Diagnosis of fatty infiltration of the liver on contrast enhanced CT: limitations of liver-minus-spleen attenuation difference measurements

Background: We investigated whether liver-minus-spleen (L-S) attenuation differences can accurately diagnose fatty infiltration of the liver on contrast-enhanced computed tomography (CT). Methods: A group of 78 patients administered a fast injection (90-s duration) of 150 mL 60% ionic contrast was compared with 81 patients given a slow injection (152.5 s). The presence or absence of fatty infiltration of the liver was diagnosed by noncontrast CT. Results: The L-S attenuation differences varied significantly, depending on both injection rate and timing of measurements. For the fast-injection group, the optimal L-S threshold for diagnosing fatty infiltration ranged from -43 to -33 Hounsfield units (HU) for early (79 s) and late measurements (106 s), respectively. For the slow-injection group, the optimal threshold ranged from -31 to -25 HU (80 and 112 s, respectively). In addition, sensitivity was not very high (range = 0.54-0.71) for either injection protocol at any measurement time because of significant overlap of L-S values between normal and fatty infiltration patients. Moderate and severe fatty infiltration were more reliably diagnosed than mild fatty infiltration by this method. Conclusions: Contrast injection rate and timing of measurements significantly influence the optimal L-S threshold for diagnosing fatty liver. This limits the clinical usefulness of such measurements.