BrainVoyager 22.0 Release Notes

New Features

 

DNN SegmentatorThis version introduces brain segmentation using a deep neural network (DNN) with an advanced "Tiramisu" architecture. This DNN Segmentator has been developed for high-resolution sub-millimeter data to segment grey matter with very high accuracy so that both inner (white-grey) and outer (pial) boundaries are obtained that can be used with minimal editing to construct corresponding cortex meshes. The high-quality segmentations are especially suited for cortical thickness measurements and mesoscopic (laminar and columnar) fMRI. The tool has been developed in Tensorflow using Python. The tool can be easily launched from the new "DNN Segmentation" dialog that runs the neural network via the embedded Python interface (see below). After the segmentation process has been completed, the "DNN Segmentation Postprocessing" dialog can be used to fine-tune the segmentation, to separate left and right hemispheres and to create standard segmented VMRs and reconstructed cortex meshes. For details see section "Deep Neural Network Segmentation" of the "Brain and Cortex Segmentation" chapter in the updated User's Guide. Also note that using this function requires an installation of Python 3.6 and TensorFlow 2.0, for details see section "Enabling Python" in the "General Information" chapter of the User's Guide.
BrainVoyager NotebooksThis release adds BrainVoyager Notebooks supporting reproducible data analysis by documenting performed analysis steps in the form of executable (auto-generated or self-written) Python code. BV Notebooks not only offer (script) programmers a means to write and document code but they also support reproducible analyses for non-coders by converting essential GUI actions into corresponding code. If, for example, the "GO" button in the "FMR Preprocessing" dialog is clicked, all selected preprocessing operations with (non-default) arguments are converted in BV Python code and added to the current notebook (if enabled). Note that code can be easily enriched with images, animations, rendered Markdown text as well as embedded BrainVoyager viewers so that users can create rich documents. For more details on how to use BrainVoyager Notebooks, see chapter "BV Notebooks" in the updated User's Guide.
Image AnnotatorThe new Image Annotator provides a streamlined approach to annotate images with text, rectangles, circles and arrows. The annotations are vector based and stored as overlays that can be edited at any time. The annotator can be called from BV Notebooks that store made annotations automatically for later editing. The annotator can also be used to annotate images in BV's "Image Reporter" that may be used as frames in created movie animations (see below). More details can be found in topic "Image Annotator" in the updated User's Guide.
Animation Player and Movie GenerationThe new "Animation Player" is available from the "Options" menu and can play simple movies that are either generated and saved automatically (e.g. during motion correction) or manually composed in the enhanced "Image Annotator". Movie animations are stored in Google's efficient "webp" format and can also be embedded in BV Notebooks (and web sites), e.g. to toggle before / after images that highlight (pre-)processing effects. More details can be found in topic "Animation Player" in the updated User's Guide.
Python PluginPython support is now enabled via the new "EnablePythonPlugin" plugin that is only loaded when Python is needed. BrainVoyager can now be launched on any system no matter whether (a supported) Python is available or not. Since Python becomes increasingly important for BrainVoyager, it is, however, strongly advised to install a supported standard Python version. Separating Python in a plugin also now allows to switch between multiple Python versions and environments (currently only 3.6 based environments) using the "Select Python On Disk" dialog. For details on how to setup one or more Python environments with support of Tensorflow for the DNN segmentation tool, see topic "Enabling Python" in the "General Information" chapter of the User's Guide.

 

Enhancements

 

