Continuous EEG Monitoring for Early Detection of Delayed Cerebral Ischemia in Subarachnoid Hemorrhage: A Pilot Study

Abstract

Introduction

Early identification of delayed cerebral ischemia (DCI) in patients with aneurysmal subarachnoid hemorrhage (aSAH) is a major challenge. The aim of this study was to investigate whether quantitative EEG (qEEG) features can detect DCI prior to clinical or radiographic findings.

Methods

A prospective cohort study was performed in aSAH patients in whom continuous EEG (cEEG) was recorded. We studied 12 qEEG features. We compared the time point at which qEEG changed with the time point that clinical deterioration occurred or new ischemia was noted on CT scan.

Results

Twenty aSAH patients were included of whom 11 developed DCI. The alpha/delta ratio (ADR) was the most promising feature that showed a significant difference in change over time in the DCI group (median −62 % with IQR −87 to −39 %) compared to the control group (median +27 % with IQR −32 to +104 %, p = 0.013). Based on the ROC curve, a threshold was chosen for a combined measure of ADR and alpha variability (AUC: 91.7, 95 % CI 74.2–100). The median time that elapsed between change of qEEG and clinical DCI diagnosis was seven hours (IQR −11–25). Delay between qEEG and CT scan changes was 44 h (median, IQR 14–117).

Conclusion

In this study, ADR and alpha variability could detect DCI development before ischemic changes on CT scan was apparent and before clinical deterioration was noted. Implementation of cEEG in aSAH patients can probably improve early detection of DCI.

Appendix: Used Equations

$$ {\text{Delta}} = \frac{{\sum\limits_{f = 1}^{4} {S(f)} }}{{\sum\limits_{f = 1}^{30} {S(f)} }} $$

$$ {\text{Theta}} = \frac{{\sum\limits_{f = 4}^{8} {S(f)} }}{{\sum\limits_{f = 1}^{30} {S(f)} }} $$

$$ {\text{Alpha}} = \frac{{\sum\limits_{f = 8}^{12.5} {S(f)} }}{{\sum\limits_{f = 1}^{30} {S(f)} }} $$

$$ {\text{Beta}} = \frac{{\sum\limits_{f = 12.5}^{30} {S(f)} }}{{\sum\limits_{f = 1}^{30} {S(f)} }} $$

$$ {\text{ADR}} = \frac{\text{Alpha}}{\text{Delta}} $$

$$ {\text{ABDTR}} = \frac{{{\text{Alpha}} + {\text{Beta}}}}{{{\text{Delta}} + {\text{Theta}}}} $$

$$ \sum\limits_{i = 1}^{{\text{SEF}}75} {S({f_i}) = \sum\limits_{j = 1}^{30} {S({f_j}) \times 0.75}} $$

$$ \sum\limits_{i = 1}^{{\text{SEF}}90} {S({f_i}) = \sum\limits_{j = 1}^{30} {S({f_j}) \times 0.9}} $$

$$ {\text{Alpha}} - {\text{Var}} = \frac{{\sigma_{\text{alpha}} }}{\text{Alpha}}\quad {\text{with}}\quad \sigma_{\text{alpha}} = \sqrt {\frac{1}{n - 1}} \sum\limits_{i = 1}^{n} {({\text{Alpha}}_{i} - \overline{\text{Alpha}} )^{2} } $$

with n the number of available data points

$$ {\text{BSI}}_{\text{Slow}} = \frac{1}{M}\sum\limits_{i = 1}^{M} {\left| {\sum\limits_{j = 1}^{N} {\frac{{S_{{ij,{\text{left}}}} (f_{j} ) - S_{{ij,{\text{right}}}} (f_{j} )}}{{S_{{ij,{\text{left}}}} (f_{j} ) + S_{{ij,{\text{right}}}} (f_{j} )}}} } \right|} \quad {\text{with}}\quad 1 < f_{j} < 7 $$

with M the number of channel pairs and N the number of Fourier coefficients

$$ {\text{BSI}}_{\text{Slow}} = \frac{1}{M}\sum\limits_{i = 1}^{M} {\left| {\sum\limits_{j = 1}^{N} {\frac{{S_{{ij,{\text{lef}}t}} (f_{j} ) - S_{{ij,{\text{right}}}} (f_{j} )}}{{S_{{ij,{\text{lef}}t}} (f_{j} ) + S_{{ij,{\text{right}}}} (f_{j} )}}} } \right|} \quad {\text{with}}\quad 1 < f_{j} < 25 $$

with M the number of channel pairs and N the number of Fourier coefficients.