The unreliability of CT numbers as absolute values.

The use of CT numbers as absolute values was examined by scanning a standard phantom on five CT scanners under a variety of conditions simulating those encountered in routine body CT scanning. The results show that: (1) There are significant differences in absolute CT numbers between most scanners (only one scanner produced CT numbers that were equal to zero for water); (2) There are significant differences in absolute CT numbers between two scanners of the same manufacturer and model that were examined; (3) There is a significant difference in CT numbers in a single phantom scan, dependent on location in the scan, and the format of this variability is not constant from one scanner to another; and (4) There may be a significant difference in absolute CT numbers depending on various physical factors (e.g., kilovoltage, phantom orientation in scan aperture, and position of the phantom in the scan aperture). The findings suggest that there is a wide range of CT numbers observed for a given tissue type as a result of scanner performance alone, and that if absolute CT numbers are to be used for diagnosis the user must document that these machine-related variations are less than the differences thought to be significant.