Evaluation of cerebral blood flow after subarachnoid hemorrhage using near-field coupling and machine learning

Abstract

Background

Bedside continuous monitoring of cerebral blood flow (CBF) has significant implications in guiding individualized management and improving the prognosis of subarachnoid hemorrhage (SAH).

Objective

This study established a CBF monitoring system based on near-field coupling (NFC) measuring periodic changes in intracranial dielectric properties.

Methods

To evaluate its basic performance, a physical experiment was performed on a simulated vascular vibration model. Twenty-one healthy volunteers were recruited to perform synchronous monitoring of the cerebral oxygen (CO) and CBF before and after caffeine intake. Furthermore, eight subjects with SAH were selected for preoperative and postoperative CBF monitoring.

Results

The results demonstrate that the frequencies of simulated vascular vibration match those of the measured pulsation. The reduction of CBF and CO results in an obvious attenuation of the measured pulsation. The waveform of the measured signal collected on the SAH-affected side is distorted. This is improved to varying degrees after surgical intervention. After feature extraction and screening, two groups of feature combinations were obtained to distinguish CBF levels in different pathophysiological states. Combined with machine learning, NFC technology enables accurate diagnosis of different CBF levels.

Conclusions: This study is expected to provide a new solution for treatment effect evaluation and personalized management of SAH.