Thea Radüntz

EEG artifact elimination by extraction of ICA-component features using image processing algorithms.

Artifact rejection is a central issue when dealing with electroencephalogram recordings. Although independent component analysis (ICA) separates data in linearly independent components (IC), the classification of these components as artifact or EEG signal still requires visual inspection by experts. In this paper, we achieve automated artifact elimination using linear discriminant analysis (LDA) for classification of feature vectors extracted from ICA components via image processing algorithms. We compare the performance of this automated classifier to visual classification by experts and identify range filtering as a feature extraction method with great potential for automated IC artifact recognition (accuracy rate 88%). We obtain almost the same level of recognition performance for geometric features and local binary pattern (LBP) features. Compared to the existing automated solutions the proposed method has two main advantages: First, it does not depend on direct recording of artifact signals, which then, e.g. have to be subtracted from the contaminated EEG. Second, it is not limited to a specific number or type of artifact. In summary, the present method is an automatic, reliable, real-time capable and practical tool that reduces the time intensive manual selection of ICs for artifact removal. The results are very promising despite the relatively small channel resolution of 25 electrodes.

Automated EEG artifact elimination by applying machine learning algorithms to ICA-based features

OBJECTIVE Biological and non-biological artifacts cause severe problems when dealing with electroencephalogram (EEG) recordings. Independent component analysis (ICA) is a widely used method for eliminating various artifacts from recordings. However, evaluating and classifying the calculated independent components (IC) as artifact or EEG is not fully automated at present. APPROACH In this study, we propose a new approach for automated artifact elimination, which applies machine learning algorithms to ICA-based features. MAIN RESULTS We compared the performance of our classifiers with the visual classification results given by experts. The best result with an accuracy rate of 95% was achieved using features obtained by range filtering of the topoplots and IC power spectra combined with an artificial neural network. SIGNIFICANCE Compared with the existing automated solutions, our proposed method is not limited to specific types of artifacts, electrode configurations, or number of EEG channels. The main advantages of the proposed method is that it provides an automatic, reliable, real-time capable, and practical tool, which avoids the need for the time-consuming manual selection of ICs during artifact removal.

Signal Quality Evaluation of Emerging EEG Devices

Electroencephalogram (EEG) registration as a direct measure of brain activity has unique potentials. It is one of the most reliable and predicative indicators when studying human cognition, evaluating a subjects health condition, or monitoring their mental state. Unfortunately, standard signal acquisition procedures limit the usability of EEG devices and narrow their application outside the lab. Emerging sensor technology allows gel-free EEG registration and wireless signal transmission. Thus, it enables quick and easy application of EEG devices by users themselves. Although a main requirement for the interpretation of an EEG is good signal quality, there is a lack of research on this topic in relation to new devices. In our work, we compared the signal quality of six very different EEG devices. On six consecutive days, 24 subjects wore each device for 60 min and completed tasks and games on the computer. The registered signals were evaluated in the time and frequency domains. In the time domain, we examined the percentage of artifact-contaminated EEG segments and the signal-to-noise ratios. In the frequency domain, we focused on the band power variation in relation to task demands. The results indicated that the signal quality of a mobile, gel-based EEG system could not be surpassed by that of a gel-free system. However, some of the mobile dry-electrode devices offered signals that were almost comparable and were very promising. This study provided a differentiated view of the signal quality of emerging mobile and gel-free EEG recording technology and allowed an assessment of the functionality of the new devices. Hence, it provided a crucial prerequisite for their general application, while simultaneously supporting their further development.

