NeuroElf Features

While this list might not be complete, it tries to give an overview of what the NeuroElf toolbox can do (and, to some extent, what it cannot do), linking to other wiki pages containing information on how the user can achieve the required tasks with the toolbox.

Next to the purely file-operation (xff class to read and write file formats commonly used in fMRI data analyses) functionality, features can be mainly split into two categories: analysis (which includes any kind of processing and manipulation of data, such as preprocessing, filtering, etc.) and result generation and visualization (which is at the “far end” of the pipeline between coming up with a project/study idea and submitting a paper).

• limited capabilities for data import with functions dcm2nii (DICOM to NIftI import via SPM5/8 code) or dicom2nii (import without SPM code, still not fully functional), createfmr (DICOM to FMR import, not all types of DICOM images supported), createvmr (DICOM to VMR import, not all types of DICOM images supported)
• fMRI quality control with fmriquality (computation) and fmriqasheet (assessment)
• batched preprocessing of fMRI data (re-using SPM5/SPM8 code) with spm5_preprojobs
• attempting to correct for magnetic field inhomogeneities (intensity bias field) using the pmbfilter function or the VMR::InhomogeneityCorrect method
• reconstructing a segmentation as a surface using the VMR::DBReco method to generate a BrainVoyager QX SRF object
• morph (smoothing, inflation, sphere-morph) surface objects using the SRF::Morph method
• SPM-to-BVQX data import with importvtcfromanalyze (time courses), importvmpfromspms (maps), and importrfxglmfromspms (entire set of first-level stats)
• performing robust regression and computing t-statistics based on those results with the fitrobustbisquare (single sample regression), fitrobustbisquare_img (robust regression with one common model over multi-dimensional data), and fitrobustbisquare_multi (robust regression with variable model over multi-dimensional data) functions, followed by using the robustt function
• first-level regression (beta-estimation), either via the Compute Multi-Study GLM dialog or the MDM::ComputeGLM method on the command line (which creates a BrainVoyager QX-compatible RFX-GLM object)
• extract time-course data from VTC files using either the VTC::VOITimeCourse (single file) or MDM::VOITimeCourses (multiple files) methods
• VOI-based GLM computation (regression) using the mdm.ComputeVOIGLM method
• single-run ICA using the FastICA algorithm of the group of Aapo Hyvärinen either using the ne_fastica function or the VTC::ICA method
• RFX contrast computation and correlation with (behavioral) covariates using the Contrast Manager UI or the GLM::RFX_tMap and GLM::RFX_rMap methods (incl. group comparison as well as robust regression and rank correlation features)
• RFX (single-level) mediation analysis (whole-brain mediation mapping) using the RFX mediation analysis dialog for data stored in BrainVoyager QX's GLM format (or the mediationpset function on the command line for extracted data)
• computing conjunction maps
• estimating true-Null cluster-size thresholds using using alphasim
• classifying data with a Support-Vector-Machine (SVM) classifier using the svmtrain and svmpredict functions (which re-use the libSVM code by Chih-Jen Lin and his group)

• creating cluster tables with the main GUI or using the VMP::ClusterTable method
• extracting beta values from clusters with the main GUI or using the GLM::VOIBetas method

• creating high-resolution slice images (overlaid statistics, montage images) using the Image Montage UI, incl. multi-map overlay and integration into variable background
• creating high-resolution surface images (overlaid statistics) using the screenshot feature of the NeuroElf GUI satellite window
• visualizing activation maps as flatmaps using the files created by neuroelf_makefiles and the SMP sampling either via the main GUI or the VMP::CreateSMP method
• creating high-quality 3D renderings of brains (incl. statistics) using the Rendering UI
• creating GLM beta bar or scatter plots using the GLM beta plotter feature (supports subject groups and plotting of contrast values)
• creating trial/event-related averaging plots using the MDM VOI condition average UI
• flexible time-course plotting using the tcplot function
• plotting of (conditional) confidence ellipses using the cellipse function

• segmentation (only via SPM's preprocessing)
• inter-subject normalization (only via SPM's preprocessing)
• group-based ICA
• multi-level mediation analysis
• Granger Causality (network analysis and whole-brain mapping)

Please note, that at the moment I have too little time to develop and/or integrate these features into the toolbox!