Tissue characterization using intelligent adaptive filter in the diagnosis of diffuse and focal liver disease

Abstract

In ultrasonic imaging an adaptive two-dimensional filter (ATDF) can suppress randomly generated speckle using the ratio of the local variance to the local mean as the speckle recognition feature (R). The degree of smoothing depends on the difference between the recognition feature in the region to be filtered and the selected reference tissue. We have investigated the clinical application of ATDF for ultrasound B-mode images of liver abnormalities. Using the R values of normal liver as reference values, the ATDF images were displayed. Normal livers (n = 17, R = 2.19 +/- 0.14 M +/- SEM), fatty livers (N = 16, R = 1.89 +/- 0.15) and those with acute hepatitis (N = 10, R = 2.25 +/- 0.18) appeared smooth after application of the adaptive filter, but those diseases with higher R values, such as chronic hepatitis (N = 10, R = 3.04 +/- 0.30), cirrhosis (n = 16, R = 4.44 +/- 0.30), metastases (N = 16, R = 6.43 +/- 0.53) and hepatocellular carcinomas (N = 8, R = 7.92 +/- 0.85), were largely unsmoothed. In conclusion, ATDF allows differentiation of some forms of liver disease and may be helpful in the detection of microfocal echogenic textural lesions.