BrainVoyager QX 2.4 Release Notes

New Features

Automatic AC-PC and Talairach TransformationIn previous versions, the normalization of 3D anatomical VMR data sets into AC-PC and Talairach space required manual determination of the AC and PC landmarks, the specification of angles to orient the brain with the mid-sagittal plane (MSP) and finally the determination of the borders of the cerebrum. These tasks can now be performed automatically for most VMR data sets. The AC-PC - TAL transformatiol can be executed in in one step ("Auto-ACPC-TAL" button) or in two separate steps from the "Talairach" tab of the "3D Volume Tools" dialog. The automatic normalization not only saves time but also allows users to perform ACPC and Talairach transformation that may not have the neuroanatomical knowledge about the location of relevant landmarks. While other (e.g. template-based) automatic alignment tools will be available in the next release of BrainVoyager QX, the new automatic AC-PC/Talairach transformation tools will remain useful in case that rigid body brain normalization (AC-PC space) is desired, e.g. in the context of (TMS) neuronavigation or when a "true" Talairach procedure is requested. For further details, consult the "Brain Normalization" section in the "Basic (f)MRI Data Analysis" chapter of the User's Guide.
Relative Depth Cortex Mesh and Grid SamplingSeveral improvements in this version have the goal to allow optimal processing of high-resolution (sub-millimeter) functional data and associated anatomical data obtained from ultra-high field scanners (7 Tesla and beyond). This version introduces high-resolution sampling of funcional data within grey matter (GM) at arbitrary relative depth levels. Two versions are introduced that both are based on the results of advanced segmentation and cortical thickness calculation, preferentially at 0.5mm resolution. The first approach ("Meshes -> Cortical Depth Sampling" menu) creates a series of meshes at different relative depth levels from a mesh reconstructed at the surface running through the middle of the WM-GM and GM-CSF boundary. These created meshes are then used to sample the functional data at the respective depth values. For even more precise sampling of small folded cortical regions, a high-resolution grid sampling tool is available ("Volumes -> High-Resolution Cortex Grid Sampling" menu) allowing to sample functional data at arbitrary relative cortical depth values on regularly spaced two-dimensional grids providing precise detailed topographic information. For further details, consult the "Tools for High-Resolution Data" chapter of the User's Guide.
EPI Distortion Correction (New Plugin)This functionality makes it possible to apply EPI distortion correction using the pixelshift algorithm (Jezzard & Balaban, 1995) for images that are geometrically distorted due to the susceptibility artifact. This approach requires fieldmaps (either phase and magnitude or real and imaginary), being acquired before the functional run. Secondly, a difference image of two FMRs can be created to evaluate the undistortion/unwarping result visually. Finally, some calculations can be made as well with this plugin, for example to compute the phase encoding bandwidth or to convert between radians and degrees; also Cartesian (real and imaginary) fieldmaps can be converted to polar coordinates (phase and magnitude). Please consult the manual (PDF) for usage and background information.
GLM/GLS Modeling ToolThis tool offers the possibility to create and edit custom design matrices allowing to implement specific models that are not available in the provided ANCOVA module. It also can be used as a teaching tool to better understand existing models. The tool performs standard (ordinary least squares, OLS) GLM estimation on provided (ROI) time course data and also allows to specify the error covaraince matrix for generalized least squares (GLS) estimation.
Image ReporterThe "Image Reporter" serves as a blackboard to store and visualize images. The stored images may be posted from computational routines to present intermediate and final processing results that allow to better understand ongoing processes and to evaluate the performance of intermediate calculations. The tool is also available for plugin writers as a quick means to visualize bitmaps.

 

Enhancements

CBAv2.4.2. If a CBA run has been completed, the program now automatically creates group files that can be used to average aligned curvature maps (.CAL file) and aligned shapes (.SAL file), i.e. without the need to manually select individual files. These files are also directly applied at the end of CBA, i.e. the program shows a folded mesh as the result of group-aligned shape averaging with an overlay of group-aligned curvature maps. The created mesh and curvature information can be directly inspected to judge the quality of cortex-based alignment.
