Development of a real-time cortical rhythmic activity imaging technology and its applications

Abstract

The principal aim of this dissertation is to develop a real-time cortical rhythmic activity imaging technology and to apply this technology to a variety of potentially practical applications, such as real-time brain activity monitoring, diagnosis of brain diseases, advanced neurofeedback, brain-computer interface (BCI), and classification of human thoughts. To this end, the author first implemented an EEG-based, real-time, cortical rhythmic activity monitoring system to investigate whether or not a real-time cortical rhythmic activity imaging is feasible. In the monitoring system, a frequency domain inverse operator is preliminarily constructed, considering the subject’s anatomical information and sensor configurations, and then the spectral current power at each cortical vertex is calculated for the Fourier transforms of successive sections of continuous data, when a particular frequency band is given. A preliminary offline simulation study using four sets of artifact-free, eye-closed, resting EEG data acquired from two dementia patients and two normal subjects demonstrates that spatiotemporal changes of cortical rhythmic activity can be monitored at the cortical level with a maximal delay time of about 200 ms, when 18 channel EEG data are analyzed under a Pentium4 3.4 GHz environment. The first pilot system is applied to two human experiments– (1) cortical alpha rhythm changes induced by opening and closing eyes and (2) cortical mu rhythm changes originated from the arm movements– and demonstrated the feasibility of the developed system. The developed real-time cortical rhythmic activity monitoring system was utilized as a motor imagery training system for EEG-based brain-computer interface (EEG). Ten healthy participants took part in this study, half of whom were trained by the suggested training system and the others did not use any training. All participants succeeded in performing motor imagery after a series of trials to activate their motor cortex without any physical movements of their limbs. To confirm the effect of the suggested system, EEG signals were recorded for the trained group around sensorimotor cortex while they were imaging either left or right hand movements according to the experimental design, before and after the motor imagery training. For the control group, EEG signals were also measured twice without any training sessions. The participants’ intentions were then classified using a time-frequency analysis technique, and the results of the trained group showed significant differences in the sensorimotor rhythms between the signals recorded before and after training. Classification accuracy was also enhanced considerably in all participants after motor imagery training, compared to the accuracy before training. On the other hand, the analysis results for the control EEG data set did not show consistent increment in both the number of meaningful time-frequency combinations and the classification accuracy, demonstrating that the suggested system can be used as a tool for training motor imagery tasks in BCI applications. With just slight modifications of the real-time cortical rhythmic activity monitoring system, the author developed an EEG-based, real-time, cortical functional connectivity imaging system capable of monitoring and tracing dynamic changes in cortical functional connectivity between different regions of interest (ROIs) on the brain cortical surface. To verify the implemented system, the author performed three test experiments in which the author monitored temporal changes in cortical functional connectivity patterns in various frequency bands during structural face processing, finger movements, and working memory task. The author also traced the changes in the number of

connections between all possible pairs of ROIs whose correlations exceeded a predetermined threshold. The quantitative analysis results were consistent with those of previous off-line studies, thereby demonstrating the possibility of imaging cortical functional connectivity in real-time. The cortical source imaging was used to decode various mental states more accurately than sensor-level analyses. Eight participants took part in this study; their EEG data were recorded while they performed four different cognitive imagery tasks. The spectral power at each preliminarily determined cortical ROIs was estimated, and then a 2D spatiospectral pattern map was constructed for each task, of which each element was filled with 1, 0, and -1 reflecting the degree of event-related synchronization (ERS) and event-related desynchronization (ERD). Consistent ERS/ERD patterns were observed more frequently between trials in the same class than those in different classes, indicating that these spatiospectral pattern maps could be used to classify different mental states. Classification of a specific mental state was performed through the similarity evaluation between a current 2D pattern map and the template pattern maps, by taking the inner-product of two pattern matrices. The classification accuracy was evaluated using the leave-one-out cross-validation (LOOCV) and that for sensor-level analysis using the raw EEG signals was also calculated for comparison. An average accuracy of 76.31% ( 12.84%) was attained for the cortical-level analysis; whereas an average accuracy of 68.13% ( 9.67%) was attained for the sensor-level analysis, demonstrating cortical-level analysis can interpret various human thoughts more correctly than sensor-level analysis. In summary, the author developed a real-time cortical rhythmic activity imaging technology and demonstrated the usefulness of the developed technology by successfully realizing a variety of practical applications.