Starting with NIfTI Files in Data Analysis ManagerIn previous releases, the "Data Analysis Manager" only allowed to start with the "Create Document Workflow" that generates NIfTI files from raw DICOM files. It was, however, not possible to start directly from NIfTI files. The new "Import NIfTI Document Workflow" now supports to use NIfTI files as input, which makes it, for example, possible to start with NIfTI data from public repositories; it also works well with NIfTI / BIDS data created with the updated "Create Document Wizard" (see next point). For more details, consult the "Import NIfTI Document Workflow" topic in the "Data Analysis Management" chapter of the User's Guide.
BIDS / NIfTI SupportThis release further improves BIDS support. The "Create Document Wizard" now allows not only to save a new dataset as a NIfTI file but to store it in the correct sub-folder of a specified (existing or new) project, for details consult the "Create Functional Documents" and "Create Anatomical Documents" topics in the "Getting Started" chapter of the User's Guide. Also when linking a VTC from a participants VMR file in a BIDS anat folder, the VTC dialog now opens in the corresponding "..fmr/" folder as default instead of the anat folder. Furthermore, import of anatomical NIfTI has been improved by now also reading (2 and 4 byte) integer and floating point values instead of only byte values. These formats are used to create both a VMR and a V16 dataset when importing NIfTI files; if possible, intensity values will not be rescaled which improves compatibility with atlas-related region (label) files.
Movie Creation with Image ReporterThe "Image Reporter" has been substantially improved to serve now as a composer of simple movies (animations) that can be stored in "webp" format and played in the new "Animation Player " (see above) or embedded in notebooks. Images can be easily added to "Image Reporter" by several keyboard shortcuts catching the whole program window, the current dialog or, most importantly, the view of the current document. To create a movie, all accumulated images must have the same size. Frames can also be annotated using the new "Image Annotator" before creating a movie. For more details, see topics "Image Cells", "Image Annotator" and "Animation Cells" in the "BV Notebooks" chapter of the User's Guide.
BV Python API and IDEPython integration and tools have been substantially enhanced in this release. The BrainVoyager commands have been rewritten, extended and simplified. Names of commands follow now Python style (e.g. correct_motion instead of CorrectMotion). In BV Notebooks, integrated documentation highlights function signatures as tooltips and a more elaborate description of parameters are displayed in the "API Help Output" pane when hovering with the mouse over commands or when typing commands. See the updated "Python Developer Guide" (WIP) for details.
VTC PreprocessingWhile the "VTC Preprocessing" dialog offered the possibility to run high-pass temporal filtering, only the FFT option was supported. The updated dialog now also supports the GLM-Fourier (default) and GLM-DCT high-pass filter options in the same way as is available for FMR-STC data. Note that the GLM-based high-pass filter procedures also remove linear trends since the created design matrix not only contains the sine/cosine basis functions but also a linear trend predictor.
VOI VMR FunctionsThe Volumes-Of-Interest dialog now is initially smaller. If expanded, it offers new VOI VMR tools including creation of intensity histograms and creation of surfaces from selected VOIs. The main dialog part offers a new Max Component operation that is useful to remove small sub-parts from a (segmented) VOI keeping only the largest connected part.
Zoom ViewThe Zoom View now shows the same voxel information panel when hovering across voxels than the VMR view including intensity, map and VOI details. This allows to inspect e.g. segmentation results and supports editing operations. In the context of the DNN segmentation tool, one can, for example easily study tissue probabilities and intensity values across tissue borders by hovering the mouse.
Calculate FDR in Maps DialogWhen a statistical volume map did not have a FDR table, the FDR approach for thresholding was not available. The only way to get a FDR table was to (re-)run / overlay a GLM. It is now possible to recalculate a FDR table from the map's data using the new "Calculate" button next to the "Use FDR" option in the "Thresholding" field of the "Statistics" tab of the "Volume Maps" dialog. Note that this only works in case that the map represents a supported statistic and has properly defined degrees of freedom.
Continous Curvature LUTA new curvature LUT has been added that changes continously from near white to dark grey when curvature changes from high convex to high concave. Values in the transition from convex to concave are colored mid-grey. Note that this color LUT does not highlight the transition from concave to convex as the standard curvature color but it provides a useful visualization, especially for pial meshes, where gyri are colored bright and sulci colored dark. Besides loading the new LUT, the "Background and Curvature Colors" dialog now also has an option to use the new LUT.
Import of FreeSurfer FilesIt is now possible to read essential FreeSurfer files (meshes, curv, annot and .mgz). Depending on the format, a file is converted into a BV VMR, SRF, SMP, or POI file. To import a anatomical file, the "Import Freesurfer MGH/MGZ File" item in the "Options" menu can be used. To import a mesh file, the "Binary Freesurfer File" in the "Import Mesh" sub-menu of the "Meshes" menu can be used. A mesh-compatible surface map can be imported by selecting the "Freesurfer 'Curv' files" files filter in the "Open SMP File" dialog (invoked by clicking the "Load" button in the "Surface Maps" dialog). A mesh-compatible annotation file can be imported by selecting the "Freesurfer Annot files (*.annot)" files filter in the "Select POI File" dialog (invoked by clicking the "Load" button in the "Patches-Of-Interest Analysis" dialog).
Auto-Scaling Voxel Beta Bar PlotsThe voxel beta bar plots showed data in a fixed value range in order to allow comparison of bar height when hovering the mouse across voxels. When however values exceeded the upper y axis range value, bars were clipped. This release updates bar plots by keeping a default or specified upper value as long as all displayed bars have smaller values. If at least one bar exceeds the threshold, the upper range value is automatically reset to a value a bit higher than the maximum value. This allows to interpret relative heights within the plot of a voxel and allows comparisons across voxel for most cases. If for example the maximum percent signal change value is set to 5.0 (default), most voxel's beta values will be smaller than this value.
PSC in Time Course PlotsWhen selecting regions with the mouse to invoked time course windows, the displayed values are shown as raw signal values instead of more useful percent signal change (PSC) values. If FMR-STC and VMR-VTC data refer to a protocol, the time course plots now show PSC values as default. The program uses non-defined time points and/or time points defined in a baseline condition to determine (time shifted) periods used to calculate the mean signal of the baselines. Note that the correct baseline setting (no baseline condition or 'Use first condition as baseline' or 'Use last condition as baseline') need to be made to get correct results, which is usually done when performing a "single-study GLM" or using the "Settings" dialog.
VOIs as MasksWhen running a spatial mask-based (e.g. brain or cortex-based) statistical analysis, a special MSK mask file can be provided. It requires however an extra step to calculate MSK, e.g. when one has defined a VOI file. It is now possible to directly use a VOI file (and a specified VOI inside the file) as a spatial mask when running linear correlation tests and (single study so far) GLMs.

 

Bug Fixes

 

POI Labelswhen POIs were shown with 3D text labels and then turned off in the POI Options dialog followed by a click on Hide POIs in the POI dialog, BV crashed. This issue has been fixed.
Outline POIsThe function to create outline (border) POIs from more conventional filled POIs did not work in the previous version. This issue has been fixed.
POI GLM and POI ANOVAThe functions to run POI GLM and POI ANOVA analyses could not be launched from the "POI Analysis Options" dialog in previous releases. These functions have been re-enabled in this release.
Probabilistic Maps from VOIsCreating probabilistic maps, including MPMs, from overlapping VOIs did not work in the previous release. This has been fixed.
Verification of Functional CoverageThe function to check the region of consistent overlap of functional (VTC) data ("Verify Functional Overlap" item in "Options" menu) did not properly work in the last version; there was for example some noisy coloring for the 100% overlap case and individual volume maps could not be visualized (unless the maps were saved and reloaded from disk). This issue has been fixed.