New Scripting Commandsv2.4.1. It is now possible to perform temporal high-pass filtering (drift removal) using the GLM approach using Fourier or discrete cosine transform (DCT) basis functions. In previous versions, only the FFT-based high-pass filtering was available. The new commands are "TemporalHighPassFilterGLMFourier()" and "TemporalHighPassFilterGLMDCT()" with one parameter that specifies the number of cycles (pairs of two basis functions) used to build an appropriate design matrix. The installed script "HighPassFilterUsingGLM.js" shows how to use the new commands. It is now also possible to specify a bounding box for VTC creation in any supported target space using the properties "UseBoundingBoxForVTCCreation" and "TargetVTCBoundingBox[XYZ]Start" and "TargetVTCBoundingBox[XYZ]End"; consult the "CreateVTCfiles.js" script for details. Furthermore the command "SaveVTC()" has been added allowing to save, for example, to save a VTC file after linking a protocol with the "LinkStimulationProtocol()" function. The "SaveVTC" command accepts a string (file name) as input; if the string is empty (""), the name of the current VTC file will be used, i.e. the VTC file will be ovewritten (see the example script "LinkPRTSaveVTC.js"). Other new scripting commands allow to interrogate information about the running BrainVoyager version including the build number and whether the program is running in 32 or 64 bit mode. The installed script "VersionScript.js" shows how these commands can be used
Custom Default Scripts Folderv2.4.1. Scripts are installed in a standard location within the user's "Documents" folder ("BVQXExtensions/Scripts"). For some scenarios, it would be beneficial if scripts could be accessed from a custom folder as default, i.e. when written scripts are made available to members of a research group in a shared network folder. For such scenarios it is now possible to change the default scripts folder in the "Scripts" tab of the "Global Preferences" dialog.
Splitting VTCs and MTCsv2.4.1. The possibility to split VTC and MTC files has been added. This feature may be sometimes helpful, e.g. for creating sub-time courses for cross-validation purposes. Besides creating 2 or more partial time course data sets, the protocol of the source VTC/MTC file is also splitted accordingly and linked to the respective VTC/MTC split data sets (at present, this only is supported for protocols with "Volumes" resolution). The "Split VTC" dialog is availabe in the "Link 3D Volume Time Course (VTC)" dialog, and the "Split MTC" dialog is available in the "Mesh Time Courses" dialog.
POI Details and Peak Vertex Dialogsv2.4.1. The "POI Details" dialog now shows p values of available maps next to the map (e.g. t) value; the x/y/z coordinates are no longer in internal BV coordinates but in Dicom/Talariach convention. The "Peak Vertex" table shows now also Talairach coordinates for the respective peak vertex positions in case that the underlying VMR is in Talairach space. Besides showing a peak vertex table, the "Peak Vertex" option always also created a new set of POIs automatically containing the peak vertices extracted from larger POIs. Although the new POIs were saved under a new name, the user had no control over this step. In this version, a new option (default: off) allows to decide whether new peak vertex POIs should be created.
AR(2) ModelSerial correlations need to be removed from measured fMRI time course data in order to get unbiased statistical results. While the method used in previous versions to remove serial correlations (pre-whitening with a voxel-wise first-order autoregressive (AR(1)) model) works well, it has been recently shown that a second-order (AR(2)) model outperforms all tested approaches currently used in fMRI. The AR(2) model is now available as the default approach to remove serial correlations from residual time courses. For more details, consult the User's Guide.
