Fully Automatic Segmentation and Three-Dimensional Reconstruction of the Liver in CT Images

Simple diameter measurement as predictor of liver volume and liver parenchymal disease

To investigate the accuracy of liver diameters for estimation of liver size and to evaluate their application as tool for assessment of parenchymal liver disease. In the course of a population-based study, (SHIP) one thousand nine hundred thirty-nine volunteers underwent magnetic resonance imaging (MRI) of the liver including 3D gradient echo MRI sequences. Maximum liver diameters were measured in cranio-caudal (CC), anterior–posterior (AP), medial–lateral (ML) orientation. Diameters were compared with true liver volume assessed by liver segmentation. Additionally, age-dependent reference values for diameters were defined. Finally, accuracy of liver diameters was assessed to discriminate volunteers with healthy livers and participants with parenchymal changes, measured by MRI and laboratory. Reference values of liver diameters within the healthy population (n = 886) were defined as follows (mean ± standard deviation, confidence interval CI in cm): CC 17.2 ± 2, CI 13.6/21.2; AP 15.8 ± 1.9, CI 12.6/19.8; ML 19.7 ± 2.3, CI 15.8/24.6. There was a poor correlation using linear regression between liver diameter and true liver volume; CC 0.393, AP 0.359; ML 0.137. The AP direction shows the best correlation to discriminate between healthy and pathologic liver changes; AUC 0.78; p < 0.001, CC AUC 0.53; p < 0.001 and ML AUC 0.52; p = 0.008. Measurement of liver diameter, especially in the anterior–posterior direction is a simple option to detect chronic liver disease but less suitable for prediction of liver volume.