By Yingying Su, MD, Neuro ICU Professor, Xuanwu Hospital, Capital Medical University
In 1998, when the neuro ICU of Xuanwu Hospital of Capital Medical University was established, we initiated a research project with the Beijing Municipal Science and Technology Commission—a study of bedside brain injury monitoring, evaluation, and brain protection treatment. Since then, with the support of nine scientific research projects, we have carried out studies of neuro-electrophysiology (EEG, EPs and ERPs), neuro-biochemical markers (S-100 and NSE), cerebral blood flow and cerebral perfusion (TCD, ICP and CPP), brain tissue oxygen technology (SjvO2, AVDO2 and OEF), heart rate variability (HRV), and quantitative clinical assessments (GCS and FOUR scores) in the neuro ICU. Our results showed that (1) the specificity of each index or parameter in predicting a poor prognosis for comatose patients (including brain death) was very high, but the sensitivity was low. (2) The combination of at least two techniques could improve the accuracy of predicting a poor prognosis. (3) The specificity and sensitivity of predicting a good prognosis (including awakening) for comatose patients remain unsatisfactory, findings which were not significantly different from most other studies. As a result of these findings, we began to promote the application of these methods in the prediction of poor outcomes for comatose patients in China's neuro ICUs, in order to help influence medical decision-making and allocate medical resources appropriately.
Bedside resting-state EEG functional brain network information collection.
Starting in 2016, we shifted our focus to predicting the awakening of comatose patients. First, we interfaced traditional EEG technology with functional brain network concepts (coherence, phase synchronization, phase latency index, cross relation) combined with computational neuroscience. The results showed that the stronger the coherence and synchronization of the whole brain in patients who were comatose after cardiopulmonary resuscitation (CPR) and those with large hemispheric infarction (LHI), the more likely they were to regain consciousness. This differs from previous imaging-based functional brain network studies (i.e., using fMRI) in the following respects: (1) We advanced the timing of coma prediction to earlier stages of onset (within 1-2 weeks), so as to help make earlier medical decisions and provide more tailored treatment. (2) We brought coma prediction strategies straight to the bedside in the neuro ICU, so as to reduce the risks associated with transport and interruption of treatment in critically ill patients. (3) We were able to detect certain neural oscillations that were not possible with fMRI, and studied the conscious perception and information transfer between regions based on synchronization communication.
In the coming year, we will extend our work to predicting coma recovery based on resting-state EEG and functional brain networks before and after intervention. We hope that our findings can ultimately help comatose patients regain consciousness as soon as possible and return to their families and society.
Neural information processing and calculation, feature extraction and mapping of functional brain networks.
A: Patients who received CPR; B: Patients with LHI.