Browsing by Supervisor "Tangermann, Michael"
Now showing 1 - 7 of 7
Results Per Page
Sort Options
Item A hybrid BCI design using cVEP and eye gaze tracking(2022-06-27) Pamboer, LucaThe code modulated visual evoked potential (cVEP) is a particular response found in electroencephalography (EEG) recordings that can be linked to a specific code represented by a flickering rectangle on an interface. Using this response, a user can control a brain-computer interface (BCI) by shifting gaze location. Eye-tracking (ET) is a method in which the gaze position of a person is predicted based on the position of the pupils. In this study, the benefit of combining cVEP with ET is analysed. The hybrid BCI design makes use of ET to limit the number of commands in both stimulation and classification. In an offline experiment, data from an existing experiment was used, in which 12 participants tested a 6×6 matrix speller BCI. In an analysis on this data, an accuracy of 93% was achieved using a hybrid BCI design with a simulated eye-tracker. Furthermore, an average trial length of 4.29 seconds was reached. Consequently, significant accuracy and ITR improvements were found over a BCI solely based on cVEP. In an online experiment, data was recorded from 8 participants who tested the same 6 × 6 matrix speller BCI. In a post-hoc analysis of the recorded data, a significant increase of 7% in accuracy and 9.70 bits/min for ITR was found when only limiting the classification space. An increase in accuracy and ITR when limiting both stimulation and classification was present but found to be insignificant. The results suggest that limiting the number of commands during stimulation is sub-optimal for the performance of a BCI. In contrast, limiting the commands during classification results in a significant performance improvement in terms of accuracy and ITR. Webcam-based ET could be used to make this limitation.Item Extending a German Aphasia Rehabilitation BCI into Dutch and English Domains(2022-03-11) Wit, de, KimBrain-computer interfaces provide a way for brain-controlled devices, eliminating the need for limb control. Auditory BCIs further give rise to possibility for motor-impaired patients, specifically since some auditory BCIs do not require eye-gaze control. German native post-stroke aphasic patients suffering from damage in language-specific neural correlates have been shown to regain a higher level of communicative capabilities and increased scores in clinical aphasia assessments after training with a novel auditory ERP-based BCI paradigm for aphasia rehabilitation. This BCI paradigm utilizes a spatial speaker set-up, effectively introducing a spatial component to stimuli, which was shown to increase classification on target versus nontarget stimuli based on the classic oddball paradigm. Stimuli are comprised of a preceding cueing sentence, with the last word missing. The present study aims to gather apt stimuli sets in Dutch and English domains, and setting up the experiment such that pilot studies can be held shortly after. Stimuli sets were found in both Dutch and English, and recordings of these have been made with native speakers. Furthermore, organizational requirements to commence an EEG study have been fulfilled, and a detailed description of future steps for conducting a pilot study in Dutch has been provided.Item How to Build a Faster Code-Modulated Visual Evoked Potential based Brain-Computer Interface: The Impact of Codes Bachelor’s Thesis in Artificial Intelligence(2023-01-20) Cornielje, GijsBrain-computer interface (BCI) speller applications are a way for physically limited people to communicate. The type of BCI in this research will use electroencephalography (EEG) to measure the users’ visually evoked potentials (VEP) to decode what character the user is looking at. These VEPs are induced by presenting flash patterns generated by pseudorandom patterns. To be able to distinguish these flashing patterns from one another, they are required to have low correlation. Generating these flash patterns is not trivial, which is why several types of code sets have been used in earlier research, with m-sequences being used the most. This research aims to find alternatives to this code set which achieve higher Information Transfer Rates (ITRs), or induce less eyestrain. These alternatives are: de Bruijn, Gold and Golay code sets. Additionally, it aims to find the impact modulation has on these metrics. Results are obtained by performing a 320 trial copyspelling task in addition to an eyestrain rating across 10 conditions for 12 participants. Results are analysed using reconvolution and canonical correlation analysis (CCA). Not modulating appeared to perform significantly better than not modulating in ITR (122.0 bits per minute > 111.4 bits per minute, p-value < 0.05). It also caused significantly less eyestrain (not modulated: 3.9 < modulated: 5.5, p < 0.001). No significant differences were found in ITR or eyestrain across the different code sets. The highest performing code set reached an ITR of 138.5 bits per minute averaged over all participants. Higher performance was found by not modulated codes, whilst no differences were found between different types of code sets. Variations in setup and decoding could potentially be limitations and explain differences with other research.Item Optically pumped magnetometers for a brain-computer interface based on event-related desynchronization(2022-10-07) Zerfowski, janBackground Stroke is one of the leading causes of disability worldwide and often responsible for impairments of hand motor function. Rehabilitation and restoration of motor functions can be significantly improved using devices controlled by brain signals, so called brain-computer interfaces (BCIs). Most current BCI systems are based on electroencephalography (EEG), which provides only limited spatial resolution and thus limited versatility of control commands. Compared to EEG, optically pumped magnetometers (OPMs) measure cortical magnetic fields without contact to the scalp and provide a higher spatial resolution and bandwidth. In contrast to superconducting quantum interference device (SQUID)-based magnetoencephalography (MEG), OPMs have low maintenance cost and allow movement in the scanner, making them more applicable in clinical contexts. Methods We quantify the signal characteristics of a commercially available OPM system (FieldLine Inc., USA) in terms of noise floor, dynamic range and bandwidth to verify its suitability for cortical measurements. We then develop an experiment contrasting resting and right hand grasping imagery to measure modulations of the