Individuals with severe impairments may use brain-computer interface (BCI) technologies in order to interact with their external environment. One non-invasive brain-monitoring technology which may be suitable for this purpose is transcranial Doppler ultrasound (TCD). Previous research has shown that TCD is useful in detecting changes in cerebral blood flow velocities after the performance of cognitive tasks which are often lateralized towards a specific hemisphere of the brain. However, to date, TCD has not been used in a BCI system. This thesis first explores TCD in an offline study, showing that on average, accuracies of 80.0% are attainable with user-specific training data and 74.6% with user-independent training data. Furthermore, consecutive sequential lateralizations do not decrease classification accuracies. In a subsequent online experiment, a TCD-BCI system yielded an average accuracy of 61.4%, but revealed key findings about the effects of user motivation and error streaks in an online system.

Individuals with severe impairments may use brain-computer interface (BCI) technologies in order to interact with their external environment. One non-invasive brain-monitoring technology which may be suitable for this purpose is transcranial Doppler ultrasound (TCD). Previous research has shown that TCD is useful in detecting changes in cerebral blood flow velocities after the performance of cognitive tasks which are often lateralized towards a specific hemisphere of the brain. However, to date, TCD has not been used in a BCI system. This thesis first explores TCD in an offline study, showing that on average, accuracies of 80.0% are attainable with user-specific training data and 74.6% with user-independent training data. Furthermore, consecutive sequential lateralizations do not decrease classification accuracies. In a subsequent online experiment, a TCD-BCI system yielded an average accuracy of 61.4%, but revealed key findings about the effects of user motivation and error streaks in an online system.

Many individuals with severe motor disabilities may not be able to use conventional means of emotion expression (e.g. vocalization, facial expression) to make their emotions known to others. Lack of a means for expressing emotions may adversely affect the quality of life of these individuals and their families. The main objective of this thesis was to implement a non-invasive means of identifying emotional arousal (neutral vs. intense) and valence (positive vs. negative) by directly using brain activity. In this light, near infrared spectroscopy (NIRS), which optically measures oxygenated and deoxygenated hemoglobin concentrations ([HbO2] and [Hb], respectively), was used to monitor prefrontal cortex hemodynamics in 10 individuals as they listened to music excerpts. Participants provided subjective ratings of arousal and valence. With respect to valence and arousal, prefrontal cortex [HbO2] and [Hb] were characterized and significant prefrontal cortex hemodynamic modulations were identified due to emotions. These modulations were not significantly related to the characteristics of the music excerpts used for inducing emotions. These early investigations provided evidence for the use of prefrontal cortex NIRS in identifying emotions. Next, using features extracted from [HbO2] and [Hb] in the prefrontal cortex, an average accuracy of 71% was achieved in identifying arousal and valence. Novel hemodynamic features extracted using dynamic modeling and template-matching were introduced for identifying arousal and valence. Ultimately, the ability of autonomic nervous system (ANS) signals including heart rate, electrodermal activity and skin temperature to improve the identification results, achieved when using PFC [HbO2] and [Hb] exclusively, was investigated. For the majority of the participants, prefrontal cortex NIRS-based identification achieved higher classification accuracies than combined ANS and NIRS features. The results indicated that NIRS recordings of the prefrontal cortex during presentation of music with emotional content can be automatically decoded in terms of both valence and arousal encouraging future investigation of NIRS-based emotion detection in individuals with severe disabilities.

