Tom Maullin


Post-doctorate Researcher
in fMRI Statistics


Contact

Contact name


Contact description


Academic Department


Academic Institution




Software


This page contains information about previous, current, and upcoming coding projects I have been involved with. If you are looking for code that has not been package-released, please see my GitHub. If you have any queries regarding current or upcoming projects, please feel free to reach out via email.


Upcoming Projects

  • UK Biobank Deconfounding
    Role: Lead Developer for Python Implementation, Project Status: Under Development (Coming Soon...), Language: Python.

    In collaboration with the FMRIB research centre, University of Oxford, I am currently working on a portable Python implementation of the UK Biobank deconfounding code that is currently used for inhouse FSL Biobank analyses.
    References: [1].

Current Software

  • Confidence Regions Toolbox
    Role: Lead Developer, Project Status: Initial Implementation Active, Language: Python.

    Despite its central role in interpreting results, very little attention is given to the variability in signal location in fMRI analysis. To address this, this project aims to provide tools for generating spatial confidence regions: regions which act as probabilistic bounds for the locale of observed clusters and excursion sets. Such regions provide useful insight into fMRI uncertainty quantification and increase the transparency of analysis results. Tools under active development include confidence regions for raw (%BOLD), and standardized (Cohen’s D) effect size images, as well as for conjunction inference.
    References: [2], [3], [4], [12], [13].

    Source: https://github.com/TomMaullin/crtoolbox, Demo: https://github.com/TomMaullin/crtoolbox_demo


  • BLMM: Big Linear Mixed Modelling
    Role: Lead Developer, Project Status: Active, Language: Python.

    Methods conventionally employed for drawing inference on neuroimaging data (e.g. Ordinary Least Squares regression, t-tests and F-tests) often cannot account for the grouping structures present in large datasets and, consequently, can produce erroneous inferences. An alternative approach is to employ the Linear Mixed Model (LMM) for analysis. The LMM is a flexible method applicable for analyzing longitudinal, heterogeneous or unbalanced clustered data. BLMM ("Big" Linear Mixed Models) acts as an efficient tool for performing large-scale fMRI LMM analyses, implemented in python and designed for use on SGE computer clusters.

    Source: https://github.com/TomMaullin/BLMM
    References: [10], [11].


  • BLM: Big Linear Modelling
    Role: Lead Developer, Project Status: Active, Language: Python.

    As datasets grow larger even performing a standard linear regression analysis is becoming increasingly demanding for image-based fMRI, with heavy overheads being placed on memory and computation time. In addition, variability in the patterns of missing data, especially near the edge of the brain, can become progressively problematic as sample sizes increase. To combat this, the BLM ("Big" Linear Models) toolbox provides an efficient way to conduct large-scale MRI GLM analyses, with parallel computing, whilst accounting for mask variability.

    Source: https://github.com/TomMaullin/BLM
    Reference: [11].

Previous Work

  • SwE-Toolbox
    Role: Former Contributor, Project Status: Active, Language: MatLab.

    The SwE toolbox is a toolbox for longitudinal and repeated measures neuroimaging data. It fits a simple "marginal model" with no need for per-subject dummy variables, and instead of iterative computation of variance components it uses the noniterative "sandwich estimator" to find standard errors. The core work is described in Guillaume et al. (2014), with improved small sample adjustments as well as a Wild Bootstrap for non-parametric inferences covered in Guillaume & Nichols (2015) and Guillaume (2015).

    Source: http://www.nisox.org/Software/SwE/
    References: [5], [6], [7].


  • IBMA Toolbox
    Role: Former Contributor, Project Status: Legacy, Language: MatLab.

    While most neuroimaging meta-analyses are based on peak coordinate data, the best practice method is an image-based meta-analysis that combines the effect estimates and the standard errors from each study. While the latter is the preferred approach in the statistical community, often only standardised estimates are shared, reducing the possible meta-analytic approaches. In response to the increasing availability of image data for neuroimaging analyses, the Image Based Meta-Analysis (IBMA) toolbox for SPM provides a range of tools for synthesizing image-based summary statistics drawn from multiple datasets.

