A Quantitative Electroencephalographic Index for Stroke Detection in Adults

Area of Science:

• Neurology
• Neurophysiology

Background:

• Electroencephalography (EEG) is underutilized in stroke characterization.
• Quantitative EEG metrics can aid in stroke assessment.

Purpose of the Study:

• To evaluate the EEG Correlate Of Injury to the Nervous system (COIN) index's ability to differentiate large from small acute ischemic strokes in adults.
• To assess COIN's performance in stroke size discrimination using continuous EEG monitoring.

Main Methods:

• Retrospective analysis of 35 adult patients with acute ischemic stroke and continuous EEG monitoring.
• Stroke size categorized as large (>100 mL) or small using ABC/2 method.
• COIN index calculated from EEG data; logistic regression and random forest classification used for analysis.

Main Results:

• Patients with large strokes exhibited significantly higher COIN values compared to those with small strokes (P = 0.0001).
• Logistic regression achieved 83% accuracy, 70% sensitivity, and 88% specificity in classifying stroke size.
• Random forest classification demonstrated robust performance with limited EEG data (5-30 minutes).

Conclusions:

• The COIN index successfully differentiated large from small acute ischemic strokes in this adult cohort.
• Further prospective, multicenter studies are needed to validate COIN's utility for bedside stroke detection, especially when neuroimaging is inaccessible or the neurological exam is limited.