Main updates : March 2009 - June 2010

Introduction

Brainstorm evolution overview in the last 3 years:

February 2007 – February 2009: Design and development of a completely new Java-based interface, making the software much more intuitive and stable, and allowing much more interactivity with the recordings. But development and diffusion was still limited to a few researchers and students in Paris.

March 2009 – February 2010: Debugging and testing, elaboration of a website for diffusion, support and collaborative documentation, and many additional developments to make interactions with other software packages easier.

February 2010 - June 2010: Development of new important features, such as the computation and visualization of time-frequency decompositions using Morlet wavelets.

The new version of Brainstorm was made available for download in June 2009, at the occasion of the 15th Human Brain Mapping Conference in San Francisco, where it was first publicly presented. About 850 new users registered on the website and started using Brainstorm between June 2009 and June 2010, in addition with the 4,000 accounts already existing.

The software was presented in various conferences worldwide, including Biomag 2010, HBM 2010, and several events in Europe and North America. Almost all the users discovering or using daily the program gave very enthusiastic feedback. Many are impressed by how easy, fast, and reliable the program is. For many applications, Brainstorm can now represent an interesting alternative to commercial software packages.

Here are the main improvements of the past 18 months.

Documentation

• Website: Wiki-based website for collaborative documentation and distribution: [[|http://neuroimage.usc.edu/brainstorm]]

• Tutorials: Redaction of detailed and well illustrated step-by-step tutorials to help new users discovering all the features of the interface.

Distribution

• Visibility: Improved visibility on the web by registering on specialized websites (NITRC, INCF, Matlab Central)

• Portability: Fixed lots of compatibility issues (file systems, access rights, interfaces, different Matlab behaviors)

  • All the features are now available for most of the common operating systems: Linux, Unix, Windows, MacOS

• Updates: Automatic update system for Brainstorm, that downloads a new version every month (and easy manual update)

• Compilation: Integrated the deployment functions in the general distribution. Now any user with a Matlab Compiler licence can generate easily a stand-alone version of Brainstorm

• Validation: Development of a set of functions that test automatically that all the interface elements are still working before releasing a new version online.

• Logs: Now logging all the logins / downloads / automatic updates in a MySQL table on Neuroimage server

Source modeling

• SSP: now read, store, and use SSP projectors in forward model

• Noise covariance matrix:

  • Explicit computation and storage in the database
  • Now possible to compute it from: continuous raw recordings, single trials, and averaged recordings.
• Re-coding, optimization, stabilization and improvement of many of the models: overlapping spheres, minimum norm estimator, beamformer
• Recent addition of a minimum norm function that gives the same results as MNE (Rey)
  • Also computes dSPM and sLORETA
  • Various dipoles orientation constraints : unconstrained, loose, truncated, normal to the cortex
• Use of the coils definition file for CTF recordings (coil_def.dat)

New features

• Time-frequency: Computation and visualization of time-frequency decompositions of MEG/EEG and source signals using Morlet wavelets.

• Graphic batching system (Processes tab)

  • Apply a process to complex structures of data with just a few mouse clicks and drag’n’drops
  • Easy to add new processes to the list

• Scripting system: automation of many processes without using the GUI

• Projection of sources on a different surface with a basic registration of the surfaces (smooth + adaptative ICP)

  • Between two meshes with different resolutions from the same subject
  • From an individual cortex mesh to a template (Colin27, or user defined)

• Digitized head points:

  • Interface to import, save, view digitized head points
  • Automatic MEG/MRI registration based on them
  • Deformation of an anatomy (MRI + surfaces) based on a set of points (still requires some adjustments...)

• Clusters: define and analyze groups of sensors

• Spherical models:

  • Redesigned the interface for the definition of the sphere
  • Estimates based on existing information (scalp, cortex, head points, EEG, MEG...) are now much more accurate

• Dipoles: Display of dipoles estimations from the FDA-approved software published by Elekta-Neuromag. This is very helpful in a clinical environment to compare the results obtained with clinical and research software.

Display

• Magnetic extrapolation: Display of recordings in 2D/3D are now much accurate and smoother (Rey's bst_extrapm function)

• Display of recordings in columns

  • Selection of electrodes similar to MNE (support for .sel files)

• MRI Viewer:

  • Completely recoded in order to display sources, MIP, support mouse wheel, colorbars
  • Made more physician-friendly (black background, useless controls can be hidden, neurological/radiological orientation...)

I/O

• Recoded all the I/O subsystem for MEG/EEG import
  • Now only two frontend functions for all the files formats: in_fopen and in_fread
• New unique interface for importing recordings, that can do additionnally:
  • Epoching
  • Resampling
  • Splitting long files in small time blocks
  • Baseline removal
• Full support for new acquisition systems:
  • Elekta-Neuromag MEG
  • Neuroscan EEG
  • 4D-Neuroimaging
• Other file formats added:

  • ?FreeSurfer surfaces

  • Electrodes positions from many formats
• Integration with MNE:
  • Use of the MNE Matlab toolbox for reading FIF files
  • Import from FIF files: recordings, surfaces, head points, noise covariance matrix
  • Import / export of sensors selections in MNE format (.sel files)
• Improved exportability of the data manipulated in Brainstorm:
  • Export of recordings, meshes and sensors definition in many formats, including ASCII, and some of the common file formats
  • Improved the integration with Matlab environment

Various details

• Smart loading / unloading for : recordings, sources, surfaces, MRI volumes
• Average reference for EEG: now always computed on the fly
• Support for multiple screens (one for the database, one for the data)
• Interface to edit some of the preferences of the application