    Source: https://github.com/NeuroimagingMetaAnalysis/ibma
    Reference: [8].


  • NIDM Results
    Role: Former Contributor, Project Status: Active (Specific tools worked on: Deprecated), Language: MatLab.

    The three major software analysis packages for neuroimaging, SPM (Statistical Parametric Mapping), FSL (FMRIB Software Library) and AFNI (Analysis of Functional NeuroImages), provide implementations of mass univariate analyses. While there are many commonalities across software packages, there are also software-specific outputs that can be of interest for the end-user. The NeuroImaging Data Model (NIDM) Results specification aims to provide a unified representation of neuroimaging results across analysis software. When a piece of information is only available in a specific software, software-specific extensions are provided.

    Source: http://nidm.nidash.org/specs/nidm-results_130.html
    Reference: [9].

References

  1. Alfaro-Almagro, F. et al. (2020). Confound modelling in UK Biobank brain imaging. NeuroImage, 224, 117002. https://doi.org/10.1016/j.neuroimage.2020.117002
  2. Bowring, A. (2019). Spatial confidence sets for raw effect size images. NeuroImage, 203, 116187. https://doi.org/10.1016/j.neuroimage.2019.116187
  3. Bowring, A., Telschow, F. J. E., Schwartzman, A., & Nichols, T. E. (2021). Confidence Sets for Cohen’s d effect size images. NeuroImage, 226, 117477. https://doi.org/10.1016/j.neuroimage.2020.117477
  4. Davenport, S., Nichols, T. E., & Schwarzman, A. (2022). Confidence regions for the location of peaks of a smooth random field. arXiv. https://doi.org/10.48550/arxiv.2208.00251
  5. Guillaume, B. (2015). Accurate non-iterative modelling and inference of longitudinal neuroimaging data. ULiège-Université de Liège [Médecine vétérinaire].
  6. Guillaume, M., & Nichols, T. E. (2014). Fast and accurate modelling of longitudinal and repeated measures neuroimaging data. NeuroImage, 84, 287-298. https://doi.org/10.1016/j.neuroimage.2014.03.029
  7. Guillaume, B., Nichols, T., et al. (2017). Non-parametric Inference for Longitudinal and Repeated-Measures Neuroimaging Data with the Wild Bootstrap. figshare. https://doi.org/10.6084/m9.figshare.5478229.v1
  8. Maumet, C., & Nichols, T. (2014). IBMA: An SPM toolbox for NeuroImaging Image-Based Meta-Analysis. Frontiers in Neuroinformatics, 8. https://doi.org/10.3389/conf.fninf.2014.18.00025
  9. Maumet, C., Auer, T., Bowring, A., et al. (2016). Sharing brain mapping statistical results with the neuroimaging data model. Scientific Data, 3(1), 160102. https://doi.org/10.1038/sdata.2016.102
  10. Maullin-Sapey, T., & Nichols, T. E. (2021). Fisher scoring for crossed factor linear mixed models. Statistics and Computing, 31(5), 53. https://doi.org/10.1007/s11222-021-10026-6
  11. Maullin-Sapey, T., & Nichols, T. E. (2022). BLMM: Parallelised computing for big linear mixed models. NeuroImage, 264, 119729. https://doi.org/10.1016/j.neuroimage.2022.119729
  12. Maullin-Sapey, T., Schwartzman, A., & Nichols, T. E. (2022). Spatial Confidence Regions for Combinations of Excursion Sets in Image Analysis. arXiv. https://doi.org/10.48550/arxiv.2201.02743
  13. Sommerfeld, M., Sain, S., & Schwartzman, A. (2018). Confidence regions for spatial excursion sets from repeated random field observations, with an application to climate. Journal of the American Statistical Association, 113(523), 1327-1340. https://doi.org/10.1080/01621459.2017.1341838