Impact of Clinical Environment on Embolus Detection: A Comparison of Automated and Manual Detection of Doppler Embolic Signals

Abstract

Background: Transcranial Doppler ultrasound detection of weak embolic signals is inhibited by intrinsic limitations within the human auditory system. Psychoacoustics effects are likely to be exacerbated in a clinical environment, where automated embolus detection has potential to surpass manual detection. In this study we quantify the impact of clinical environment on manual detection of Doppler embolic signals following carotid surgery. We also discuss the implications of psychoacoustics considerations for the evaluation of automated detection systems. Method: Concurrent monitoring by vascular technologists and an automated embolus detection system were performed for 50 consecutive patients during postoperative recovery. Both detection methods were evaluated against a majority decision human expert panel analyzing under ideal conditions. Results: Clinical environment reduced the overall sensitivity of manual monitoring by ∼23%, mainly due to a ∼2-dB increase in the lower threshold for detection. Clinical environment was also associated with a reduction in positive predictive value for manual detection of ∼9% compared to ideal conditions. Automated monitoring, which is not affected by environment, was marginally more sensitive for detection of weaker embolic signals. Conclusions: One in 4 weak embolic signals was missed during routine clinical monitoring compared to ideal conditions. Automated detection (in this study) performed slightly better than human observers but did not approach the performance of the majority decision panel.