Magnetism

fMRI Processing Software

What is the best software for processing fMRI data?

Most major MR manufacturers offer basic integrated fMRI processing software (e.g., GE's BrainWave, Philips I View BOLD) as an optional purchase. These are probably adequate for the majority of simple task-based fMRI studies used for clinical cortex mapping, but more complex experiments will likely require third-party solutions. Many third-party software programs are free and open source. Some provide robust tool sets for experimental design, processing, and data analysis, while others are more "bare-boned". I won't take sides here as to which ones are the best, but I did grow up with SPM and am most familiar with it. Below is a summary of the most widely used independent fMRI software listed alphabetically.

AFNI (Analysis of Functional Neuroimaging) is an extensive free set of C-based programs for processing, analyzing, and displaying fMRI data. Originally developed by Robert Cox at Medical College of Wisconsin in the mid-1990s, AFNI is now hosted at the National Institute of Mental Health (NIMH) in Bethesda, MD. Add-on tools include a skull-stripping program and SUMA for cortical surface-based fMRI analysis. Runs on Unix, SGI, Solaris, Linux and Mac OS X. Available at this link.

Brain Connectivity Toolbox contains over 120 brain connectivity and statistical functions primarily designed for complex network (graphical) analysis, most widely used to analyze resting-state fMRI studies. This free site is the work of multiple investigators, primarily Olaf Sporns at Indiana University and Mikail Rubinov at Cambridge. MATLAB is required, although some functions will run in Octave. Available at this link.

BrainSuite is a collection of open source software tools for largely automated brain MRI processing. It was produced and distributed as a collaborative project between the research groups of David Shattuck at UCLA and Richard Leahy at USC. The major functionality is for extraction and parameterization of the inner and outer surfaces of the cerebral cortex and to analyze diffusion MRI data. It is available for 64-bit Mac OS X, Windows, and Linux. Available at this link.

BrainVISA, from the Institut Fédératif de Recherche n°49 (IFR), France, is a free neuroimaging software platform for mass data analysis. Works on Windows, Unix/MacOS, and LInux systems and interfaces with FreeSurfer. Customizable modules include toolboxes for brain segmentation, cortical sulci recognition and morphometry, interactive 3D visualization, and cortical surface parameterization and analysis. Available at this link.

Brain Voyager is a comprehensive suite of commercial programs for functional imaging analysis and visualization, including not only fMRI, but also DTI, EEG, and MEG. Add-on modules for TMS neuronavigation, fMRI biofeedback, and neural network simulation are also available. Based in the Netherlands, CEO Rainer Goebel originally developed this platform in the mid-1990s. It runs on all major computer platforms, including Windows, Linux and Mac OS X. It has excellent on-line manuals that explain the operation of the software and are quite educational about fMRI processing in general. Available at this link.

FreeSurfer is an open source software suite for processing and analyzing brain MR images. It was developed by Bruce Fischl and his team in the Laboratory for Computational Neuroimaging at the Harvard/MGH Martinos Center. The software is comprehensive and includes modules for skull stripping, image registration, segmentation, cortical topology, fMRI analysis, and tractography. PET data analysis can also be performed. The real power of this software lies in its cortical and subcortical analysis tools for measuring, inflating or flattening the cortical surface and performing multi-subject fMRI statistics thereon. FreeSurfer integrates (albeit somewhat clumsily) with FSL and the MNI's Brain Imaging Software Toolbox. It runs on Linux, Mac OS X, and Windows (via VirtualBox). An extensive Wiki and on-line tutorials are available. Here is the link.

FSL (FMRIB Software Library) is a free suite of applications from Oxford's Functional Magnetic Resonance Imaging of the Brain (FMRIB) laboratory. The first release of this software was in 2000, and FSL has continued to grow exponentially since that time. It is designed primarily for Mac OS X and Linux, but can be used on Windows in a Virtual Machine environment. Several dozen tool sets are available that run as separate programs. The most widely used modules include FEAT (a model-based analysis for task fMRI), MELODIC (a model-free ICA-based analysis for resting state fMRI), BET (for brain extraction), FAST (for tissue segmentation), and MCFLIRT (for motion correction). Tools are also available for diffusion tractography and analysis, as well as ASL-based fMRI and perfusion measurements. Many third party plug-ins are available. FSL can be accessed at this link.

MIALAB (Medical Image Analysis Lab), headed by Vince Calhoun at the University of New Mexico, has developed a set of free MATLAB-based software for brain imaging analysis. Their group's independent component analysis toolbox (GIFT) is the most widely used. Here is the link.

