Showing papers by "Gert Pfurtscheller published in 1997"

Adaptive autoregressive modeling used for single-trial EEG classification.

Abstract: An adaptive autoregressive (AAR) model is used for analyzing event-related EEG changes. Such an AAR model is applied to single EEG trials of three subjects, recorded over both sensorimotor areas during imagination of left and right hand movements. It is found that discrimination between both types of motor-imagery is possible using linear discriminant analysis, but the time point for optimal classification is different in each subject. For the estimation of the AAR parameters, the Least-mean-squares and the Recursive-least-squares algorithms are compared. In both methods, the update coefficient plays a key role: it determines the adaptation ratio as well as the estimation accuracy. A new method, based on minimizing the prediction error, is introduced for determining the update coefficient.

Using adaptive autoregressive parameters for a brain-computer-interface experiment

Abstract: In an online EEG discrimination task, continuous feedback was presented. The EEG was recorded during imagination of left and right hand movement and analyzed with adaptive autoregressive parameters. The parameter discrimination was fed back in form of a rectangular bar on a computer screen over a period of four seconds. An online classification result of more than 90% was obtained after a few sessions.

Timing of eeg-based cursor control

Abstract: Summary:Recent studies show that humans can learn to control the amplitude of electroencephalography (EEG) activity in specific frequency bands over sensorimotor cortex and use it to move a cursor to a target on a computer screen. EEG-based communication could be a valuable new communication and con

EEG-based communication: evaluation of alternative signal prediction methods.

Abstract: Individuals can learn to control the amplitude of EEG activity in specific frequency bands over sensorimotor cortex and use it to move a cursor to a target on a computer screen. For one- dimensional (i.e., vertical) cursor movement, a linear equation translates the EEG activity into cursor movement. To translate an individual's EEG control into cursor control as effectively as possible, the intercept in this equation, which determines whether upward or downward movement occurs, should be set so that top and bottom targets are equally accessible. The present study compares alternative methods for using an individual's previous performance to select the intercept for subsequent trials. In offline analyses, five different intercept selection methods were applied to EEG data collected while trained subjects were moving the cursor to targets at the top or bottom edge of the screen. In the first two methods - moving average, and weighted sum - a single intercept was selected for the entire 1-2 sec period of each trial. In the other three methods - blocked moving average, blocked weighted sum, and blocked recursive sum (a variation of the weighted sum) - an intercept was selected for each 200-ms segment of the trial. The results from these methods were compared in regard to their balance between upward and downward movements and their consistency of performance across trials. For all subjects combined, the five methods performed similarly. However, performance across subjects was more consistent for the moving average, blocked moving average, and blocked recursive sum methods than for the weighted sum and blocked weighted sum methods. Due to its consistent performance and its computational simplicity, the moving average method, using the five most recent pairs of top and bottom trials, appears to be the method of choice.

Effects of handedness on movement-related changes of central beta rhythms.

Abstract: Summary:The effects of handedness on the movement-related changes in β rhythms (14-30 Hz) in the left and right perirolandic area were analyzed in 12 right-handed and 11 left-handed subjects. The motor task consisted of unilateral brisk or slow self-paced extension of the right or left index finger.

Post-movement Beta synchronization studied with linear estimation

Abstract: Event-related desynchronization (ERS) describes a short-lasting and localized amplitude enhancement of specific frequency components. The spatial distribution of a post-movement beta ERS can be visualized by computing the local average reference (LAR). The Linear estimation (LE) method can also be applied to study the spatiotemporal ERS patterns. As source space an hemisphere was used with equally distributed radially oriented current dipoles. The lead field matrix is normalized to make sure that all dipoles have the same average impact on the sensors. A distributed source solution is found for each timestep and for each trial. Event-related Desynchronization calculations are carried out for every dipole (squaring of amplitude, averaging over all trials and time averaging over 16 time points). Both methods were conducted for the study of voluntary hand movement. The results are similar but in contrast to the LAR maps, the LE maps show a better spatial resolution. This is not surprising since the LAR method is limited to the electrode sites whereas with LE the EEG activity is projected onto the source space. Furthermore, the LE method counteracts the deblurring caused by the poorly conducting skull. Linear Estimation depends on several assumptions about the source space, volume conductor and the regularization parameter. Further investigation is needed to evaluate the application of LE for the study of Event-Related EEG phenomena.

Subject specific EEG patterns during motor imaginary [sic.: for imaginary read imagery]

Abstract: Left- and right-hand movement imagery is accompanied by an EEG event-related desynchronization (ERD) over the contralateral hand area The time courses of the ERD from two brain-computer interface experiments were investigated by the calculation of instantaneous band power changes and by adaptive autoregressive model parameters combined with linear discriminant analysis Subject-specific differences of the EEG reactivity patterns were observed

Post-Movement EEG Synchronization Studied with Different High Resolution Methods

Abstract: In this paper we present a study of spline surface Laplacian (LP), linear estimation (LE) and analytical deblurring (AD) utilized to improve the spatial resolution of single trial EEG data. AD is a method to reconstruct the potential distribution on the cortical surface. The dependency of AD on the electrode grid size as well as the sensitivity to uncorrelated noise and errors in the volume conductor model are investigated in detail and compared with LP. Finally, all methods (LP, LE and AD) are applied to single trial EEG data recorded in three subjects during voluntary and self-paced extension and flexion movements of the right index finger. In each subject postmovement beta oscillations were found in specific frequency bands. Cortical dipolar source strengths were reconstructed by LE and cortex potentials were estimated with AD. Both results are compared with LP calculated from the scalp EEG. All methods, although having different theoretical basis, yield similar results and reveal a maximal event-related synchronization over the left sensorimotor area approximately 500-875 ms after termination of the movement.

Cortical Imaging Based on an Analytic High Resolution EEG

Abstract: It is well known that the EEG is a blurred and spatially low-pass filtered representation of the cortical activity. Since additional recording electrodes will not necessarily lead to an improved spatial resolution, further steps have to be taken. We derived the analytic downward continuation of the scalp potential field to an arbitrary inner surface for a spherical volume conductor model with piecewise constant conductivities. The basic idea of the Analytic High Resolution BEG (AHREEG) is the fact that a function defined on a sphere can be expressed as a weighted sum of spherical harmonics. Considering the spatial transfer function between the cortex and the scalp surface, the potential distribution on the cortex is theoretically computed by the application of the inverse transfer function to the scalp potential field. Compared to source localization procedures, the AHREEG is not based on any source distribution or on the nature of the sources. Furthermore the proposed method is unique, though ill-posed. Due to this fact and the fact that real world data are always contaminated by noise, a regularization based on general cross validation is performed to stabilize the inverse solution. Simulation results as well as the application of the AHREEG to EEG recorded during median nerve stimulation will be presented and critically discussed.