Aim of the study was to evaluate feasibility and confounding factors of a new quantitative sonographic method based on a combination of attenuation and backscatter coefficients (UDFF: Ultrasound derived fat fraction) in correlation with the fat fraction measured by MRI (PDFF: Proton density fat fraction). *Methods and Materials Examinations were done with 2 US-machines (Sequoia and S2000: Siemens) with 5C1, DAX and 6C1 transducers with an algorithm based on a phantom corrected combination of attenuation (AC) and backscatter coefficient (BSC). Measurements of a ROI of 9 cm2 were correlated with MRI PDFF values of segmented liver (segPDFF) and a ROI in the right liver lobe (RLL) in the same position (Vida, Siemens - LiverLab®). To test the reproducibility we examined in a pilot study 20 patients with two UDFF-measurements on different days. The effect of fasting state was tested by performing UDFF before, 1, 3 and 5 hours after a standard meal. The main study consists of 165 consecutive patients (98 male, 67 female), who received a liver MRI for clinical reasons.*Results In the pilot study PDFF of the segPDFF (10.3±9.4%), was significantly higher than in the corresponding ROI (8.65±9.6%). The correlation of PDFF-measurements with scan heads (5C1, DAX, 6C1) were r=0.94 /0.91 and 0.85. Repeated measurements showed values of r=0.99/0.97 and 0.89 with best reproducibility of the DAX. Measurements before, 1, 3 and 5 hours after meal showed no significant differences between the time points (Friedmann-Test). The main study shows a significant difference in segPDFF (11.6±9.1%) and the ROI (8.9±9.5%). Although AC and BSC are not linear related to PDFF, the model with both parameters shows a significant correlation between UDFF and PDFF of the whole liver and the corresponding ROI and Voxel for all patients r=0.80/0.7 and 0.6 (6C1). Bland-Altmann shows, that main confounding factors are severe liver cirrhosis with a small RLL, patients after chemotherapy / lipiodol application, sarcoidosis and amyloidosis. Problems in MRI arose in severe Hemochromatosis, patients over 150 kg due to artifacts. *Conclusions UDFF shows a strong positive correlation with PDFF. Confounding factors could be identified hypothetically allowing for improvement in the future.*Clinical Relevance/Application UDFF could be a screening tool for early diagnosis of NAFLD and a biomarker for therapy control.

In the pilot study PDFF of the segPDFF (10.3±9.4%), was significantly higher than in the corresponding ROI (8.65±9.6%). The correlation of PDFF-measurements with scan heads (5C1, DAX, 6C1) were r=0.94 /0.91 and 0.85. Repeated measurements showed values of r=0.99/0.97 and 0.89 with best reproducibility of the DAX. Measurements before, 1, 3 and 5 hours after meal showed no significant differences between the time points (Friedmann-Test). The main study shows a significant difference in segPDFF (11.6±9.1%) and the ROI (8.9±9.5%). Although AC and BSC are not linear related to PDFF, the model with both parameters shows a significant correlation between UDFF and PDFF of the whole liver and the corresponding ROI and Voxel for all patients r=0.80/0.7 and 0.6 (6C1). Bland-Altmann shows, that main confounding factors are severe liver cirrhosis with a small RLL, patients after chemotherapy / lipiodol application, sarcoidosis and amyloidosis. Problems in MRI arose in severe Hemochromatosis, patients over 150 kg due to artifacts.

UDFF could be a screening tool for early diagnosis of NAFLD and a biomarker for therapy control.