Baseline Z-Standardization for RFX AnalysisWhile z standardization of time course data is a useful step to normalize the influence of different subjects when performing RFX analyses, the variance during baseline episodes is an even better measure for normalization since it leads to better RFX analyses that are comparable to mixed effects (MFX) analyses incorporating the variability of beta estimates at the fist level. The baseline z-transformation feature was added for whole-brain analyses since it was only available for ROI analyses in previous versions. In case that not enough baseline points are available, the baseline z normalization step will automatically be switched to a standard z normalization (reported in the Log pane).
ANCOVA ImprovementsIn previous versions, the beta values from the baseline condition was included when starting from a GLM data file. While usually ignored, the baseline could in principle be used as a level of a within factor. Since this is not optimal for running ANOVA models, the baseline beta is now dropped when opening a GLM file in the "ANCOVA" dialog. When running a 1 within-subjects factor design for ROI data, the cell means are now plotted as part of the generated results. Furthermore, the simple "correlation of covariate with dependent variable" design is now better supported: In case of one dependent variable per subject (no full within factor) and no between-subjects factor specified, one can add covariates (w of simple Correlation model.
Creation of POIs from SMP Map ClustersIt is often useful to get a list of regions-of-interest (ROI) automatically from a calculated individual or group map. While this function is available for volume data, it was missing for surface data. This version introduces this possibility for the currently selected map. Clicking the "Convert" button in the "Create POIs from map clusters" field in the "Advanced" tab of the "Surface Maps" dialog creates a list of POIs from surface map clusters that pass the current map threshold as well as a specified cluster area threshold in order to avoid inclusion of many small active patches.
POI Map Peak TableTo summarize results, it is often useful to create a table containing the location of clusters providing the coordinates of the "peak" voxels within each cluster. While this feature is available in volume space, it was missing for surface data. This versin introduces this feature for patches-of-interest (POIs) in the "POI Map Peak Vertices" field in the "POI Functions" tab of the "POI Analysis Options" dialog.
VOI TransformationsIt is now possible to directly apply spatial transformations to VOIs including rigid transformations (e.g. from native to ACPC space or back) and from ACPC to TAL space or back. The transformation tools are available in the new "Transformations" tab of the "VOI Analysis Options" dialog.
ACPC / TAL Framing SpaceIt is now possible to run rigid spatial transformations (e.g. ACPC transformation) and Talairach transformation in dimensions and voxel resolutions that differ from the default 256 framing space; this is especially useful for high-resolution data, e.g. with 0.5 mm voxels in a 512 framing space; if transformation matrices or Talairach landmarks are defined in a (resampled) 256 space, they will be automatically valid also in a higher resolution space (and vice versa). This is supported by storing information in spatial transformation (.TRF) and Talairach landmark (.TAL) files about the framing space and voxel resolution used during definition of a transformation (see also next point). This version also adds support for using a framing space with 768 voxels per dimension.
Sub-MM VTC SpacesThe possibility has been added to create VTCs in arbitrary (e.g. sub-millimeter) resolution for ACPC, TAL and sub-TAL (bounding box) space. The final resolution of a created VTC data set is determined by the resolution of the used native space VMR file and the relative resolution with respect to this file, i.e. if the relative resolution is "1", the VTC file will have the same resolution as the used VMR file. FMR-VMR alignment and ACPC / Talairach transformation may be performed in 1 mm space and applied to any other (e.g. higher) resolution (see point above). For details, consult the topic "Creation of High-Resolution VTCs" in the chapter "Tools for High-Resolution Data" of the User's Guide..
Intensity Inhomogeneity CorrectionThe intensity inhomogeneity correction (IIHC) tools have been enhanced. While the 16 bit results of IIHC were kept in working memory, they were not saved to disk for later use; in this version the option "Save resulting IIH corrected .V16 data to disk" has been added that is turned on as default, i.e. V16 results are automatically saved after IIHC if not turned off. The mapping from 16 to 8 bit data and the IIHC procedure work now robustly also in case that the input V16 data contains extremely high (outlier) values. The proton-density based pre-IIHC step has been improved allowing to scale the resulting data and it now stores resolution information in the resulting VMR/V16 data also in case that no mask VMR is provided.
