The effect of blood pressure calibrations and transcranial Doppler signal loss on transfer function estimates of cerebral autoregulation

• Published in: Medical engineering & physics Vol. 33; no. 5; pp. 553 - 562
• Main Authors: Deegan, Brian M., Serrador, Jorge M., Nakagawa, Kazuma, Jones, Edward, Sorond, Farzaneh A., ÓLaighin, Gearóid
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.06.2011; Elsevier
• Subjects: Biological and medical sciences; Blood Pressure; Brain - blood supply; Brain - physiology; Calibration; Cardiovascular system; Cerebral autoregulation; Cerebral blood flow; Data loss; Female; Humans; Investigative techniques, diagnostic techniques (general aspects); Linear Models; Male; Medical sciences; Middle Aged; Physiocal; Radiology; Signal Processing, Computer-Assisted; Transfer function; Ultrasonic investigative techniques; Ultrasonography, Doppler, Transcranial; Physiocal; Cerebral autoregulation; Blood pressure; Transfer function; Cerebral blood flow; Data loss; Sonography; Doppler ultrasound study; Central nervous system; Calibration; Blood flow; Encephalon; Transcranial route; Self regulation; Arterial pressure; Hemodynamics; Biomedical engineering
• ISSN: 1350-4533; 1873-4030; 1873-4030
• DOI: 10.1016/j.medengphy.2010.12.007

There are methodological concerns with combined use of transcranial Doppler (TCD) and Finapres to measure dynamic cerebral autoregulation. The Finapres calibration mechanism (“physiocal”) causes interruptions to blood pressure recordings. Also, TCD is subject to signal loss due to probe movement. We assessed the effects of “physiocals” and TCD signal loss on transfer function estimates in recordings of 45 healthy subjects. We added artificial “physiocals” and removed sections of TCD signal from 5 min Finapres and TCD recordings. We also compared transfer function results from 5 min time series with time series as short as 1 min. Accurate transfer function estimates can be achieved in the 0.03–0.07 Hz band using beat-by-beat data with linear interpolation, while data loss is less than 10 s. At frequencies between 0.07 and 0.5 Hz, transfer function estimates become unreliable with 5 s of data loss every 50 s. 2 s data loss only affects frequency bands above 0.15 Hz. Finally, accurate transfer function assessment of autoregulatory function can be achieved from time series as short as 1 min, although gain and coherence tend to be overestimated at higher frequencies.