A micro-embolic energy detector based on sub-band decomposition

Cerebrovascular Accidents can be caused by cerebral emboli blocking brain blood vessels. Analysis of transcranial Doppler signals practically aids the detection of emboli. Signal processing methods have been proposed for emboli detection. In this study, we introduce a new micro-embolic energy detector composed of N detectors associated to N Doppler frequency sub-bands. To test our detectors, we propose a training phase during which we adjust the optimal number of sub-bands and detection thresholds and a testing phase through which we measure performances. Using real signals, we show that in terms of the number of sub-bands, 4 sub-bands provide the highest detection rate and lowest false alarm. Moreover, compared to standard detectors, the 4 sub-band energy detector reduces the false alarm rate from 44% to 36%, increases the detection rate from 66% to 79% and augments the Embolus to Blood Ratio from 24 dB to 40 dB. This new energy detector permits detecting smallest micro-emboli, precursors of coming large emboli with high stroke risks.