Turbo-BrainVoyager v4.4

Turbo-BrainVoyager
User's Guide

Turbo-BrainVoyager (TBV) is a highly optimized software package for real-time analysis and advanced visualization of functional and structural magnetic resonance imaging data sets enabling neurofeedback and other brain-computer interface (BCI) applications. Turbo-BrainVoyager allows quality assurance of ongoing fMRI measurements by assessing head motion and time course drifts, and by incrementally computing statistical maps as contrasts of a General Linear Model (GLM). Incrementally calculated (statistical) maps are visualized on original slices (multi-slice view) as well as in orthographic 3D mode using original slices as well as coregistered anatomical scans of the subject. Anatomical data sets of the subject can be prepared in TBV prior to the real-time functional measurement to enable visualization of maps and regions-of-interest (ROIs) in MNI space. With the help of BrainVoyager, maps and ROIs can also be visualized on cortex (mesh) representations. TBV allows to create ROIs on any brain view and plots time courses, estimated beta values and event-related averages dynamically. TBV also allows to use machine learning tools to classify and predict distributed patterns of activity ("brain reading"). Due to its powerful computational and visualization capabilities, TBV enables advanced real-time applications such as fMRI neurofeedback, brain computer interfaces (BCIs) and activation-driven adaptive experimental designs.

TBV and the new TBV EDU version provide a sophisticated data simulation and experiment preparation mode that can be used to prepare and test protocols, and to assess expected localizer and neurofeedback results before moving to the scanner room.

Turbo-BrainVoyager is based on the BrainVoyager software package with the following unique or adjusted features:

• Easy to use graphical user interface (GUI) with essential elements to control online and offline analysis.
• Incremental processing routines written in optimized C++ code or GPU shaders, including CPU and GPU-based 3D motion correction, and 3D spatial smoothing.
• Selection of ROIs using mouse selection or via prepared Volumes-Of-Interest (VOIs) in normalized (MNI or Talairach) space.
• Inspection of ROI data in time course plots, event-related averaging plots and dynamically estimated beta values at any time during as well as after fMRI scanning.
  
• Detrending of displayed time courses using incrementally added confounds to GLM design matrices, such as linear and non-linear drift predictors as well as 6 predictors filled with motion parameters as they are calculated online during incremental rigid-body motion correction.
  
• Percent-signal change transformation of multiple detrended ROI time courses that can be directly used as cleaned input for fMRI neurofeedback or other BCI applications.
• Design matrices, contrasts and other relevant statistical data are created automatically from a defined protocol.
• Robust incremental GLM statistical data analysis of block- and event-related designs with multiple conditions.
• Optimized routines for multi-voxel pattern classification (MVPC) using support vector machines (SVMs).
• Instead of using a pre-dermined sequence of condition events, real-time protocols make it possible to build up the main predictors of the design matrix on the fly during scanning supporting flexible designs that depend on behavioral responses or brain activation data of participants. 
• Interactive GUI that allows to explore incoming data while running the actual measurement, including selection of multiple contrasts as well as selection of conjunction of contrasts.
• Advanced volume and surface visualizations with fast statistical updates ("movie") by using transformation matrices to project calculated statistical data from/to original image space.
• Incorporation of Volumes-Of-Interests (VOIs) in MNI (or Talairach) space to (re-)load ROIs across sessions.
• Integrated neurofeedback module to display selected and processed brain activation data to subjects as time courses or thermometer-like displays, and to save ROI data incrementally for use with custom software providing feedback.
• Integrated letter spelling BCI tool (BOLD Decoder).
• Storage of fMRI raw data on local hard drive in BrainVoyager format after functional run has been completed allowing an easy transition for in-depth offline data analysis with BrainVoyager or other fMRI software packages.
• A powerful plugin interface that supports integration of custom computations during real-time processing; the plugin interface has access to raw and processed (e.g. motion-corrected) data, to ROIs and statistical maps that can be processed or exported incrementally to disk. 
• Support for real-time ICA analysis implemented as a plugin using the plugin map visualization functions.

Previously recorded runs can also be reloaded and inspected at any time. Turbo-BrainVoyager is optimized for real-time analysis and is not a replacement for BrainVoyager or other offline software. There are, for example, no routines for statistical group analyses. While anatomical document creation from DICOM files, anatomical preprocessing and MNI normalization is available since TBV 4.0, some of the advanced visualization features require processing (e.g. Talairach transformation, surface mesh creation) in BrainVoyager prior to real-time analysis in TBV. Turbo-BrainVoyager 4.4 runs on Microsoft Windows 10/11, Linux (2.6+ kernel), and macOS 10.15+.

The next section provides information how to setup TBV in a scanner network environment and how to prepare neurofeedback experiments. This is followed by an overview of how to use TBV. An in-depth description of its major features is then provided in subsequent topics. If you are new to Turbo-BrainVoyager, you may also want to run the provided sample data sets to make yourself familar with the software. It is also recommended to have a look at the release notes describing features added in this and previous versions.

Major analysis steps implemented in Turbo-BrainVoyager are described in the publication:

Goebel, R. (2021). Analysis methods for real-time fMRI neurofeedback. In: M. Hampson (Ed.). fMRI Neurofeedback, pp. 23-55. Academic Press.

Copyright © 2002 - 2024 Rainer Goebel. All rights reserved.