Adaptive noise reduction process based on gray-level concentrative filter for CW Doppler spectrogram

An adaptive noise reduction process based on a gray-level concentrative filter is proposed for CW Doppler spectrogram. A gray-level concentrative filter and a convolution filter were tested to separate cardiac Doppler signals from the background noise in the histogram of the spectral amplitude as a function of time. By setting the cut-off threshold to the peak of the histogram of background noise amplitude after filtering, the background noise was significantly reduced without deteriorating the spectral signal. By setting the threshold to the valley between the histogram peaks of the background noise and spectral amplitudes after filtering, the maximum velocity was automatically traced in the spectrum as a function of time with a simple algorithm. A gray-level concentrative filter always showed a slightly better performance than a convolution filter with the same sliding window size in histogramically separating the Doppler signals from the background noise.