Quantitative estimation of attenuation in ultrasound video images: correlation with histology in diffuse liver disease.
RATIONALE AND OBJECTIVES: To determine the relationship between the attenuation of backscatter intensity in B-scan images of the liver and diffuse liver disease in order to assess the usefulness of this method in providing quantitative objective characterization of diffuse liver diseases in general and in fatty liver in particular. METHODS: Twenty-four healthy volunteers and 28 patients with elevated liver enzyme levels who underwent liver biopsy were included in this study. An automatic far-field slope (FFS) algorithm that estimates the decrease in amplitude of the backscattered echo as a function of beam depth was implemented on the noncompensated image that was acquired on a commercial phased-array ultrasound system fitted to a custom-built interface card. The images were processed at a work-station. All scans were acquired repeatedly, read, and graded blindly by experienced ultrasound radiologists. Histology obtained via needle biopsy was reviewed without knowledge of the ultrasound findings. RESULTS: Analysis of the FFS data for fatty infiltration in all patient groups yielded a sensitivity of 67%, a specificity of 77%, a positive predictive value (PPV) of 77%, negative predictive value (NPV) of 67%, and an accuracy of 71%. The mean score of the ultrasound reviewers showed a sensitivity of 82%, a specificity of 66%, a PPV% of 68%, an NPV of 81%, and an accuracy of 72%. Normal FFS values (false-negative) were found in five patients with proved fatty infiltration. All of these patients had coexistent moderate to severe hepatic inflammation. However, FFS data in patients with uncomplicated (pure) fatty infiltration revealed a sensitivity of 100%, a specificity of 80%, a PPV of 89%, an NPV of 100%, and an accuracy of 92%. The best ultrasound score yielded a sensitivity of 100%, a specificity of 60%, a PPV of 80%, an NPV of 100%, and an accuracy of 85% in the same patients. CONCLUSIONS: The data demonstrate an excellent sensitivity (100%) of the FFS values in patients with uncomplicated fatty infiltration. This was also the only group of patients in whom the FFS score was superior to the radiologists' best score. The FFS method can be used as a tool to follow up the response to a clinical or research treatment and to obtain standardization of pattern interpretation independently of the individual reader.