Multiple wavelet denoising for embolic signal enhancement

Transcranial Doppler ultrasound can be used to detect circulating cerebral emboli. Embolic signals have characteristic transient chirps suitable for wavelet analysis. We have implemented and evaluated the first online selective wavelet transient enhancement filter to amplify embolic signals in a preprocessing system. Our approach is similar to wavelet de-noising for signal enhancement, but, in order to retain blood flow information, we do not use traditional threshold methods. The selective wavelet amplifier uses the matched filter properties of wavelets to enhance embolic signals significantly and improve classification performance using a novel noise tolerant approach. Even the smallest embolic signals are enhanced. We show an increase of over 2dB (on average) in embolic signal strength and a significant improvement in detection accuracy when our filter is applied both to a commercially available detection system and an in house frequency based detection system. The implementation of the filter is simplified by using an optimized matrix form. The block matrix form is significantly faster than the normal recursive discrete wavelet transform implementation.