중증 근신경계 질환 환자들의 의사소통을 위한 뇌파 기반 뇌-컴퓨터 인터페이스의 개발

Abstract

For the past decades, various brain-computer interface (BCI) systems have been developed for the patients with neuromuscular diseases. However, there has been several challenging issues in terms of practicality in the previous BCI studies. Firstly, there was an issue related to optimization of tasks for BCI. Secondly, The BCI studies for patients with severe neuromuscular diseases have not been sufficiently performed. Lastly, possible EEG-based BCI systems for long-term use have not been developed yet. To overcome these challenging issues of the previous BCI studies, the author performed three studies. First of all, for optimizing mental tasks for EEG-based BCI systems, the author investigated best or worst combinations of mental tasks. Nine healthy participants were enrolled in this study and conducted eight different mental tasks. While they perform the mental tasks, the EEG data were recorded using thirty electrodes on their whole head. The classification accuracies for all possible mental task combinations were calculated using the EEG data. From the results of this study, the author could confirm that the combinations of distinct non-motor imagery tasks tended to result in higher classification accuracy (e.g., combinations of “mental character writing” with “mental multiplication” or “mental subtraction”) than combinations of solely motor imagery tasks. Certain combinations of motor and non-motor imagery tasks also showed relatively high classification accuracy (e.g., combinations of “mental character writing” and/or “mental multiplication” with “any motor imagery tasks (left/right/tongue)” and of “mental subtraction” with “two motor imagery tasks (left/tongue)). On the other hand, brain activity patterns elicited by “left hand motor imagery”, “mental singing”, and “right hand motor imagery” were not well discriminated when these tasks were performed simultaneously. Moreover, performing similar type of mental tasks at the same time (e.g., mental multiplication with subtraction) did not result in high classification accuracy even though they were frequently included in the best mental task combinations. It is expected that our results reporting the best and worst combinations of mental tasks will be a useful reference to reduce the time needed for preliminary tests when discovering individual-specific mental task combinations. Also, the author implemented an EEG-based BCI paradigm for online binary communication by patients with severe neuromuscular diseases and evaluated its feasibility with a female patient in completely locked-in state (CLIS) due to severe amyotrophic lateral sclerosis (ALS), who had not communicated even with her family for more than one year. An online classification accuracy of 87.5 % was achieved when Riemannian geometry-based classification was applied to real-time EEG data recorded while the patient was performing one of two distinct mental imagery tasks for 5 s. To the best of the author’s knowledge, this is the first report of successful application of an EEG-based BCI for online communication with a patient in CLIS. Finally, the author developed an asynchronous hybrid BCI paradigm for long-term use. The users could operate the respiration-based PPG switch by holding their breath for about ten seconds. The PPG switch carried out a role converting operation modes such as no control (NC) or intentional control (IC) state. Once the PPG switch was operated by the users, the participants could control four environmental devices (an electric fan, a heater, a lamp, and an emergency alarm) using a steady-state visual evoked potential (SSVEP)-based BCI paradigm. From the experimental results, the author could confirm that the proposed BCI paradigm had very low FPR. The averaged FPR of the healthy subjects and the FPR of a ALS patient were 0.02 and 0.20 FPs/min. And also, the healthy subjects and the ALS patient could control four environmental devices with high classification accuracies of 88.57 and 100.00 %, respectively. These results show that the asynchronous BCI paradigm with the respiration-based PPG switch might be successfully used in practical environment for patients in LIS. In summary, the author performed three different BCI studies to develop practical BCI technologies for patients with neuromuscular diseases. The author hopes that these experimental results can improve the life’s quality of ALS patients with neuromuscular diseases.