This study aims at designing and implementing a single channel stand-alone Brain-Controlled Switch (BCS) device, which records the electroencephalography (EEG) signals from the scalp using electrodes, amplifies it, eliminates interferences (associated with the EEG signals) and processes the EEG signals to extract and decode temporal signal features to determine user’s intention of regulating an external switch. The design of our “brain-controlled switch” device is implemented using a bio-potential amplifier and a microcontroller. The bio-potential amplifier amplifies the EEG signals to a level sufficient for processing, eliminates interferences and ensures patient safety. The microcontroller (dsPIC30F4013) digitizes the amplified and conditioned analog EEG signals from the bio-potential amplifier, extracts the desired signal features for decoding and prediction of user’s intention and accordingly operates the external switch. When the user concentrates on an external visual stimulus or performs externally triggered movement (hand movement or motor imagery movement), a reproducible pattern appears in user’s EEG frequency bands. The analysis of these patterns is used to decode and predict user’s intention to operate an external switch. To realize our “brain-controlled switch”, we explored two EEG sources: steady-state visually evoked potentials (SSVEP) and beta rebounds, which are patterns generated in the EEG frequency bands associated with focusing on an external visual stimulus or performing externally triggered movements. In case of SSVEP based brain controlled switch, a repetitive visual stimulus (LED flickering at a specified frequency) was used. When the user concentrates on the flickering LED, a dominant fundamental frequency (equivalent to the flickering frequency) appears in the spectral representation of the EEG signals recorded at occipital lobes. Our microcontroller implemented a digital band pass filter to extract the frequency band containing this fundamental frequency and continuously took an average of the amplitude power every predetermined time interval. Whenever the amplitude average power exceeded the preset power threshold the external switch was turned ON. A healthy subject participated in this study, and it took approximately 3.14 ± 1.81 seconds of active concentration for the subject to turn ON the switch in real time with a false positive rate of 1.17%. In case of beta rebound based brain controlled switch, the subject was instructed to perform a brisk hand movement following an external synchronization signal. Our design focused on the post-movement beta rebound which occurs after the cessation of the movement to operate the external switch. Our microcontroller in this case implemented a digital band pass filter to extract the beta band and continuously took an average of its amplitude power every predetermined time interval. Whenever the amplitude average power exceeded the preset power threshold the external switch was turned ON. It took approximately 12.23 ± 7.39 seconds of active urging time by the subject to turn ON the switch in real time with a false positive rate of 9.33%. Thus we have designed a novel stand-alone BCS device which operates an external switch by decoding and predicting user’s intentions.

Brain Computer Interface (BCI) technology can be important for those unable to communicate due loss of muscle control. The P300 Speller allows communication at a rate up to eight selections per minute. Given this relatively slow rate of communication highly accurate classification is of great importance. Previous studies have shown that alternative stimuli (e.g., faces) can improve BCI speed and accuracy. The present study uses two new alternative stimuli, locations and graspable tools in a two-stimulus paradigm. Functional MRI studies have shown that images of familiar locations produce brain responses in the parahippocampal place area and graspable tools produce brain responses in premotor cortex.The current study shows that location and tool stimuli produce unique brain responses that can be used for classification in the two-stimulus paradigm. This study shows proof of concept for using two unique stimuli to improve speed and accuracy of the P300 Speller.

Self-paced Brain computer interface (SBCI) systems allow individuals with motor disabilities to use their brain signals to control devices, whenever they wish. These systems are required to identify the user’s “intentional control (IC)” commands and they must remain inactive during all periods in which users do not intend control (called “no control (NC)” periods). This dissertation addresses three issues related to the design of SBCI systems: 1) their presently high false positive (FP) rates, 2) the presence of artifacts and 3) the identification of a suitable evaluation metric. To improve the performance of SBCI systems, the following are proposed: 1) a method for the automatic user-customization of a 2-state SBCI system, 2) a two-stage feature reduction method for selecting wavelet coefficients extracted from movement-related potentials (MRP), 3) an SBCI system that classifies features extracted from three neurological phenomena: MRPs, changes in the power of the Mu and Beta rhythms; 4) a novel method that effectively combines methods developed in 2) and 3 ) and 5) generalizing the system developed in 3) for detecting a right index finger flexion to detecting the right hand extension. Results of these studies using actual movements show an average true positive (TP) rate of 56.2% at the FP rate of 0.14% for the finger flexion study and an average TP rate of 33.4% at the FP rate of 0.12% for the hand extension study. These FP results are significantly lower than those achieved in other SBCI systems, where FP rates vary between 1-10%. We also conduct a comprehensive survey of the BCI literature. We demonstrate that many BCI papers do not properly deal with artifacts. We show that the proposed BCI achieves a good performance of TP=51.8% and FP=0.4% in the presence of eye movement artifacts. Further tests of the performance of the proposed system in a pseudo-online environment, shows an average TP rate =48.8% at the FP rate of 0.8%. Finally, we propose a framework for choosing a suitable evaluation metric for SBCI systems. This framework shows that Kappa coefficient is more suitable than other metrics in evaluating the performance during the model selection procedure.

Individuals with severe physical impairments have a reduced ability to communicate through movement and speech. We investigated transcranial Doppler ultrasound as a potential measurement modality for a novel brain-computer interface. It was hypothesized that cognitive activity would result in detectable changes in cerebral blood flow velocity within the middle cerebral arteries. Nine able-bodied participants alternated between rest and two different mental activities - silent word generation and mental rotation. Two analyses were performed to assess the feasibility and practicality of a TCD-based brain-computer interface. Both mental activities were independently differentiated from rest with high accuracy. Intuitive time-domain features were sufficient for classification. Data transmission rate was quadrupled by differentiating all three classes simultaneously using shorter state durations. Transcranial Doppler ultrasound can be used to automatically detect cognitive activity and may be useful as the basis of a brain-computer interface.

