Reliability of automated detection of microembolic signals with multigated Doppler ultrasound

• Published in: Japanese Journal of Stroke Vol. 22; no. 3; pp. 395 - 402
• Main Authors: Kondo, Kimito, Yamagami, Hiroshi, Yasaka, Masahiro, Naritomi, Hiroaki, Nagatsuka, Kazuyuki
• Format: Journal Article
• Language: Japanese
• Published: The Japan Stroke Society 2000
• Subjects: antiplatelet; ischemic stroke; microembolic signals; multigated Doppler ultrasound; transcranial Doppler
• ISSN: 0912-0726; 1883-1923
• DOI: 10.3995/jstroke.22.395

Background and Purpose:Multigated Doppler ultrasound (MDU) permits the automated detection of microembolic signals (MES). Such automated MES detection is, however, influenced by various factors, such as the Doppler gain and cut-off values of the signal intensity. We evaluated the reliability of an automated MES-detecting system in healthy controls and stroke patients. Methods:In 4 stroke patients with carotid artery stenosis and 4 healthy control subjects, the middle cerebral artery flow was monitored for 30 min to detect MES with the automated system of MDU. MES was defined as unidirectional, high-intensity transient signals (HITS) with a chirp sound according to the criteria recommended by the Consensus Committee of the Ninth International Cerebral Hemodynamic Symposium. The reliability of automated MES detection was evaluated by two experienced examiners who later analyzed the Doppler spectrum recorded on an off-line system. In the stroke patients, MDS studies were repeated before and after antiplatelet administration. Results:The automated system with a cut-off signal intensity of 4 dB detected 1-308 signals in healthy controls and 47-1217 signals in stroke patients as MES. The automatically detected signals in the healthy controls all showed a signal intensity of below 10 dB. In stroke patients, the Doppler spectrum demonstrated 3 HITS with a chirp sound (MES) and 14 HITS without a chirp sound (weak HITS) prior to antiplatelet therapy. The automated system failed to detect 8 of the 17 signals. The background flow velocity in the 8 undetected signals was significantly slower than that in the 9 detected signals. Following antiplatelet therapy, the MES and weak HITS decreased in number or disappeared completely in all patients regardless of automated detection. On the other hand, automatically detected signals other than MES or weak HITS were never reduced following antiplatelet therapy. Conclusion:The data for the antiplatelet effects suggest that weak HITS reflect embolic signals like MES in spite of no association of a chirp sound. The automated system in MDU often dismisses real embolic signals or detects non-embolic signals erroneously. The false-negative or false-positive detection appears to result from various factors such as a slow background flow velocity or excessively low cut-off signal intensity. The present automated system needs to be improved further to obtain a higher accuracy.