I would recommend using Brainstorm. It uses some sLORETA toolboxes and is really user friendly and intuitive to use. It works as a Matlab GUI so you can go back and forth between Matlab and Brainstorm to manipulate the data as you wish. Brainstorm is originally developed with MEG in mind but all their algorithms are also work for EEG as well. The MEG focus is mostly evident in the tutorials rather than on the usability.

Have a look at MNE. It is free, open source and has both standalone and python versions: http://martinos.org/mne/stable/index.html. IT is under active development (see https://github.com/mne-tools/mne-python). Here is a recent companion paper that discusses some of its capabilities: http://www.sciencedirect.com/science/article/pii/S1053811913010501.  The following is a link to the forum/mailing-list to discusses any issues related to the methods or their implementation in MNE: https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis.

I would suggest EEGlab for its general high quality and easy use. I have no experiences myself with source reconstruction, but know that this toolbox is capable to do so. Interesting in this context might be the capability to compute independent components and locate them.

Thanks folks, for the feedback. I will have a look at those toolboxes. I have checked NFT toolbox , a plugin for EEGLAB for solving forward problem, however it still needs much improvement. The EEGLAB itself seems to provide dipole fitting on ICA , however I would also like to have distributed source localization algorithms. Thanks.

I would suggest sLORETA. The package includes routines expressively designed for rhythm localization. It uses a standard MR for head model reconstruction, while channel locations are defined by the user.  By the way, it's really user-friendly.

I'm very familiar with all the methods mentioned by the former fellows, but if you really would dig into source reconstruction deeply, and apply several different methods, the most comprehensive suite to me is BESA Research. Although it is commercial and somewhat expensive, the money you spent is worth it. Here is s small list of its source reconstruction methods:

Highly interactive graphical user interface for fast hypothesis testing

Spatio-temporal multiple dipole and regional source modeling

Image-weighted source fitting

Tomographic source sensitivity display

Automated multiple source probe scan (MSPS) for model validation

Automated source fitting: RAP-MUSIC, Genetic algorithm

PCA analysis with generation of spatial source components

ICA components obtained in the main window can be used as spatial components in the source analysis window

Realistic standardized FEM and multi-shell ellipsoidal head models

Individual FEM head models can be imported from BESA MRI 2.0

Source imaging

3D imaging of oscillatory activity using the multiple source beamformer (MSBF) or Dynamical Imaging of Coherent Sources (DICS)

Distributed volume images: LORETA, sLORETA, swLORETA, LAURA, SSLOFO

Iterative, focusing 3D imaging by CLARA (Classical LORETA Analysis Recursively Applied)

User-defined 3D imaging methods

Minimum norm images based on the individual brain surface

3D movie of maps and dipoles using source waveforms

Source activity projected onto standardized or individual MRI

Source import and export:

Direct transfer of source models between subjects and conditions

Transfer of source models, waveforms, and 3D images to MATLAB

Export of source models and source waveforms in ASCII

Creation of source montages

Coordinate systems: Talairach / Head / Device / Unit Sphere

Batch commands for many source analysis functions

Addtionally, the reconstruction works with spontaneous and event-related EEG and MEG data and you can merge EEG and MEG reconstruction.

Thank you Marco for the suggestion. I will try it as well. Currently, I'm exploring the brainstorm toolbox, so far so good. 

Found these slides by Dr. Tadel which compares different tools for source reconstruction. 

these slides are really old. I use MNE for EEG source reconstruction routinely exploiting the automatic anatomical pipeline for segmentation and the individual BEM model for more accurate forward modeling. This is crucial for beamforming. MNE implements time frequency beamformers which seems to be what you're looking for. See examples such as:

http://martinos.org/mne/dev/auto_examples/inverse/plot_dics_source_power.html#example-inverse-plot-dics-source-power-py

http://martinos.org/mne/dev/auto_examples/inverse/plot_tf_lcmv.html#example-inverse-plot-tf-lcmv-py

http://martinos.org/mne/dev/auto_examples/inverse/plot_tf_dics.html#example-inverse-plot-tf-dics-py

hope this helps

Alex

Thank you Alex! Very useful. I think I've found what I was looking for in your toolbox. 

Cheers,

Berdakh.

Having spent half a year on this journey, I strongly recommend to anyone using both BrainStorm and FieldTrip. 

if you don't restrict yourself to matlab you can have a look at MNE.

http://martinos.org/mne/stable/index.html

it supports many inverse methods (MNE, dSPM, sLORETA, LCMV, DICS, MxNE, single dipole fit and soon RAP-MUSIC)

The language used for scripting is Python.

See http://www.sciencedirect.com/science/article/pii/S1053811913010501

Thank you. I took a look at MNE and it seems to be a powerful package. 

Hi Alexandre, Could you point us to an example of using MNE for source localization of EEG? All the examples I see online, show how to do it for MEG signals not EEG. Thanks!

Same for Brainstorm. We also have an EEG-specific tutorial documentation and dataset available online. Note also some courses coming up later this spring in Montreal.

http://neuroimage.usc.edu/brainstorm/Tutorials/Epilepsy

Thank you all. I have been able to successfully perform the analysis on each individual subject and I have two specific questions

 - How can I perform group analysis to detect significant differences between healthy and non-healthy group of subjects? Add your answer

- How can I get Loreta results for a given analysis window of my ERPs. Currently I can either specify an exact time or let it generate the results for the maximum activity. 

Thank you for your time. 

We have a full tutorial on group analysis:

http://neuroimage.usc.edu/brainstorm/Tutorials/VisualGroup

It's MEG and a vision study but the principles are foundations for a large range of studies, incl. in EEG.

http://neuroimage.usc.edu/brainstorm/Tutorials/VisualGroup

Hi, you can use EEGLAB and BCILAB efficiently for EEG signal analysis.

How the band of EEG is divided in sLoreta software, such as delta,theta,alpha1,alpha2,beta1,beta2,beta3

It has been a while since this question was asked. I have been using MNE python for EEG data analysis for some time (https://martinos.org/mne/stable/index.html).

Given the power of Python ecosystem, the integration of MNE with ScikitLearn + NILearn + Pytorch + thousands of packages are straightforward.

If someone has the same question then use >>> MNE.

I am wondering if anyone has ever made a comparison between the inverse solution of SPM and MNE. In particular what puzzles me a lot is that averaging across SPM's single trial solutions is not equivalent to compute the inverse solutions of the mean EEG signals (with MNE it is a linear operation, that is the two approaches yield the same solution).

I think we can use LORETA key, EEG to sloreta, then can be viewed in LORETA VIEWER, then you can save the image as .nii, thus, we can open it in SPM and do localization.

if you are interested in using MxNE in C++, you can find it in the iSDR package

https://github.com/BBELAOUCHA/iSDR_p