Reproducibility and observer variation at computed tomography and ultrasound of the normal pancreas

In order to evaluate the reproducibility of computed tomography (CT) and ultrasound (US), 13 patients with a normal pancreas were examined twice with the two techniques by the same observers. The results of the measurements of the pancreas were reproducible using both CT and US. In estimating the intraobserver variation at CT in 20 patients, the location and spread of the measurements of the diameters of the pancreatic head were reproducible, but the diameters of the pancreatic body and tail were significantly larger at the second observation. Further, the differences in the measurements of the size and dispersion of the pancreatic diameters at CT between two observers were significant, apart from the diameter of the pancreatic tail, where no significant difference was noted. The differences between two observers were greater than for the same. Using US, the results may indicate that the measurements were even less reproducible than using CT, both for the same observer and between two observers. However, all the measured diameters within the 95 per cent confidence intervals, were within the normal range. Consequently, even if most measurements differed significantly, these differences may probably not lead to any false conclusions concerning the limits of the normal size of the pancreas.