Micro-embolic signal characterization based on long time Doppler time-frequency image processing and statistics

Micro-embolic signal (MES) can be an important indicator of stroke risk. Transcranial Doppler technique provides a non-invasive way to monitor the dynamic mechanisms of the emboli. Existing detection methods are based on signal processing viewpoint, to detect high intensity transient signals (HITS) and to differentiate of MES and artefact. This paper presents a method to detect HITS based on Doppler time-frequency image processing. Several characteristics of emboli and various artefacts are also established, based on statistical results from 150 hours of recordings on 97 patients, to differentiate emboli and artefacts. Statistical results show that emboli can be effectively separated from artefacts and be detected, using criteria of the velocity and HITS width-length ratio.