Noise reduction in ultrasonic images by digital filtering.

Ultrasonic images exhibit a characteristic granular pattern, called speckle, which bears little resemblance to the actual acoustical tissue microstructure (Burckhardt, 1978). This phenomenon is due to the phase-sensitive addition, by the transducer, of echoes scattered by tissue inhomogeneities of size and spacing too fine to be resolved (Morrison et al, 1980). In the literature, speckle has been treated as a distracting artefact (Wells & Halliwell, 1981) which limits the detection of lowcontrast lesions (Smith & Lopez, 1982). The mean greyscale level of speckle provides information about the tissue echogenicity and is a feature one wants to preserve. However, the spatial variation of speckle (texture) characterises the imaging system as well as the tissue itself. Factors such as grey-scale mapping of the scanner, transducer focusing pattern and relative position, or even the intervening tissue have a significant effect on image texture (Jaffe & Harris, 1980; Kimme-Smith & Jones, 1984) so that it becomes ...