EEG/MEG Analysis in Continous ModeIn the EEG-MEG module, a continous EEG/MEG data preparation (including artifact detection and time-frequency transform) is now possible allowing to perform cortical source distributed EEG/MEG analysis on arbitrary data segments (starting at any given time point specified in the protocol) or even on the entire CTC (if no protocol is specified). Before processing, the program performs a number of checks on the current settings to prevent unwanted huge file generation and endless calculations. In this "continous" mode, the resulting ACT files will contain not the event-related averages but the average power spectrum density (PSD) of the analyzed segment, and the cortical source analysis can be performed on the desired spectral (ACT) or spectro-temporal (TFD) measurements without specifying a target and a baseline interval, an important option for studying, e. g., resting-state and block-design EEG and MEG experimental data. For further details, consult the User's Guide.
EEG/MEG Channel Configuration Editor (Plugin Update)In order to create, update and synchronize the channel configuration data (CCD) and the surface head (SFH) data, the old plugin "EEG-MEG Channel Configuration Update" (available from the Plugins menu, in the EEG-MEG plugins submenu) has now evolved to a fully-fledged 3D digital point and EEG configuration editor, that can be used to create new SFH files from scratch or modify existing ones (e. g. from standard configuration). In addition, it is possible to import digitized point configurations from a Polhemus device (if exported as ASCII "DAT" files).
Parameter-Assisted Selection of Temporal ICA Components (Plugin Update)Selection of EEG and MEG temporal ICA components is not automatic. After the extraction step, all ICA components are now listed in a table and assigned with an individual check box to be used for retaining or not a component during the selection step. Besides the "Retain" column, useful spatial and temporal parameters are displayed in the same table for aiding the selection process. These include the spatial kurtosis, the time 1-lag autocorrelation, the spectral gradient and the maximal correlation with reference channels (if provided), all scaled to their z-scores over the entire set of components to quickly identify outliers. For further details, consult the EMEG chapter in the User's Guide.
Removal of BCG artifacts from EEG with Non-Linear Time Warping (Plugin Update)When analyzing EEG data acquired simultanouesly with fMRI, the BCG artifacts are traditionally subtracted within a fixed time interval. However, besides the magnitude and the shape, the BCG is poorly stable in its duration as well, i. e. the ideal time scale is also variable across all detected BCGs. In order to correct for the different durations, a non-linear approach is now available, that applies a non-linear time warping (NLTW) procedure to equalize the duration of all BCGs prior to their subtraction. After NLTW and subtraction, the original time scale is restored and the BCG-clean data segments are replaced in the data .For further details, consult the EMEG chapter in the User's Guide.
Shortest Paths on Folded MeshesIn order to measure distances between vertices on a mesh, the paths created in previous versions were not necessarily shortest paths. With this release an optimized shortest path calculation (based on the Dijkstra algorithm) has been added that allows measurement of distances on folded meshes. After Ctrl/Cmd-clicking on two vertices, the "Shortest Path" dialog can be invoked from the local conext menu. By default, the calculated shortest path is represented as a (line) POI and its distance is written to the Log pane and can be also recalculated in the updated "Edit Patches-Of-Interest (POI)" dialog. For additional options, consult the "Shortest Path Creation" topic in the "Miscellaneous Tools" chapter of the User's Guide.
CBA Special ToolsPairwise distance analysis for single-vertex (e.g. peak map vertex) POIs. Changed pairwise distance analysis of (peak map) single-vertex POIs to deviation from mean vertex as method to assess CBA funcROI improvement objectively, i.e. without thresholding and size issues.
