Fourteenth EEGLAB Workshop
Mallorca, Spain, Sept. 22-25, 2011 preceding ICON
Overview
The 14th EEGLAB Workshop took place from Thursday, September 22 through Sunday, September 25, 2011 on the Spanish island of Mallorca preceding ICON XI. Participants in the first two parts of the Workshop were expected to bring laptops with MATLAB installed so as to be able to participate in the practical sessions. The workshops consisted of three parts:
1. On Thursday, Sept. 22 there was a full-day Novice EEGLAB Workshop for those interested in learning the basics of using EEGLAB and independent component analysis (ICA) to analyze their EEG (or related) data.
2. Friday, Sept. 23 through noon on Sunday, Sept. 25, the first Advanced EEGLAB Workshop introduced and demonstrated the use of EEGLAB-linked tools for performing advanced analyses of EEG and related data, with detailed method expositions and practical exercises.
3. Sunday afternoon, Sept. 25, there was a (free) Open Discussion Session of evolving directions in EEG/ECoG research and free / open source data analysis, collection, and archival software. All interested were welcome to attend or participate in this discussion.
Workshop Agendas
Key: Lecture, Practicum, Break
I. Novice EEGLAB Workshop – Thursday, Sept. 22
This workshop is designed for researchers who would like to learn how to process their EEG or related datasets using the ICA, time/frequency, and other tools provided in the EEGLAB software environment for MATLAB (http://sccn.ucsd.edu/eeglab). The workshop instructor will be Julie Onton, Ph.D., long-time EEGLAB user and SCCN laboratory member. Scott Makeig, Director of the Swartz Center for Computational Neuroscience, UCSD, and originator of EEGLAB methods will give an introductory talk on evolving methods for analyzing EEG dynamics.
Novice workshop topics will include: • Data import and preprocessing options
• Basic independent component analysis (ICA) theory and application
• Methods for imaging IC activations (ERPs, time/frequency, coherence)
• Equivalent dipole source localization of independent components
• Introduction to MATLAB scripting using EEGLAB structure
Note: Because of time limitations, the Novice workshop will NOT include: • New, more advanced toolboxes (NFT, SIFT, BCILAB, MPT)
• Mathematical derivations of the algorithms discussed
These topics will be covered in the Advanced EEGLAB workshop (described below).
08:30 – 09:15am – Mining event-related brain dynamics (Scott Makeig) (PDF)
09:15 – 10:15am – Introduction and getting started with EEGLAB (Julie Onton) (PDF)
Data import - Preprocessing tools and pipeline - Running ICA decomposition
10:15 – 10:30 BREAK
10:30 – 11:15 – Evaluating ICA components (ICs) (PDF)
Apply ICA weights - IC scalp map interpretation - Basic IC evaluation - Identify artifact ICs
11:15 – 12:15 – IC analysis tools (PDF)
Removing ICs and back-projection - IC ERP envelope - IC ERP images - Time-frequency analysis - IC Event-related spectral perturbations (ERSPs) - IC Cross coherence
12:15 – 13:30 LUNCH
13:30 – 14:30 – EEGLAB ‘EEG’ structure and basic MATLAB scripting (PDF)
‘EEG’ structure overview - ‘EEG’ structure overview - Search EEG.event structure - MATLAB functions - Converting from ‘pop’ functions to output functions
14:30 – 15:15 – Equivalent dipole modeling (PDF)
Co-registration of electrodes with head model - Dipole fitting using Fieldtrip’s dipfit function - Co-registration for 3D headplots
15:15 – 15:30 BREAK
15:30 – 16:30 – Introduction to EEGLAB STUDY structure (PDF)
Build a STUDY - Precompute, precluster, and cluster ICs across subjects - Plot and edit STUDY clusters
15:30 – 17:30 – Advanced STUDY scripting (PDF)
Build a STUDY from the command line - STUDY structure overview - Cluster ERP image - Accessing raw STUDY data measures
II. Advanced EEGLAB Workshop – Friday-Sunday, Sept. 23-25
This 2.5-day workshop will focus on emerging computational methods for EEG/ECoG analysis that have recently been made available within the EEGLAB environment as plugin toolboxes. The lectures and practica will be more technically advanced than previous EEGLAB workshops. Participants will be expected to have at least passing familiarity with concepts such as linear regression, matrix inversion and other basic linear algebraic operations, Fourier transforms, and basic probability theory.
