Institute of Medical Psychology and Behavioral Neurobiology Germany

Institute of Medical Psychology and Behavioral Neurobiology (Germany) Wadsworth Center (USA) BCI 2000: A General-Purpose Software Platform for BCI

Acknowledgement I would like to thank all the author(s) for providing their material which I have used in this presentation and at the same time I would also apologize for unintentional copyrights violations. Asad 2

Motivation 3

Outlines • • Introduction & Methodology Technical Details System Working (Applications) Experimental Results Conclusions References Open Questions for Discussion 4

Introduction (BCI) • Uses brain signals to convey intent • Does not use peripheral nerves or muscles • Can provide communication and control to people who are totally paralyzed 5

Motivation to BCI 2000 • Most of the Brain signals processing systems do not readily support systematic research and development, like • Evaluating and comparing alternative signals and combinations of signals, alternative feature extraction methods, alternative translation algorithms, and control mechanism for different users. • In recognition to this situation, BCI 2000 was developed to provide and facilitate the systematic studies. 6

BCI 2000 (1/4) • Factors determine BCI System performance • Brain signals measured, the signal processing to extract features, the algorithms to translate features into device commands, the output devices to execute commands, the feedback to the user, and user characteristics. 7

BCI 2000 (2/4) • Essential Features • Common Model: Using TCP/IP, so modules can be developed in any programming language. • Interchangeability and Independence: Good flexibility, low coupling and high cohesion. • Scalability: No constraints on the no of signal channels or their sampling rate, no of system parameters, and etc. • Real-Time Capability: To achieve minimum latency jitter, BCI 2000 is designed in a way to minimize the response time of RT-OS, HD, and other sources of interruptions. 8

BCI 2000 (3/4) • System Variables • System Parameters: SP are those variables that do not change throughout a data file i. e. during a specified period of online operation. • Event Markers: EM record events that occur during operation and can change from one data sample to the next. EM’s allows full reconstruction of a session. • System Signals: SS’s are functions of user’s brain signals that are received and modified by the modules. 9

Defines system parameters; trial length in a specific application, Converts signals from the into or specific signal processing variable. Also defines brainonset signals that It receives control signals Stores and passes it to signal processing module. output device from Signal processing and control an and offset of operation. Where these definition are made by usesconversion is in two steps; 1) feature extraction It have data acquisition and storage components The them to derive an application automated algorithms or byapplications are presented display on computer the Userfiring investigator. Can visually File contains ASCII header followed by binary rate of a cortical neuron, amplitude of an like information from any module, which allow the investigator to signal samples and then an event marker screen and consists of selection of targets, letters, evoked potential, and etc 2) translation control the experiment and receive real time information about algorithms icons and etc. signal features into control signals that online events (display oftranslates the brain signals) unprocessed are sent to user application module Sys Para Sig Proc BCI 2000 (3/4) • Modules Display and control Info 10

Technical Details

Initial Implementations of BCI 2000 • Platform • Language and Development Environment : BCI 2000 system model accommodates any programming language and any development environment for implementation. C++ with development environment of Borland C++ Builder is used for BCI 2000 implementation. • Operating System: MS Windows. TM 2000/XP are used for the BCI 2000 development. • Device drivers facility • Real time processing requirements 12

Initial Implementations of BCI 2000 • Source Module • Five source module implementations created to date. Three of them control A/D converter boards, one for EEG recoding system, and the fifth one is for signal generator for use in system development and testing. • In the case of A/D converter boards, brain signals must first be conditioned i. e. band-pass filtered and amplified, so that they can be detected by the A/D hardware. • The data storage component incorporated in these source implementation is highly optimized so that many data channels and /or high digitization rates can be used with minimal effects on the latency of real-time operation. 13

Initial Implementations of BCI 2000 • Signal Processing Module • The first stage is feature extraction from the digitized brain signals. Three signal operators are applied here; 1) linear transformation of the input matrix i. e. sample block, 2) Linear spatial filters (Laplacian derivation or common average, IC, and Common Spatial Patterns), 3) Temporal Filters (slow wave filter, autoregressive spectral estimation, finite impulse response filter and a peak detection routine) are applied here. • In second stage, the translation algorithms translates the signal features to control signals to be used by user application. 14

