SPM 25: open source neuroimaging analysis software
Tim M. Tierney, Nicholas A. Alexander, Nicole Labra Avila, Yaël Balbastre, Gareth Barnes, Yulia Bezsudnova, Mikael Brudfors, Korbinian Eckstein, Guillaume Flandin, Karl Friston, Amirhossein Jafarian, Olivia S. Kowalczyk, Vladimir Litvak, Johan Medrano, Stephanie Mellor, George O’Neill, Thomas Parr, · 2025
Read SPM 25: open source neuroimaging analysis software by Tim M. Tierney, Nicholas A. Alexander, Nicole Labra Avila, Yaël Balbastre, Gareth Barnes, Yulia Bezsudnova, Mikael Brudfors, Korbinian Eckstein, Guillaume Flandin, Karl Friston, Amirhossein Jafarian, Olivia S. Kowalczyk, Vladimir Litvak, Johan Medrano, Stephanie Mellor, George O’Neill, Thomas Parr, (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-12081.txt.
Introduction
Tim Tierney and a large international team announce SPM 25, the newest release of the open source neuroimaging analysis software. Statistical Parametric Mapping offers integrated methods for testing hypotheses about brain structure and function from scanner data. The package has been in continuous development for more than thirty years by a worldwide community.
Release 25.01 is a major version. It incorporates novel analysis methods and optimizes existing ones. Practices for open science and software development are improved as well. The paper reports what changed and why it matters for researchers who rely on the toolbox. A mature, openly developed platform underpins reproducibility across thousands of brain imaging studies.
Catalog record
| CREATOR | Tim M. Tierney, Nicholas A. Alexander, Nicole Labra Avila, Yaël Balbastre, Gareth Barnes, Yulia Bezsudnova, Mikael Brudfors, Korbinian Eckstein, Guillaume Flandin, Karl Friston, Amirhossein Jafarian, Olivia S. Kowalczyk, Vladimir Litvak, Johan Medrano, Stephanie Mellor, George O’Neill, Thomas Parr, |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 9 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-12081
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
SPM 25: open source neuroimaging analysis software Tim M. Tierney, Nicholas A. Alexander, Nicole Labra Avila, Yaël Balbastre, Gareth Barnes, Yulia Bezsudnova, Mikael Brudfors, Korbinian Eckstein, Guillaume Flandin, Karl Friston, Amirhossein Jafarian, Olivia S. Kowalczyk, Vladimir Litvak, Johan Medrano, Stephanie Mellor, George O’Neill, Thomas Parr, Adeel Razi, Ryan Timms, Peter Zeidman Abstract 1 Summary Statistical Parametric Mapping (SPM) is an integrated set of methods for testing hypotheses about the brain’s structure and function, using data from imaging devices. These methods are implemented in an open source software package, SPM , which has been in continuous development for more than 30 years by an international community of developers. This paper reports the release of SPM 25.01 , a major new version of the software that incorporates novel analysis methods, optimisations of existing methods, as well as improved practices for open science and software development. 2 Statement of need SPM introduced many of the statistical foundations that underpin cognitive and clinical neuroimaging research today, including: • The voxel-wise application of General Linear Models (GLMs) to neuroimaging data ( [ ref-friston1994statistical ] ). • Convolution modelling of functional MRI (fMRI) signals using haemodynamic response functions ( [ ref-friston1994analysis ] ). • Correction for multiple comparisons using topological inference (Random Field Theory, RFT) ( [ ref-worsley1996unified ] ). • Event-related fMRI ( [ ref-josephs1997event ] ). • Voxel-Based morphometry (VBM) for detecting changes in anatomy ( [ ref-ashburner2000voxel ] ). • Dynamic Causal Modelling (DCM) for state-space modelling using variational Bayesian methods ( [ ref-friston2003dynamic ] ). • Source localisation for M/EEG data using variational Bayesian methods ( [ ref-phillips2005empirical ] ). These methods share certain key principles: the use of generative models, the application of well-motivated parametric statistics and a commitment to open science practices. They are included in a major new release of SPM , which addresses a series of needs in the neuroimaging community, set out below. 2.1 Open development SPM was previously developed and tested using a private Subversion server within University College London. To enable community engagement in the future development of SPM and to increase transparency, development has recently moved to a public GitHub repository . SPM 25.01 is the first release of the software following the move to GitHub. The key advantages of using GitHub thus far have been: • Introducing automated unit and regression tests across platforms. • Automating the build process to conveniently generate and release source code and compiled versions. • Issue tracking and distributing tasks among developers. 