Individuals with severe physical impairments have a reduced ability to communicate through movement and speech. We investigated transcranial Doppler ultrasound as a potential measurement modality for a novel brain-computer interface. It was hypothesized that cognitive activity would result in detectable changes in cerebral blood flow velocity within the middle cerebral arteries. Nine able-bodied participants alternated between rest and two different mental activities - silent word generation and mental rotation. Two analyses were performed to assess the feasibility and practicality of a TCD-based brain-computer interface. Both mental activities were independently differentiated from rest with high accuracy. Intuitive time-domain features were sufficient for classification. Data transmission rate was quadrupled by differentiating all three classes simultaneously using shorter state durations. Transcranial Doppler ultrasound can be used to automatically detect cognitive activity and may be useful as the basis of a brain-computer interface.

A commercially available brain-body interface, the `Cyberlink' which was developed by a Dr Andrew Junker, has been evaluated as a potential interface device for persons with a severe disability such as traumatic brain injury. The literature concerning brain computer interfaces and other input devices is surveyed and it is shown there is a need to investigate the Cyberlink as an assistive technology device for persons with a disability. The investigation was carried out in four phases, using forty-four persons with and without physical, mental and sensory impairments as participants. The first phase consisted of a survey of common assistive technology devices along with the Cyberlink. This demonstrated that many users were able to operate alternative devices. The second phase identified a group of distinct users that could only use a Cyberlink to both recreate and communicate with the outside world. These participants formed the focus group. A modified contextual inquiry and design was performed at the same time as the phase two studies. The data collected from the contextual inquiry and design drove the design for a communication application, developed in phase three, that gave the focus group the opportunity to select yes and no answers to questions. Phase four was the testing phase of the new yes / no application. This identified some design flaws that were addressed following a target acquisition study which showed that some paths in the design were difficult to steer through. New prototypes were created and tested using this data. The final yes / no program allowed the focus group to select yes and no answers on prompting, albeit with a les's than 100% success rate. Success appeared to depend on the focus group not beirighampered by the inconsistent debilitation of their injuries and medications. The utility of the Cyberlink for the focus group for recreating and performing elementary communication is thus demonstrated for occasions when settings are relevant, medications are not dampening bio-signals, and the inconsistencies of the brain injury allow them to control the cursor.

Self-paced Brain computer interface (SBCI) systems allow individuals with motor disabilities to use their brain signals to control devices, whenever they wish. These systems are required to identify the user’s “intentional control (IC)” commands and they must remain inactive during all periods in which users do not intend control (called “no control (NC)” periods). This dissertation addresses three issues related to the design of SBCI systems: 1) their presently high false positive (FP) rates, 2) the presence of artifacts and 3) the identification of a suitable evaluation metric. To improve the performance of SBCI systems, the following are proposed: 1) a method for the automatic user-customization of a 2-state SBCI system, 2) a two-stage feature reduction method for selecting wavelet coefficients extracted from movement-related potentials (MRP), 3) an SBCI system that classifies features extracted from three neurological phenomena: MRPs, changes in the power of the Mu and Beta rhythms; 4) a novel method that effectively combines methods developed in 2) and 3 ) and 5) generalizing the system developed in 3) for detecting a right index finger flexion to detecting the right hand extension. Results of these studies using actual movements show an average true positive (TP) rate of 56.2% at the FP rate of 0.14% for the finger flexion study and an average TP rate of 33.4% at the FP rate of 0.12% for the hand extension study. These FP results are significantly lower than those achieved in other SBCI systems, where FP rates vary between 1-10%. We also conduct a comprehensive survey of the BCI literature. We demonstrate that many BCI papers do not properly deal with artifacts. We show that the proposed BCI achieves a good performance of TP=51.8% and FP=0.4% in the presence of eye movement artifacts. Further tests of the performance of the proposed system in a pseudo-online environment, shows an average TP rate =48.8% at the FP rate of 0.8%. Finally, we propose a framework for choosing a suitable evaluation metric for SBCI systems. This framework shows that Kappa coefficient is more suitable than other metrics in evaluating the performance during the model selection procedure.