Background and Rationale Cerebral autoregulation (CA) is the ability of the cerebrovascular system to maintain relatively constant cerebral blood flow when cerebral perfusion pressure is extremely high or low. In neurocritical patients with acute intracerebral hemorrhage, subarachnoid hemorrhage, or ischemic stroke, impaired CA is closely associated with secondary brain injury, intracranial hypertension, neurological deterioration, and poor outcomes. The pressure reactivity index (PRx), calculated as the moving Pearson correlation between invasively measured intracranial pressure (ICP) and mean arterial pressure (MAP), is the current gold standard for bedside assessment of CA. However, invasive ICP monitoring requires surgical placement of an intraventricular or intraparenchymal catheter, which carries risks of infection, hemorrhage, cerebrospinal fluid leakage, and tissue damage, and requires specialized expertise and intensive care resources, limiting widespread use.
Electrical impedance tomography (EIT) is a non-invasive, radiation-free, real-time biomedical imaging technique that measures regional changes in bioimpedance, which reflect variations in cerebral blood volume and tissue perfusion. A custom-built cerebral EIT system (EC100-PRO) has been developed by the research team. It uses 16 scalp electrodes placed in a ring configuration to continuously acquire impedance signals. Through standardized signal processing and image reconstruction algorithms, EIT-derived parameters that track cardiac-synchronous cerebral blood volume pulsations are generated. Preclinical and pilot clinical data indicate that these EIT-derived parameters correlate with changes in cerebral perfusion and autoregulatory status. This study aims to validate the agreement between EIT-derived parameters and the reference standard PRx, thereby establishing a non-invasive, continuous alternative for CA monitoring in neurocritical care.
Study Design and Technical Framework This is a single-center, prospective, observational, self-controlled clinical study conducted in the Neurosurgical Intensive Care Unit (NICU) of Xijing Hospital, The First Affiliated Hospital of Air Force Military Medical University. The study is observational: EIT monitoring serves only as an additional non-invasive signal acquisition and does not interfere with, terminate, or affect the patient's original clinical diagnosis and treatment plan. It is self-controlled because invasive ICP/ABP data and non-invasive EIT data are collected synchronously from the same patient, using the patient's own invasive standard (PRx) to verify the accuracy of the non-invasive index.
All participants receive standard clinical care according to current international neurocritical care guidelines (including blood pressure management per the 2024 Chinese Expert Consensus on Hypertension Management in Stroke Patients, sedation, analgesia, and ICP-directed therapy). No experimental drugs, additional surgical procedures, or treatment modifications are applied for research purposes. EIT monitoring is add-on only and does not interrupt routine care.
Continuous synchronous multimodal data acquisition is performed during the participant's NICU stay. Invasive MAP (via femoral arterial catheter) and ICP (via intraventricular catheter) are measured as standard of care. The EIT system applies a 1 mA, 50 kHz alternating current through 16 Ag/AgCl electrodes placed equidistantly around the scalp at the level of the upper border of the eyebrow arch to the external occipital protuberance. Electrode-skin impedance is kept below 5 kΩ. EIT data are acquired at 40 frames per second.
Signal preprocessing includes low-pass filtering at 0.05 Hz to remove respiratory and high-frequency artifacts and resampling to 10-second averages. The reference index PRx is calculated as the Pearson correlation between MAP and ICP within a 300-second sliding window (step 60 seconds). From the EIT image sequence, a parameter reflecting cardiac-synchronous cerebral blood volume pulsation is extracted. Within the same 300-second sliding window, the correlation between MAP and this EIT-derived parameter (called the EIT-CA index) is computed. Data segments with electrode contact impedance >5 kΩ or motion artifacts exceeding 10% of the recording duration are excluded. At least 2 hours of continuous valid data per participant are required for analysis.
Study Design and Technical Framework This is a single-center, prospective, observational, self-controlled clinical study conducted in the Neurosurgical Intensive Care Unit (NICU) of Xijing Hospital, The First Affiliated Hospital of Air Force Military Medical University. The study is observational: EIT monitoring serves only as an additional non-invasive signal acquisition and does not interfere with, terminate, or affect the patient's original clinical diagnosis and treatment plan. It is self-controlled because invasive ICP/ABP data and non-invasive EIT data are collected synchronously from the same patient, using the patient's own invasive standard (PRx) to verify the accuracy of the non-invasive index.
All participants receive standard clinical care according to current international neurocritical care guidelines (including blood pressure management per the 2024 Chinese Expert Consensus on Hypertension Management in Stroke Patients, sedation, analgesia, and ICP-directed therapy). No experimental drugs, additional surgical procedures, or treatment modifications are applied for research purposes. EIT monitoring is add-on only and does not interrupt routine care.
Continuous synchronous multimodal data acquisition is performed during the participant's NICU stay. Invasive MAP (via femoral arterial catheter) and ICP (via intraventricular catheter) are measured as standard of care. The EIT system applies a 1 mA, 50 kHz alternating current through 16 Ag/AgCl electrodes placed equidistantly around the scalp at the level of the upper border of the eyebrow arch to the external occipital protuberance. Electrode-skin impedance is kept below 5 kΩ. EIT data are acquired at 40 frames per second.
Signal preprocessing includes low-pass filtering at 0.05 Hz to remove respiratory and high-frequency artifacts and resampling to 10-second averages. The reference index PRx is calculated as the Pearson correlation between MAP and ICP within a 300-second sliding window (step 60 seconds). From the EIT image sequence, a parameter reflecting cardiac-synchronous cerebral blood volume pulsation is extracted. Within the same 300-second sliding window, the correlation between MAP and this EIT-derived parameter (called the EIT-CA index) is computed. Data segments with electrode contact impedance >5 kΩ or motion artifacts exceeding 10% of the recording duration are excluded. At least 2 hours of continuous valid data per participant are required for analysis.