In addition, participants should be comfortable with using MATLAB including performing the following operations using EEGLAB: • Performing ICA decompositions and evaluating ICA component
• Obtaining equivalent dipole models of independent components using DIPFIT
• Performing time-frequency transforms and coherence analysis in EEGLAB
• Building an EEGLAB data STUDY
Advanced workshop topics will include: • Applying adaptive-mixture ICA (Amica) to non-stationary EEG source dynamics
• Using Measure Projection analysis of multi-subject ICA-resolved EEG dynamics
• Applying statistical machine learning to EEG data analysis and Brain-Computer Interface design
• Analysis of oscillatory source network dynamics including Granger causality
• Forward and inverse modeling for EEG/ECoG source localization
Participants are expected to bring a laptop with MATLAB and EEGLAB installed to work on (detailed instructions will be sent out before the workshop). Pairs of participants may also choose to share a laptop.
08:30 – 09:00 – Welcome, introductions and brief overview (Scott Makeig)
Session A – Adaptive mixture independent component analysis (AMICA) decomposition (Jason Palmer)
This session will motivate, derive simply, and demonstrate the Adaptive Mixture ICA (Amica) algorithm of (Palmer et al., 2007) that finds more physiologically interpretable component processes in high-density EEG (or related) data and, further, detects and models changes in the spatial EEG source structure. Use of a set of tools and measures for interpreting the results of Amica decomposition will be demonstrated.
09:00 – 09:45 ICA methods overview, with motivation for and derivation of Amica (PDF)
09:45 – 10:45 Amica toolbox practicum
10:45 – 11:00 BREAK
Session B – Improving EEG source estimation using the Neuroelectromagnetic Forward Head Modeling Toolbox (NFT) (Zeynep Akalin Acar)
Both PDFs, supplementary functions and data for NFT
Supplementary functions and data ONLY, for practicum
11:00 – 12:00 – Forward head modeling overview (PDF)
12:00 – 13:30 LUNCH
13:30 – 14:30 – NFT head modeling toolbox practicum (PDF)
Session C – Comparing EEG dynamics across subjects using the Measure Projection Analysis (MPA) toolbox (Nima Bigdely Shamlo)
14:30 – 15:30 – Measure projection analysis theory (PDF)
15:30 – 15:45 BREAK
15:45 – 17:30 – Measure projection analysis practicum (PDF)
Session D – Analyzing oscillatory EEG/ECoG source dynamics and interactions using the Source Information Flow Toolbox (SIFT) (Tim Mullen)
8:30 – 10:00 am – Analyzing Oscillatory Dynamics and Effective Connectivity (PDF)
10:00 – 10:30 BREAK
10:30 –12:00 – Using the Source Information Flow Toolbox: practicum (PDF)
12:00 – 13:30 LUNCH
Session E – Statistical Learning Theory and Brain-Machine Interface Design (Christian Kothe)
Slides from Christian’s lecture.
13:30 – 15:00 – Statistical machine learning and Brain-Computer Interface design
115:00 – 15:30 BREAK
15:30 – 17:30 – Using BCILAB to design and run BCI, cognitive monitoring, and neurofeedback experiments / applications (PDF)
Session F - Workshop Review, Results, and Discussion
09:00 – 10:00 – EEG Research: Current and Future Directions (Scott Makeig)
10:00 – 10:30 – Group practica results preparation
10:30 – 10:45 BREAK
10:45 – 12:00 – Group practica results presentations and general discussion
III. EEGLAB Workshop Open Discussion Session – Sunday, Sept. 25, 13:30 - 16:30
13:30-16:30 – Evolving Methods and Tools for EEG/ECoG Research (preliminary agenda)
1. Introduction – Framing remarks and EEGLAB overview (Scott Makeig)
2. Advanced EEGLAB-compatible tools and directions (Panel)