra Selvan  11   12 , Charlotte L Rae  1
Affiliations

Measuring and reducing the carbon footprint of fMRI preprocessing in fMRIPrep

Nicholas E Souter et al. Hum Brain Mapp. .

Abstract

Computationally expensive data processing in neuroimaging research places demands on energy consumption-and the resulting carbon emissions contribute to the climate crisis. We measured the carbon footprint of the functional magnetic resonance imaging (fMRI) preprocessing tool fMRIPrep, testing the effect of varying parameters on estimated carbon emissions and preprocessing performance. Performance was quantified using (a) statistical individual-level task activation in regions of interest and (b) mean smoothness of preprocessed data. Eight variants of fMRIPrep were run with 257 participants who had completed an fMRI stop signal task (the same data also used in the original validation of fMRIPrep). Some variants led to substantial reductions in carbon emissions without sacrificing data quality: for instance, disabling FreeSurfer surface reconstruction reduced carbon emissions by 48%. We provide six recommendations for minimising emissions without compromising performance. By varying parameters and computational resources, neuroimagers can substantially reduce the carbon footprint of their preprocessing. This is one aspect of our research carbon footprint over which neuroimagers have control and agency to act upon.

Keywords: Carbon; Computing; fMRI; fMRIPrep; footprint; neuroimaging; preprocessing.

PubMed Disclaimer

Conflict of interest statement

The authors have no competing interests or conflicts of interest to disclose.

Figures

FIGURE 1
FIGURE 1
ROIs for use in FSL Featquery, including the left primary motor cortex for motor response activation, and the pre‐SMA for successful response inhibition activation. pre‐SMA, pre‐supplementary motor area; ROI, region of interest.
FIGURE 2
FIGURE 2
Experimental procedure of preprocessing and analysis steps, with outcome measures derived from each stage. CPU, central processing unit; GLM, generalised linear model; RAM, random‐access memory.
FIGURE 3
FIGURE 3
Estimated carbon emissions (dotted green line) plotted against (a) statistical task activation in regions of interest, (b) estimated data smoothness, (c) duration of preprocessing, and (d) CPU and RAM energy usage, for each pipeline. Error bars reflect one standard error of the mean. These are frequently too small to be visible. Note that the scale used varies between variables, see text below for percent changes. N = 257. CO2eq, carbon dioxide equivalent; CPU, central processing unit; kg, kilograms; kWh, kilowatt hours; mm, millimetres; pre‐SMA, pre‐supplementary motor area; RAM, random‐access memory.
FIGURE 4
FIGURE 4
The mean total size (GB) of all files generated for a given subject, split by ‘derivatives’ (final output files), ‘scratch’ (working directory), and the subject‐specific ‘FreeSurfer’ directory, for each pipeline. Error bars are one standard error of the mean, these are too small to be visible. The mean change in size relative to the baseline pipeline (P0) is highlighted for each experimental pipeline. N = 257. GB, gigabytes.
FIGURE 5
FIGURE 5
Activation count maps for each fMRIPrep pipeline. Values reflect the percentage of participants within the sample (N = 257) showing significant individual‐level activation in a given voxel for both the ‘go > successful stop’ (motor; hot colours) and ‘successful stop > go’ (response inhibition; cool colours) contrasts. Slices presented are at MNI coordinates Z = 4 and Z = 52. AROMA, automatic removal of motion artifacts; ICA, independent components analysis.
See this image and copyright information in PMC

References

    1. Andraszewicz, S. , Scheibehenne, B. , Rieskamp, J. , Grasman, R. , Verhagen, J. , & Wagenmakers, E.‐J. (2015). An introduction to Bayesian hypothesis testing for management research. Journal of Management, 41(2), 521–543. 10.1177/0149206314560412 - DOI
    1. Anthony, L. F. W. , Kanding, B. , & Selvan, R. (2020). Carbontracker: Tracking and predicting the carbon footprint of training deep learning models. arXiv, 2007.03051. https://arxiv.org/abs/2007.03051
    1. Aron, A. R. , Ivry, R. B. , Jeffery, K. J. , Poldrack, R. A. , Schmidt, R. , Summerfield, C. , & Urai, A. E. (2020). How can neuroscientists respond to the climate emergency? Neuron, 106, 17–20. 10.1016/j.neuron.2020.02.019 - DOI - PubMed
    1. Bakhtiarifard, P. , Igel, C. , & Selvan, R. (2024). EC‐NAS: Energy consumption aware tabular benchmarks for neural architecture search. International Conference on Acoustics. 10.48550/arXiv.2210.06015 - DOI
    1. Bilder, R. , Poldrack, R. , Cannon, T. , London, E. , Freimer, N. , Congdon, E. , Karlsgodt, K. , & Sabb, F. (2020). UCLA consortium for neuropsychiatric Phenomics LA5c study. OpenNeuro. [Dataset] 10.18112/openneuro.ds000030.v1.0.0 - DOI