sensorimotor rhythm (SMR) with 17 OPMs over the left motor cortex of 18 healthy participants. The BCI capabilities of the OPM acquisition system are evaluated with a modular near real-time classification pipeline, which provides visual feedback to the user. Results The sensor characterization revealed a system noise floor of about 27 fT/ √ Hz at 10 Hz, a bandwidth of 400 Hz and a dynamic range of ±15 nT, fulfilling the minimum requirements for cortical measurements. In 10 of 16 eligible participants, a difference in SMR power between resting and grasping condition could be identified. We show that OPMs are suitable to measure SMR modulations in near real-time and that the classification performance of our pipeline significantly exceeds chance level. Discussion OPMs allow for the online quantification of voluntary modulations of the sensorimotor rhythm on single-trial basis, a central requirement for many BCI systems used in the rehabilitation of stroke survivors. With their higher spatial resolution compared to EEG, OPMs could be used for more complex classification paradigms and ultimately facilitate a development towards more versatile BCI applications. The increasing availability and sensitivity of commercialized OPM systems allows for the exploration of MEG in new research areas. OPMs are projected to become an important tool in the field of cognitive neuroscience within the next few years. Keywords: brain-computer interface, optically pumpedItem Optimal Stimulus Conditions to Improve User Experience in Brain Computer Interfaces(2022-08-01) Scheppink, HannekeCurrent Brain Computer Interfaces (BCIs) used for spelling are quite fatiguing and uncomfortable for the participant to use. It is proposed that certain stimulus adaptions, such as implementing coloured stimuli, can inhibit this. This thesis studied the effect of adapting the colour and structure of checkerboard pattern stimuli on the user’s fatigue and comfort, using a code-modulated visual evoked potential (c-VEP) based BCI speller interface. The main focus was on improving this comfort while maintaining a good system performance. Using five different conditions, it was found that there is a trade-off between system performance and comfort, and that a choice needs to be made according to the purpose of the system. It was concluded that a black-white solid flashing condition was the best performing stimulus in terms of accuracy of the system, while a violet-grey checkerboard appeared to be the best condition in terms of user-comfort.Item Predicting c-VEP-based BCI performance to study BCI illiteracy(2023-01-27) Werff, van der, FlorisA brain–computer interface (BCI) is a system that can deduce the intent of a user from recordings of their brain activity. This allows users to control a computer application by brain activity, which can be measured, e.g., by electroencephalography (EEG). After approximately 50 years of BCI research, the success that BCI control can provide still greatly varies from subject-to-subject. About 15% to 30% of sensorimotor rhythm-based (SMR) BCI users do not reach the criterion level of 70% accuracy, which was determined to be the threshold for meaningful spelling. This is in the literature known as the BCI illiteracy phenomenon. A myriad of variables are studied alongside of subjects testing BCIs in order to determine if these variables correlate with BCI performance. If they do, then they could potentially be used as predictors. The development of predictors of BCI performance serves two purposes: it may lead to a better understanding of the BCI-illiteracy phenomenon and these predictors could be used as a screening tool to inform users about poor expected performance, among other things. An experiment was conducted in which six predictors were analysed by means of a linear and a multivariate regression analysis alongside of the state-ofthe- art paradigm code-modulated visual evoked potential-based (c-VEP) BCI performance. The six predictors were: relative visual alpha power during resting state with eyes open and closed, heart rate variability, attention span measured by the error rate of the Sustained Attention to Response Task (SART) paradigm and flash-VEP latency and amplitude. There were no significant (p > 0.008) Pearson’s correlation coefficients found for these predictors with N=16 and all subjects were able to obtain sufficient BCI control when they were given enough time. It was concluded that subject-to-subject variance is not in accuracy for c-VEP-based BCIs, but in time. The multivariate regression model could be used as a screening tool given its root mean square error (RMSE) of 14.084% for a trial size of 1.05 seconds.Item The online optimization of brain-computer interface stimulus parameters, a simulation(2023-08-20) Thoni, ChiaraA brain-computer interface operates by presenting stimuli to elicit brain responses that can be acted upon. As the discriminability of these responses is subject to the stimulus characteristics, the presentation of the optimal stimuli could improve the performance of the interface. Unfortunately, these optima are subject-specific. Furthermore, the optimization of said stimuli is complicated by the potentially high level of non-independent and identically distributed noise that comes with the recorded responses. The aim of this thesis is to demonstrate that these challenges can be alleviated by introducing the optimization process with heteroskedatic modelling and the replication of existing designs. To this end, various Bayesian optimization algorithms are tested on simulations that are designed to resemble the aforementioned online optimization objective. The compared optimization algorithms differ in the surrogate models, replication schemes and selection strategies that have been used. As it is impossible to change the stimulus characteristics that have been used to record existing datasets, the simulated objective functions model the response discriminability as a function of the decoding parameters. To assess the effectiveness of the optimization algorithms, the algorithms are tested on one, two and seven-dimensional objective functions. The results suggest that a Bayesian optimization algorithm that is enriched with heteroskedastic Gaussian process regression and the locationbased evaluation of existing designs significantly outperforms the random sampling baseline for all objective functions that have been considered. Even though the thesis is focused at braincomputer interfaces, application of the results is widespread since heteroskedastic noise can also be encountered with black-box optimization within other domains.