Dual Frequency Head Maps: A New Method for Indexing Mental Workload Continuously during Execution of Cognitive Tasks

One goal of advanced information and communication technology is to simplify work. However, there is growing consensus regarding the negative consequences of inappropriate workload on employees health and the safety of persons. In order to develop a method for continuous mental workload monitoring, we implemented a task battery consisting of cognitive tasks with diverse levels of complexity and difficulty. We conducted experiments and registered the electroencephalogram (EEG), performance data, and the NASA-TLX questionnaire from 54 people. Analysis of the EEG spectra demonstrates an increase of the frontal theta band power and a decrease of the parietal alpha band power, both under increasing task difficulty level. Based on these findings we implemented a new method for monitoring mental workload, the so-called Dual Frequency Head Maps (DFHM) that are classified by support vectors machines (SVMs) in three different workload levels. The results are in accordance with the expected difficulty levels arising from the requirements of the tasks on the executive functions. Furthermore, this article includes an empirical validation of the new method on a secondary subset with new subjects and one additional new task without any adjustment of the classifiers. Hence, the main advantage of the proposed method compared with the existing solutions is that it provides an automatic, continuous classification of the mental workload state without any need for retraining the classifier—neither for new subjects nor for new tasks. The continuous workload monitoring can help ensure good working conditions, maintain a good level of performance, and simultaneously preserve a good state of health.

Towards a Continuous Method for Mental Workload Registration

Continuous mental workload registration is a key technology for evaluating and optimizing work conditions in human-machine systems. Despite the urgent need for this technology, its technical measurement is still lacking. The long-term goal of this work is the establishment of precisely such an objective method. The article describes the development of a continuous method for neuronal mental workload registration during the execution of cognitive tasks. The sample consists of 54 people in paid work. The electroencephalogram as well as further workload relevant biosignal data and the NASA-TLX as a subjective questionnaire method are registered. Results from the workload classification of the EEG segments are presented. They are in concordance with the results expected from different task requirements on the executive functions. Findings from the subjective ratings, accuracy rates, and cardiovascular parameters underscore this fact.

Neuronal Mental Workload Registration during Execution of Cognitive Tasks

Neuronal workload measurement is a key-technology for optimizing work conditions in human-machine systems. Specific aims are the identification of neurophysiological parameters indicative for workload and their validation by systematic variation of external load conditions. The battery consists of tasks with diverse complexity and difficulty. The sample consists of 34 people and shows high variability in respect to the cognitive capacity and hence to the experienced mental workload. The electroencephalogram EEG as well as further workload relevant bio signal data and the NASA-TLX as a subjective questionnaire method are registered. Results from the NASA-TLX questionnaire reveal the predominant role of the mental dimension at the implemented task battery. Furthermore, the NASA-TLX indicates the existence of diverse levels of difficulty with several tasks per level. Analysis of EEG spectra demonstrates an increase of frontal theta band power and a decrease of alpha band power with increasing task difficulty level.

Research Design to Access the Mental Workload of Air Traffic Controllers

The German Federal Institute of Occupational Safety and Health in Berlin developed a method for neuronal mental workload monitoring. The so-called Dual Frequency Head Maps (DFHM) method allows defining the workload range of each person individually. The current research project describes the evaluation and condition-related verification of the DFHM method in a simulated realistic environment of an air traffic control center. During an interactive real-time simulation at the Air Traffic Validation Center of the German Aerospace Center, the load level for the controllers was varied by means of two independent variables: the traffic demand and the occurrence of a priority request. Dependent variables for registering mental workload were the DFHM index, heart rate, subjective questionnaires, and air traffic performance data.

Influence of Personal Characteristics and Device Properties on Wearable’s Rank Order

Emerging sensor technology and mobile devices offer the possibility to capture users’ current state more easily. However, the registration of users’ biosignals is still a challenge because of the lack of user acceptance.

A New Method for Mental Workload Registration

Complex and highly automated systems impose high demands on employees with respect to cognitive capacity and the ability to cope with workload. Objectively registering mental workload at workplaces with high cognitive demands would enable prevention of over- and underload. Although urgently needed, such technical measurement is currently unfeasible. Hence, the goal of this work is the establishment of precisely such an objective method.

Artifact Elimination in Neurosciences.

Artifact elimination is a central issue in neurosciences. A method that has established itself as an important part of EEG analysis is the application of independent component analysis (ICA). It decomposes the multi-channel EEG into linearly independent components (ICs) that then can be classified as artifact or EEG signal component. However, classification of the ICs still requires visual, time-intensive inspection by experts.