Resonance Based Pre-processing Method for Eliminating Artifacts in Doppler Ultrasound Signals

Abstract

Transcranial Doppler ultrasound has been successfully used in the evaluation of cerebrovascular pathologies such as stroke which is a very critical condition. Special micro embolic signals, which are caused by the clots that can block blood vessels in brain, show themselves as an increase in reflected Doppler energy and can be used as the early indicator of stroke. Fourier Transform and wavelet transform, which are the classical frequency based linear decomposition methods, cannot totally separate micro-embolic signals from other high intensity signals like artifact and Doppler speckle. Therefore, in this study, a new non-linear resonance based decomposition method is proposed for eliminating artifact signals in Doppler ultrasound signals. The results indicate that with the proposed method micro-embolic signals can be successfully extracted and high intensity artifacts can be separated even if in high noise conditions.