TY - BOOK

T1 - Design and Evaluation of a Hardware System for Online Signal Processing within Mobile Brain-Computer Interfaces

AU - Wahalla, Marc-Nils

PY - 2024/2/8

Y1 - 2024/2/8

N2 - Brain-Computer Interfaces (BCIs) are innovative systems that enable direct communication between the brain and external devices. These interfaces have emerged as a transformative solution not only for individuals with neurological injuries, but also for a broader range of individuals, encompassing both medical and non-medical applications. Historically, the challenge of neurological injury being static after an initial recovery phase has driven researchers to explore innovative avenues. Since the 1970s, BCIs have been at one forefront of these efforts. As research has progressed, BCI applications have expanded, showing potential in a wide range of applications, including those for less severely disabled (e.g. in the context of hearing aids) and completely healthy individuals (e.g. entertainment industry). However, the future of BCI research also depends on the availability of reliable BCI hardware to ensure real-world application. The CereBridge system designed and implemented in this work represents a significant leap forward in brain-computer interface technology by integrating all EEG signal acquisition and processing hardware into a mobile system. The processing hardware architecture is centered around an FPGA with an ARM Cortex-M3 within a heterogeneous IC, ensuring flexibility and efficiency in EEG signal processing. The modular design of the system, consisting of three individual boards, ensures adaptability to different requirements. With a focus on full mobility, the complete system is mounted on the scalp, can operate autonomously, requires no external interaction, and weighs approximately 56g, including 16 channel EEG sensors. The proposed customizable dataflow concept facilitates the exploration and seamless integration of algorithms, increasing the flexibility of the system. This is further underscored by the ability to apply different algorithms to recorded EEG data to meet different application goals. High-Level Synthesis (HLS) was used to port algorithms to the FPGA, accelerating the algorithm development process and facilitating rapid implementation of algorithm variants. Evaluations have shown that the CereBridge system is capable of integrating the complete signal processing chain required for various BCI applications. Furthermore, it can operate continuously for more than 31 hours with a 1800mAh battery, making it a viable solution for long-term mobile EEG recording and real-world BCI studies. Compared to existing research platforms, the CereBridge system offers unprecedented performance and features for a mobile BCI. It not only meets the relevant requirements for a mobile BCI system, but also paves the way for the rapid transition of algorithms from the laboratory to real-world applications. In essence, this work provides a comprehensive blueprint for the development and implementation of a state-of-the-art mobile EEG-based BCI system, setting a new benchmark in BCI hardware for real-world applicability.

AB - Brain-Computer Interfaces (BCIs) are innovative systems that enable direct communication between the brain and external devices. These interfaces have emerged as a transformative solution not only for individuals with neurological injuries, but also for a broader range of individuals, encompassing both medical and non-medical applications. Historically, the challenge of neurological injury being static after an initial recovery phase has driven researchers to explore innovative avenues. Since the 1970s, BCIs have been at one forefront of these efforts. As research has progressed, BCI applications have expanded, showing potential in a wide range of applications, including those for less severely disabled (e.g. in the context of hearing aids) and completely healthy individuals (e.g. entertainment industry). However, the future of BCI research also depends on the availability of reliable BCI hardware to ensure real-world application. The CereBridge system designed and implemented in this work represents a significant leap forward in brain-computer interface technology by integrating all EEG signal acquisition and processing hardware into a mobile system. The processing hardware architecture is centered around an FPGA with an ARM Cortex-M3 within a heterogeneous IC, ensuring flexibility and efficiency in EEG signal processing. The modular design of the system, consisting of three individual boards, ensures adaptability to different requirements. With a focus on full mobility, the complete system is mounted on the scalp, can operate autonomously, requires no external interaction, and weighs approximately 56g, including 16 channel EEG sensors. The proposed customizable dataflow concept facilitates the exploration and seamless integration of algorithms, increasing the flexibility of the system. This is further underscored by the ability to apply different algorithms to recorded EEG data to meet different application goals. High-Level Synthesis (HLS) was used to port algorithms to the FPGA, accelerating the algorithm development process and facilitating rapid implementation of algorithm variants. Evaluations have shown that the CereBridge system is capable of integrating the complete signal processing chain required for various BCI applications. Furthermore, it can operate continuously for more than 31 hours with a 1800mAh battery, making it a viable solution for long-term mobile EEG recording and real-world BCI studies. Compared to existing research platforms, the CereBridge system offers unprecedented performance and features for a mobile BCI. It not only meets the relevant requirements for a mobile BCI system, but also paves the way for the rapid transition of algorithms from the laboratory to real-world applications. In essence, this work provides a comprehensive blueprint for the development and implementation of a state-of-the-art mobile EEG-based BCI system, setting a new benchmark in BCI hardware for real-world applicability.

U2 - 10.15488/16149

DO - 10.15488/16149

M3 - Doctoral thesis

CY - Hannover

ER -