TY - JOUR

T1 - Assessment of Liver Fat

T2 - Dual-Energy CT versus Conventional CT with and without Contrast

AU - Xu, Jack Junchi

AU - Boesen, Mikkel Ranum

AU - Hansen, Sofie Lindskov

AU - Ulriksen, Peter Sommer

AU - Holm, Soren

AU - Lonn, Lars

AU - Hansen, Kristoffer Lindskov

PY - 2022

Y1 - 2022

N2 - We assessed the correlation between liver fat percentage using dual-energy CT (DECT) and Hounsfield unit (HU) measurements in contrast and non-contrast CT. This study included 177 patients in two patient groups: Group A (n = 125) underwent whole body non-contrast DECT and group B (n = 52) had a multiphasic DECT including a conventional non-contrast CT. Three regions of interest were placed on each image series, one in the left liver lobe and two in the right to measure Hounsfield Units (HU) as well as liver fat percentage. Linear regression analysis was performed for each group as well as combined. Receiver operating characteristic (ROC) curve was generated to establish the optimal fat percentage threshold value in DECT for predicting a non-contrast threshold of 40 HU correlating to moderate-severe liver steatosis. We found a strong correlation between fat percentage found with DECT and HU measured in non-contrast CT in group A and B individually (R-2 = 0.81 and 0.86, respectively) as well as combined (R-2 = 0.85). No significant difference was found when comparing venous and arterial phase DECT fat percentage measurements in group B (p = 0.67). A threshold of 10% liver fat found with DECT had 95% sensitivity and 95% specificity for the prediction of a 40 HU threshold using non-contrast CT. In conclusion, liver fat quantification using DECT shows high correlation with HU measurements independent of scan phase.

AB - We assessed the correlation between liver fat percentage using dual-energy CT (DECT) and Hounsfield unit (HU) measurements in contrast and non-contrast CT. This study included 177 patients in two patient groups: Group A (n = 125) underwent whole body non-contrast DECT and group B (n = 52) had a multiphasic DECT including a conventional non-contrast CT. Three regions of interest were placed on each image series, one in the left liver lobe and two in the right to measure Hounsfield Units (HU) as well as liver fat percentage. Linear regression analysis was performed for each group as well as combined. Receiver operating characteristic (ROC) curve was generated to establish the optimal fat percentage threshold value in DECT for predicting a non-contrast threshold of 40 HU correlating to moderate-severe liver steatosis. We found a strong correlation between fat percentage found with DECT and HU measured in non-contrast CT in group A and B individually (R-2 = 0.81 and 0.86, respectively) as well as combined (R-2 = 0.85). No significant difference was found when comparing venous and arterial phase DECT fat percentage measurements in group B (p = 0.67). A threshold of 10% liver fat found with DECT had 95% sensitivity and 95% specificity for the prediction of a 40 HU threshold using non-contrast CT. In conclusion, liver fat quantification using DECT shows high correlation with HU measurements independent of scan phase.

KW - liver fat quantification

KW - dual-energy CT

KW - spectral CT

KW - HEPATIC STEATOSIS

KW - COMPUTED-TOMOGRAPHY

KW - QUANTIFICATION

KW - FRACTION

KW - DISEASE

KW - ADULTS

KW - DECOMPOSITION

KW - DIAGNOSIS

U2 - 10.3390/diagnostics12030708

DO - 10.3390/diagnostics12030708

M3 - Journal article

C2 - 35328261

VL - 12

JO - Diagnostics

JF - Diagnostics

SN - 2075-4418

IS - 3

M1 - 708

ER -