Curvature-Flow SmoothingMean curvature-flow smoothing is available for smoothing initial reconstructed ("voxelated") folded mesh representations. This smoothing technique largely avoids (small) shrinkage of gyri and sulci that occurs when the standard smoothing approach is used. While the difference between the two available smoothing techniques is small, the curvature-flow smoothing is recommended when using surface meshes that sample high-resolution (e.g. in the range of 0.5 - 1.5 mm) functional data. The "Reconstructed Mesh Smoothing" dialog can be invoked from the "Meshes > Reconstructed Mesh Smoothing" menu item.
Reframing VMPsFor some processing steps, VMPs need to be in a common space with the same bounding box. The new "Reframe VMPs" dialog ("File > Reframe VMPs" menu) allows to put a set of selected VMP files in the same specified target bounding box. Note that the VMP files will not be spatially transformed, i.e. they need already be in the same target space (e.g. ACPC or TAL space) but with different bounding boxes.
Fiber TableIt is now possible to add fibers ("Add .FBR" button) to existing fibers in the "Fiber Table" dialog.
Skyra Mosaic DicomsThe Dicom header reading code for a Siemens specific part has been updated to handle the slightly modified headers of the syngo software (VD11) coming with Skyra and othe rnew scanner models.
V16 Plugin AccessNew commands have been added to the plugin API allowing direct (pointer) access to V16 data sets. The commands also include detaching, loading and saving V16 data sets as well as a convenience function to convert 16-bit data into 8-bit (VMR) data.
Bounding Box ToolsThe possibility has been added to quickly define, visualize and save values for bounding box definitions; using the "Set Bounding Box" entry in the context menu, an initial bounding box can be defined from the position of the cross in one of the orthographic VMR sub-views and fine-adjusted using the bounding box spin boxes in the "Segmentation" tab of the "3D Volume Tools"; a defined bounding box can be saved to disk as a ".bbx" file and loaded at other places such as the "Create VTC Options" dialog and the "Reframe VMPs" dialog. This enables, for example, creation of consistent bounding boxes for each participant in a group study and is especially useful in the context of (high-resolution) non-whole brain measurements.
Rearranging TabsThe tabbed documents (in the default "tabbed view mode") can now be rearranged simply by dragging the tab bars with the mouse to the left or right.

 

Bug Fixes

ANCOVAv2.4.2 There was an issue with the 2-within, 1-between factor model in previous releases: In case that the two within factors had different number of levels, the program could crash or produce incorrect results. This issue has been fixed. When using the covariate-contrast correlation model for ROIs, the program would crash. This has been fixed. In this correlation model the names of beta values were shown as "beta1", "beta2" etc. instead of names from the protocol-based predictors. This issue has been corrected.
Curvature for CBAv2.4.2 Version v2.4.0 introduced a new, mathematically more precise, calculation of mesh (mean) curvature that was also used for calculating curvature information for folded SPH meshes as a prepaaratory step of cortex-based alignment (CBA). While CBA works fine with these new curvature maps, the final quality of alignment is less good as when using the previous curvature calculation since that produced less noisy results than the new, more locally operating, curvature calculation. To get optimal CBA results, the curvature calculation used in the context of CBA has been reverted back to the old calculation. Both curvature methods are now also available in the "Background And Curvature Colors" dialog where the old method is available by using the "Fast" option (turned on as default).
Residuals VTC Optionv2.4.2 When one would turn on the option to save VTC files containing the residuals of an estimated single-study or multi-study/multi-subject GLM, the setting was kept active also in all subsequent GLM calculations of a session. The settting is now turned off after each launched GLM and needs to be explicitly turned on if it should be used in subsequent analyses.
ROI to VOI Conversionv2.4.2 When converting ROIs to VOIs in the "Region-Of-Interest" dialog of a FMR project, the position of the resulting VOIs could be displaced. This issue has been fixed.
High-Pass FFT Filter in Hzv2.4.1 When using the (non-default) FFT-based method for high-pass filtering, one can specify the cut-off frequency in Hz (instead of number of cycles). Before running the filter, this value should internally be converted to cycles but this conversion was missing in previous versions in GUI mode (it worked fine when used via scripting). This issue has been fixed.
