SONOGRAPHIC TISSUE CHARACTERIZATION OF BRIGHT LIVER

Abstract

Computerized ultrasonoh'Tl!Phic tissue characterization has become obj :diw mt: ns fur di gnusis of livt:r dist: ses. It is difficult lu differentiate diffuse liver diseases by visual inspection from the ultrasound images. Computerized ultrasonographic tissue characterization has emerged as one of the must accurate methods in classification of different liver pathologies. Using 10 tissue char cterizatiun pammeters namely; mean gray level, 9u. percentile, contrast, entropy, attenuation, correlation, angular second moment, average specular intensity, average diffuse intensity and speckle separation distance was able to differentiate between different hepatic pathologies among 45 patients with sunographically detected bright liver am!• 12 control subjects. We concluded that, Computer-assisted sonographic tissue characterization using the neural network classifier is a sensitive and specific technique for sepamting diffuse liver disease namely; cases with uitrdsonically detected bright liver from control cases.