Filtering echocardiographic image sequences in frequency domain

This work describes a new method for filtering echocardiographic image sequences, based on the assumption of an almost periodic motion of the cardiac wall. Exploiting the knowledge of the regular motion of the heart can be advantageous with respect to classical filtering methods where no assumption is made on the image dynamics. From the ideal model of a pulsing ring, a mathematical description of the corresponding gray level pattern in time has been derived, and a peculiar feature in the frequency domain has been observed. This feature has been used as the discriminating function in enhancing pixels with higher probability of lying on the ventricular wall trajectory.