GLM with AR(1) Modelv2.4.1 When switching from the default AR(2) model for serial correlation correction to the AR(1) model when running a single-study GLM, the program would crash. This issue has been fixed.
Auto-ACPC Stepv2.4.1 When performing only the automatic ACPC transformation step (using the "Detect MSP, AC, PC" button in the "Talairach" tab of the "3D Volume Tools" dialog), the correctly calculated and stored transformation file was not applied to the current VMR file for proper visualization. This issue has been fixed.
Find AC/PC Dialogsv2.4.1 When manually specifying Talairach reference points using the "Find AC" and "Find PC" dialogs, the program sometimes crashed, especially when switching to other programs and back. This issue has been fixed.
Moving POIs Up/Downv2.4.1 When moving POIs up or down using the respective icons in the "Patch-Of-Interest Analysis" dialog, the POI definitions sometimes were messed up and the program could crash. This issue has been fixed.
Rigid CBA File Namev2.4.1 When running the rigid alignment step prior to CBA, the produced ".rga" output file name contained a concatenation of all included subject initials; when using many subjects, this could lead to very long file names that may not be handled by the operating system; to avoid this issue, the file names now use a shorter string ("Group_N-[No. of subjects]_LH[or RH].rga") in case that 10 or more subjects are used during rigid CBA.
Pluginsv2.4.1 When starting the BLM plugin, a crash could occur. This issue has been fixed. Minor improvements are also implemented in the "SogICA" plugin (issue with unusual temporal similarity values) and the "ClusterThresh" plugin (consistent calculation of p values for correlation maps).
RFE with Unequal Number of Trialsv2.4.1 When running recursive feature elimination (RFE) from the "Multi-Voxel Pattern Analysis" dialog, the program could crash in case that the two classes contained an unequal number of trials. The present RFE version only supports two classes with an equal number of trials. In order to prevent the crash, the program now detects unsupported scenarios and presents an appropriate message.
Coordinates in Time Course Dialogsv2.4.1 Coordinates of selected voxels or VOIs were displayed in "ROI Time Course Options" dialogs as Talairach coordinates even in case that the respective VMR/VTC data was in native reference space. A similar problem also happened in the "VOI Details" table. These issues have been fixed.
Calling GLM from Matlab/COMv2.4.1 When calling the "ComputeMultiStudyGLM()" command via the COM scripting interface on Windows (e.g. from Matlab), the script halted with an error message "Unable to open GLM for writing!". This issue has been fixed.
Displaying Fibersv2.4.1 When loading or adding fibers from the "Fibers Table" dialog, fibers were not immediately displayed. This issue has been fixed.
Referenced Volume for MCThe reference volume that will be used during FMR motion correction can be selected in the "3D Motion Correction Options" dialog. After motion detection and correction, the motion correction procedure also saves a (confound) design matrix (SDM) file that can be included in subsequent GLM analyses in order to potentially remove residual motion artefacts. In case that the specified reference volume was not the first volume (default), the motion parameters were not set to 0.0 at the corresponding time point in the .SDM file in previous versions. This issue has been fixed.
Export of PSC Time CourseThe "ROI Signal Time Course" window allows to calculate the mean percent signal change of the plotted time course with respect to a reference function that can be specified in the "Display reference time course" field. The full percent-signal change (PSC) transformed time course can be exported with the "Export TXT" function but the exported values were slightly scaled due to rounding of reference percent signal change values. This issue has been fixed and the percent signal change values are now precisely calculated with respect to the mean value of all epochs with 0.0 values in the reference function.
Volume RendererThe real-time volume renderer has been updated to increase overall stability. Specific bugs that have been removed are crahes that could occur when clicking the "Remove" button in the list of masks or in the list of color definitions. v2.4.1 A display bug on Windows has been fixed.