Abstract
Clinical assessments of individuals with cognitive-motor dissociation (CMD) following brain injury are both challenging and prone to error. Previous studies have shown that electroencephalographic (EEG)-based Brain-Computer Interface (BCI) protocols for Motor-Command Following (MCF), along with differences in the N1 and P3 components of auditory evoked potentials (AEPs) in response to an auditory oddball paradigm, provide a more accurate and quantitative assessment of CMD in children [1]. This study explores whether the combination of motor-imagery BCI (MI-BCI) and auditory evoked Event Related Potentials (ERPs) using an oddball paradigm can produce complimentary EEG-markers, to improve the assessment of adults with prolonged disorders of consciousness (PDoC).
Methods: EEG data were collected from (N=9) individuals with PDoC, including cases of unresponsive wakefulness syndrome (UWS, n = 2) and minimally conscious state (MCS, n = 3) and four cases of locked-in syndrome (LIS), using a 16-channel g.Nautilus system (g.tec). The MI-BCI protocol [2] involved up to 12 sessions of 240 trials each. In the initial six sessions, participants were trained, with and without feedback, to consistently imagine moving one of two limbs, such as the left or right hand, in response to auditory cues. From the seventh session onward, this binary imagery task was integrated into a closed yes/no question-and-answer format. Separately, the auditory oddball protocol included at least two sessions, approximately 10 days apart. Each session consisted of two five-minute sets of auditory stimuli, featuring 340ms square-wave beeps at frequencies of 400 Hz (standard) or 575 Hz (deviant) frequencies, interspersed with novel sounds, in a standard-deviant-novel ratio of 27:8:6 per set.
Results: Significant group differences were found in mean N1 latencies, with shorter latencies for the LIS and MCS groups compared to the UWS group (LIS vs. UWS and MCS vs. UWS, p < 0.001). Additionally, N1 latencies were negatively correlated with the mean decoding accuracies (DA) from significant motor-imagery runs (i.e., runs where the peak DA during the task period was significantly higher than during the baseline period), indicating that as mean DA increased, mean N1 latencies decreased (two-tailed, p = 0.017). The magnitude of the N2 and P3 components was measured as the variance between the largest negative SNR for N2 and the largest positive SNR for P3 in response to deviant/novel stimuli versus standard stimuli. While the N2 difference wave magnitude was found to be significantly larger for the MCS group compared to both the UWS and LIS groups (p < 0.01), it did not significantly differ between the UWS and LIS groups. However, a negative correlation was observed between N2 latency and difference wave magnitude, with larger magnitudes occurring when latencies were closer to normal. Finally, no significant differences were observed in P3 magnitude or latency across groups.
Conclusion: The observed anti-correlation between N1 latencies and motor-imagery decoding accuracy underscores the value of these complementary protocols for more precise assessment of consciousness in adults with PDoC. The findings highlight the potential for the combined implementation of these MI-BCI and auditory oddball protocols, to provide objective movement-independent EEG biomarkers.
References
[1] N. Kim et al., “Objective neurophysiological markers of cognition after pediatric brain injury,” Neurol. Clin. Pract., 2022, doi: 10.1212/cpj.0000000000200066.
[2] D. Coyle et al., “Towards electroencephalography-based consciousness assessment and cognitive function profiling in prolonged disorders of consciousness,” Res. Sq., 2022, doi: 10.21203/rs.3.rs-2349135/v1.
Methods: EEG data were collected from (N=9) individuals with PDoC, including cases of unresponsive wakefulness syndrome (UWS, n = 2) and minimally conscious state (MCS, n = 3) and four cases of locked-in syndrome (LIS), using a 16-channel g.Nautilus system (g.tec). The MI-BCI protocol [2] involved up to 12 sessions of 240 trials each. In the initial six sessions, participants were trained, with and without feedback, to consistently imagine moving one of two limbs, such as the left or right hand, in response to auditory cues. From the seventh session onward, this binary imagery task was integrated into a closed yes/no question-and-answer format. Separately, the auditory oddball protocol included at least two sessions, approximately 10 days apart. Each session consisted of two five-minute sets of auditory stimuli, featuring 340ms square-wave beeps at frequencies of 400 Hz (standard) or 575 Hz (deviant) frequencies, interspersed with novel sounds, in a standard-deviant-novel ratio of 27:8:6 per set.
Results: Significant group differences were found in mean N1 latencies, with shorter latencies for the LIS and MCS groups compared to the UWS group (LIS vs. UWS and MCS vs. UWS, p < 0.001). Additionally, N1 latencies were negatively correlated with the mean decoding accuracies (DA) from significant motor-imagery runs (i.e., runs where the peak DA during the task period was significantly higher than during the baseline period), indicating that as mean DA increased, mean N1 latencies decreased (two-tailed, p = 0.017). The magnitude of the N2 and P3 components was measured as the variance between the largest negative SNR for N2 and the largest positive SNR for P3 in response to deviant/novel stimuli versus standard stimuli. While the N2 difference wave magnitude was found to be significantly larger for the MCS group compared to both the UWS and LIS groups (p < 0.01), it did not significantly differ between the UWS and LIS groups. However, a negative correlation was observed between N2 latency and difference wave magnitude, with larger magnitudes occurring when latencies were closer to normal. Finally, no significant differences were observed in P3 magnitude or latency across groups.
Conclusion: The observed anti-correlation between N1 latencies and motor-imagery decoding accuracy underscores the value of these complementary protocols for more precise assessment of consciousness in adults with PDoC. The findings highlight the potential for the combined implementation of these MI-BCI and auditory oddball protocols, to provide objective movement-independent EEG biomarkers.
References
[1] N. Kim et al., “Objective neurophysiological markers of cognition after pediatric brain injury,” Neurol. Clin. Pract., 2022, doi: 10.1212/cpj.0000000000200066.
[2] D. Coyle et al., “Towards electroencephalography-based consciousness assessment and cognitive function profiling in prolonged disorders of consciousness,” Res. Sq., 2022, doi: 10.21203/rs.3.rs-2349135/v1.
| Original language | English |
|---|---|
| Number of pages | 1 |
| Publication status | Unpublished - 3 Oct 2024 |