OBJECTIVE: To identify correlations between ultrasonographic findings and specific hepatic diseases in cats. DESIGN: Retrospective study. SAMPLE POPULATION: Medical records of 72 cats with a histopathologic diagnosis of hepatic disease and diagnostic-quality abdominal ultrasonograms between 1985 and 1997. PROCEDURE: Abdominal ultrasonographic findings in 72 cats with histologically confirmed hepatic disease (hepatic lipidosis excluded) were reviewed. Rather than attempt to combine individual ultrasonographic findings with specific hepatic diseases, 2 classification trees were created as models to correlate certain groups of abnormalities with specific hepatic diseases or with malignant and benign lesions of the liver. Sensitivity and specificity of classification trees were calculated. RESULTS: Use of a classification tree resulted in correct classification of malignant versus benign hepatic lesions in 88.9% of cats that had hepatic disease (sensitivity, 90.7%; specificity, 86.1%). Use of a classification tree to distinguish individual types of hepatic diseases resulted in mostly accurate classification of hepatic lymphosarcoma (sensitivity, 70.5%; specificity, 98.2%), cholangitis-cholangiohepatitis syndrome (sensitivity, 87%; specificity, 90%), and benign lesions of the liver (sensitivity, 84.6%; specificity, 86.4%). Criteria that helped most in differentiating among various hepatic diseases were abnormalities within other organs (spleen, lymph nodes) and appearance of the hepatic portal system. A correlation was not found between focal or multifocal appearance of hepatic lesions and specific hepatic diseases. CLINICAL IMPLICATIONS: Use of classification trees to distinguish among specific hepatic diseases or between malignant and benign hepatic lesions provides potentially useful algorithms for ultrasonographic evaluation of cats with hepatic disease.
OBJECTIVE: To identify correlations between ultrasonographic findings and specific hepatic diseases in cats. DESIGN: Retrospective study. SAMPLE POPULATION: Medical records of 72 cats with a histopathologic diagnosis of hepatic disease and diagnostic-quality abdominal ultrasonograms between 1985 and 1997. PROCEDURE: Abdominal ultrasonographic findings in 72 cats with histologically confirmed hepatic disease (hepatic lipidosis excluded) were reviewed. Rather than attempt to combine individual ultrasonographic findings with specific hepatic diseases, 2 classification trees were created as models to correlate certain groups of abnormalities with specific hepatic diseases or with malignant and benign lesions of the liver. Sensitivity and specificity of classification trees were calculated. RESULTS: Use of a classification tree resulted in correct classification of malignant versus benign hepatic lesions in 88.9% of cats that had hepatic disease (sensitivity, 90.7%; specificity, 86.1%). Use of a classification tree to distinguish individual types of hepatic diseases resulted in mostly accurate classification of hepatic lymphosarcoma (sensitivity, 70.5%; specificity, 98.2%), cholangitis-cholangiohepatitis syndrome (sensitivity, 87%; specificity, 90%), and benign lesions of the liver (sensitivity, 84.6%; specificity, 86.4%). Criteria that helped most in differentiating among various hepatic diseases were abnormalities within other organs (spleen, lymph nodes) and appearance of the hepatic portal system. A correlation was not found between focal or multifocal appearance of hepatic lesions and specific hepatic diseases. CLINICAL IMPLICATIONS: Use of classification trees to distinguish among specific hepatic diseases or between malignant and benign hepatic lesions provides potentially useful algorithms for ultrasonographic evaluation of cats with hepatic disease.