| Summary: | The dynamics of brain states under general anesthesia in infants are complex and exhibit significant developmental changes, particularly in the context of neurophysiological responses. Traditional EEG analysis has been valuable in tracking these changes, but there is a critical need for more precise, quantitative methods to assess neural synchrony and coherence in this vulnerable population. This thesis explores advanced state-space modeling techniques, specifically focusing on State Space Global Coherence (SSGC), to estimate global coherence (GC) during sevoflurane general anesthesia in an infant. Two different SSGC approaches were employed: one approach directly estimated GC from the data, while the other first estimated the covariance matrix and then used this matrix to compute GC. The SSGC approaches were first applied to a validation dataset that had been previously analyzed using SSGC for covariance estimation. This was done to ensure that my analysis was functioning correctly by validating it against a dataset with known outcomes before proceeding with exploratory analysis. Once this was certain, the next step involved applying this pipeline to EEG data from a 10-month-old infant—a dataset where SSGC had not been previously utilized. Following this, both the validation dataset and the infant dataset were used to compare the effectiveness of SSGC for covariance estimation versus direct GC estimation. The infant dataset, in particular, provided an opportunity to explore the utility of SSGC in a new context. Both datasets that the SSGC methods were applied to had a low signal to noise ratio. This revealed that direct GC estimation provided improved temporal resolution for GC and the ability to capture dynamic changes in coherence over time. In contrast, SSGC for covariance estimation produced results nearly identical to empirical GC, suggesting that it is more susceptible to noise. The resilience of direct GC estimation to noisy data highlights its potential as a robust tool for capturing the spatiotemporal dynamics of neural synchrony under anesthesia. This thesis emphasizes the importance of advanced modeling techniques in enhancing neurophysiological monitoring, with significant implications for improving pediatric anesthetic care and outcomes.
|
|---|