Using high-frequency oscillation to detect acute brain injury in extracorporeal membrane oxygenation supported patients.

OBJECTIVE

This study aims to utilize HFO analysis to enhance existing SSEP modality and develop it as a bedside diagnostic tool for acute brain injury (ABI) detection in Extracorporeal Membrane Oxygenation (ECMO) patients.

SIGNIFICANCE: Timely diagnosis of ABI in ECMO patients is challenging due to logistical complexities with computed tomography (CT) and magnetic resonance imaging (MRI). Integrating time-frequency analysis into routine SSEP monitoring for early ABI detection can facilitate timely medical decisions.

METHOD: Consecutive SSEP data were collected from Johns Hopkins Intensive Care Units (ICUs), including 31 ECMO and 45 non-ECMO patients from 2016 to 2022. ABIs were determined using CT and MRI as clinically indicated. Using wavelet techniques, two SSEP-HFO components were quantified: HFOL (80-200 Hz) and HFOH (200-600 Hz), which were later fed to a Support Vector Machine (SVM) with a linear kernel.

RESULT: ECMO patients with ABI (N = 22) exhibited suppressed HFOH (Median = -9.09, Interquartile Range (IQR) = [ -13.5; -4.73] dB) compared to patients without (N = 9, Median = -4.39, IQR = [-6.35; -3.28] dB, P = 0.035). The SVM classifier achieved an accuracy of 75 % and a sensitivity of 82 % for detecting ABI, outperforming SSEP-N20.

CONCLUSION

SSEP-HFO can potentially improve early detection of ABI in ECMO patients at the bedside.