Retrospective quantitative assessment of liver size by measurement of radiographic liver area in small-breed dogs

OBJECTIVE To determine the feasibility of radiographic measurement of liver area in small-breed dogs and to assess correlations between CT liver volume measurements (reference standard) and radiographic liver size measurements. ANIMALS 107 small-breed dogs (body weight, <= 10 kg) that had previously undergone orthogonal thoracic and abdominal radiography and abdominal CT. PROCEDURES In a retrospective study design, dogs were allocated to groups (normal liver [n = 36], microhepatia [34], and hepatomegaly [37]) on the basis of radiographic liver size and clinicopathologic findings. Radiographic liver area (RLA) was automatically calculated from archived radiographic images by free-hand outlining of the liver margins by use of DICOM viewer software, and other standard radiographic measurements were performed. Liver volume was measured on CT images. Intraoperator repeatability of RLA and CT measurements was assessed (duplicate measurements 2 weeks apart). To control for various breed conformations, radiographic values were normalized to body weight and T11 area. RESULTS Mean +/- SD ratios of RLA to T11 area and RLA to body weight for dogs with normal livers were 32.7 +/- 6.2 and 7.0 +/- 1.4, respectively. Excellent intraobserver agreement was observed in RLA measurements within groups (intraclass correlation coefficients, 0.861 to 0.989), and RLA measurements had the highest correlation with CT liver volume measurements (r = 0.94) of all radiographic measurements. CONCLUSIONS AND CLINICAL RELEVANCE Findings indicated that RLA measurement in small-breed dogs with or without liver disease was useful and accurate for estimation of liver size, compared with CT measurement, and might be particularly useful for monitoring of changes in liver size.