2.2 Documentation and training The documentation for SPM was previously spread across multiple locations, most of which could not be edited by the community. SPM 25.01 is accompanied by a new documentation website , the source code for which is hosted in a public GitHub repository . The new website has step-by-step tutorials on all of SPM’s main features, as well as freely available video recordings of lectures from previous SPM courses covering the mathematical theory. 2.3 Major new features SPM 25 includes 10 years of new developments since the last major release ( SPM 12 , dated 1st October 2014). This section highlights some of the most significant new features for different neuroimaging modalities. 2.3.1 MRI • Multi-Brain Toolbox ( [ ref-brudfors2020flexible ] ). Generates population average-shaped brains, enabling more precise spatial normalisation with the option to automatically label brain structures ( [ ref-yan2022factorisation ] ). • SCOPE Toolbox. Generates voxel displacement maps (VDMs) using phase-encode-reversed pairs of MRI images (blip-up and blip-down images) to correct geometrical distortion in MRI ( [ ref-andersson2003correct ] ). This is similar to the Topup toolbox in FSL. 2.3.2 M/EEG • Methods for spectral decomposition - SPM 25.01 offers an implementation of an existing approach called FOOOF (Specparam) in the MEEGtools toolbox, based on code from Brainstorm ( [ ref-donoghue2020parameterizing ] ), as well as a new Bayesian implementation that introduces formal statistical testing, called Bayesian Spectral Decomposition (BSD) ( [ ref-medrano2024bsd ] ). • Support for fusion of different MEG sensor types and EEG sensors in beamforming with pre-whitening ( [ ref-westner2022unified ] ). • Support for MEG BIDS for specification of events, channels and fiducials ( [ ref-westner2022unified ] ). • Proof-of-concept routines for fusing M/EEG and fMRI data under a unified physiological model, to investigate neurovascular coupling ( [ ref-friston2019dynamic ] ; [ ref-jafarian2019neurovascular ] ). 2.3.3 OPMs A major recent innovation in neuroimaging is MEG using Optically Pumped Magnetometers (OPMs), which enable free movement of the head and body during neural recordings ( [ ref-boto2018moving ] ). This makes MEG available to new experimental paradigms (e.g., experiments involving free movement ( [ ref-mellor2023real ] )), new study populations who may not be amenable to traditional MEG (e.g., people with epilepsy ( [ ref-mellor2024detection ] )) and recording of other biomagnetic fields (e.g., from the spinal cord ( [ ref-spedden2024towards ] )). Developing analysis tools for OPM data is a major focus for SPM, with recently added features including: • File IO for all major OPM manufacturers (Quspin, Cerca, Mag4Health, Fieldline). • Methods to simulate arbitrary OPM arrays of differing densities and vector measurements. • OPM interference cancellation algorithms for low channel systems: Homogeneous Field Correction ( [ ref-tierney2021modelling ] ). • OPM interference cancellation algorithms for large channel systems: Adaptive Multipole Models ( [ ref-tierney2024adaptive ] ). 2.3.4 Bayesian statistics • Parametric Empirical Bayes (PEB) ( [ ref-friston2016bayesian ] ) extends the Dynamic Causal Modelling (DCM) framework to include random effects modelling of neural connectivity parameters, enabling people to test hypotheses about the similarities and differences among research participants. • Bayesian model reduction (BMR) ( [ ref-friston2018bayesian ] ) enables statistical evidence to be rapidly scored for large numbers of competing models, where models differ only in their priors. 2.3.5 Behavioural modelling In addition to neuroimaging analysis, SPM includes a suite of tools for behavioural modelling, including a comprehensive repository for computational neuroscience using the Active Inference framework. The code in SPM 25.01 has undergone significant development, offering a range of demonstrations accessible via the SPM DEM toolbox and associated GUI, and detailed in an accompanying textbook ( [ ref-parr2022active ] ). The key features are: • A series of inversion schemes for generative models based upon Partially Observable Markov Decision Processes (POMDPs) that can be used to simulate sequential choices, decision making, and planning ( [ ref-friston2017active ] ). • An active (generalised) filtering scheme for numerical simulation of continuous movement behaviour and responses to continuous sensory signals, e.g., ( [ ref-friston2015duet ] ). • Options for hierarchical composition of the above models ( [ ref-friston2017graphical ] ) and composition with a range of other models (e.g., speech recognition ( [ ref-friston2021active ] )). • Routines to fit the above models to behavioural data ( [ ref-schwartenbeck2016computational ] ). 2.3.6 SPM without MATLAB Approximately 90% of the SPM 25.01 source code is written in MATLAB and the remainder is written in C. This code has been highly optimised and thoroughly tested over 30 years of development. We have therefore carefully considered how to capitalise on the stability of the SPM software, while making it more accessible for people who do not have access to a MATLAB license, or who prefer to write their analysis