REST (RESting-state fMRI data analysis Toolkit) is a group of applications based on MATLAB and SPM8 for evaluation of resting-state fMRI data. It was conceived and designed under a team at the Beijing Normal University. The software offers data exchange/sharing with SPM, AFNI, and FSL. Its pipeline and flexible modules can perform a full range of preprocessing steps (e.g., slice timing, motion correction, detrending), but its real strength lies in analysis of RS-fMRI data (e.g., correlation-based functional connectivity, amplitude of low-frequency fluctuation analysis, Granger causality). The main software application is available at this link. REST is also embedded in a more comprehensive website with links to other RS and connectivity applications, The R-fMRI Network (rfmr.org).

SPM (Statistical Parametric Mapping) was created in the late 1980s by Karl Friston to analyze PET images at the voxel level. Within five years SPM became the dominant method for the analysis of PET data as well as for the (fledgling) technique known as fMRI. Today SPM is in its 12th release under continuous development from members and collaborators of the Wellcome Trust Centre for Neuroimaging at University College London. SPM is a free and works under Windows, Linus, and Mac OS X as long as core MATLAB is installed. SPM has virtually every feature one might need for fMRI processing, analysis, and display. Like most other fMRI software, SPM employs the General Linear Model (GLM) for its primary analysis, as Friston was one of the early proponents of this technique. The SPM website is found at this link.

Advanced Discussion (show/hide)»

Not included in the above discussion are specialized fMRI Paradigm Design Tools. These programs help the investigator plan the timing of events and data acquisition needed for the fMRI experiment. They then send appropriate signals to various audiovisual equipment and the MRI scanner to trigger the presentation of stimuli and recording of subject responses.

Basic paradigm design tools are provided by all major MR vendors as part of their standard integrated fMRI packages. Additionally, several of the third party fMRI analysis packages described above also contain design modules. Below is a list of several commonly used toolkits whose primary function is fMRI design:

E-Prime® is a commercial suite of applications from Psychology Software Tools, Inc. of Pittsburgh, PA. The software is widely used for non-MRI psychology and neurophysiology experiments, but also has extensions to allow its use for fMRI. A broad range of routine paradigms can be created a drag-and-drop graphical user interface, but more sophisticated experiments require using their proprietary scripting language (which fortunately, is nearly identical to Visual Basic for Applications). Further information is available at this link.

nordicAktiva is commercially available clinical stimulus and workflow software solution from the Norwegian-based company NordicNeuroLab. They offer not only design software but a complete package of visual, auditory, and response hardware for fMRI experiments. We have historically used this system for clinical eloquent cortex mapping studies. Writing complex new paradigms, however, is moderately difficult as it requires direct editing of the XML code. Further information is available at this link.

Presentation from Neurobehavioral Systems, Inc. of California, is perhaps the easiest to use stimulus delivery and experiment control program for neuroscience with a customizable graphical user interface. It is capable of simultaneously and independently presenting a wide range of stimuli, including images, 2D and 3D graphics, audio, and video. The main limitation is that the program works only under Windows. Further information is available at this link.

The above software is all commercially based, with individual user licenses ranging from several hundred to several thousand dollars. A few free, open-source solutions are available, including PsychoPy from the University of Nottingham (Website) and Psychtoolbox (Website). PsychoPy is written in Python while Psychtoolbox runs on Matlab or Octave.

References
    Ashburner J. SPM: a history. NeuroImage 2012; 62:791-800.
     Fischl B. FreeSurfer. NeuroImage 2012; 62:774-781.
Indrajit IK. Radiology websites: Functional MRI websites. Indian J Radiol Imaging 2014; 24:92-93.
     Jenkinson M, Beckmann CF, Behrens TEJ, et al. Review: FSL. NeuroImage 2012; 62:782-790.
   List of functional connectivity software. Wikipedia, the free encyclopedia. (Contains a fairly complete list of software for connectivity analysis useful for resting-state fMRI applications)
List of neuroimaging software. Wikipedia, the free encyclopedia. (Contains a fairly complete list of software for fMRI analysis, as well as others tangentially related to fMRI, such as EEG and DTI)
NITRC Resource Repository (NITRC-R). This is a major on-line collaboration environment for neuroinformatics tools and software, including not only fMRI, but PET/SPECT/CT/EEG/MEG/Optical imaging, computational neuroscience, and genomics.
     Rubinov M, Sporns O. Complex network measures of brain connectivity: uses and interpretations. NeuroImage 2010; 52:1059-1069. (methods used in the Brain Connectivity Toolbox).
     Song X-W, Dong Z-Y, Long X-Y, Li S-F, Zuo X-N, et al. (2011) REST: A Toolkit for Resting-State Functional Magnetic Resonance Imaging Data Processing. PLoS ONE 6(9): e25031. doi:10.1371/journal.pone.0025031
The FIL Methods Group. SPM12 Manual. Wellcome Trust Center for Neuroimaging, London, 2014.

Related Questions
How is fMRI data processed and analyzed?

← Previous Question

Next Question →

↑ Complete List of Questions ↑