Initial Implementations of BCI 2000 • User Application Module • Seven different application are implemented • Four cursor movement, one evaluating prospective user, present user selectable auditory and visual stimuli, and a spelling application based on evoked potentials. 15

Initial Implementations of BCI 2000 • Operator Module • Provides a graphical interface that displays current system parameters and real-time analysis results (e. g. , frequency spectra) communicated by other modules. It allows the investigator to start, stop, suspend, resume, or reconfigure system operation. In a typical BCI 2000 configuration, user feedback is displayed on one monitor and the operator module’s graphical interface is displayed on a second monitor. 16

User Interface of BCI 2000

A Typical BCI 2000 System 18

BCI 2000 Features • Supported Hardware • Guarantee support for all devices from g. tec; • Data Translations, National Instruments, Measurement Computing, Modular-EEG, • Neuroscan, Brain Products, and Micromed • Supported Brain Signals • Evoked Potentials, EEG mu/beta Rhythms (ERD/ERS), ECo. G Oscillations, Slow Cortical Potentials, Single-Neuron Action Potentials • Supported Feedback Protocols • Auditory & Visual Stimulus Presentation, 1 D/2 D/3 D Cursor Movement, n-Choice Speller, P 300 Matrix Speller 19

BCI 2000 Applications

BCI 2000 Application Demonstration

BCI 2000 Applications • • • Two-Dimensional Control of a Robotic Arm Spelling using EEG and Dasher Four-Target Selection Using Tactile Feedback BCI Control of a Domotic Assistive System Rapid Identification of Motor Cortex Overall View

BCI 2000 Evaluation

BCI 2000 Performance 1. 2. 1. 4 GHz, 256 Mb RAM, IDE I/O, DT 3003, Windows 2000 2. 53 GHz, 1 Gb RAM, SCSI I/O, NI 6024 E, Windows XP 24

BCI 2000 Performance 1. 2. Topographical distribution of P 3000 potential at 340 ms after stimuli Solid line for desired char and dashed line for un-desired char 25

BCI 2000 Benefits

Benefits of BCI 2000 • To The Investigator Availability of a complete economical system already allows for a range of studies that • To The Software Engineer Concentration on the aspects that are unique to the new BCI method that is to be developed • To End User Large community which helped achieving a stable release made custom solutions available. 27

Future Development of BCI 2000 • Platform • Modules • Source Module: Currently support data acquisition hardware from four different vendors • Signal Processing: Currently, it extract features from scalp -recorded sensori-motor cortex rhythms, cortical surface rhythms, slow cortical potentials, cortical single neurons, and P 300 evoked potentials. Other brain signals, such as the error potential, cortical field potentials, or intra-cortical neuronal activity are not implemented. • User Application Module: Auditory plus visual feedback 28

Conclusions Bad decisions are tomorrow’s constraints. BCI (with BCI 2000) is an Good decisions are tomorrow’s o. PEN d. OOR to the Future. …. opportunities. BCI 2000 with executables and source code, and documentation is available for 29 free for research and educational purposes at http: //www. bci 2000. org

References • • • Schalk, G. , Mc. Farland, D. , Hinterberger, T. , Birbaumer N. , Wolpaw J. "BCI 2000: A General-Purpose Brain-computer Interface (BCI) System. " IEEE Trans. Biomed. Eng. , vol. 51, pp. 1034 -1043 (2004) http: //bci 2000. org/BCI 2000/Home. html. http: //www. bciresearch. org/BCI 2000/download. html http: //www. bci 2000. org/wiki/index. php/User_Tutorial: BCI 2000_Tour http: //bci 2000. org/BCI 2000/Videos. html Towards Brain-Computer Interface, Edited by: Guido Dornhege, Jose del R. Millan, Thilo Hinterberger, Dennis J. Mc. Farland, and Klaus-Robert Muller, The MIT Press 30

Questions? ? 31

Open Discussion Questions • Experiments • Music

Future

BCI 2000 Architecture 34