Quantitative Assessment of Fatty Liver using Ultrasound with Normalized Local Variance Technique

Purpose To assess the diagnostic performance of the normalized local variance (NLV) ultrasound technique in the detection of the fatty liver using histopathology as a reference standard. Materials and Methods We prospectively enrolled 194 consecutive patients with clinical suspicion of diffuse liver disease or history of liver transplantation. Conventional grayscale ultrasound and NLV examinations were performed and immediately followed by liver biopsies. The degrees of fatty liver, necroinflammatory activity, and fibrosis stage were evaluated by histopathological assessment. The diagnostic performance of the NLV values in detecting each grade of fatty liver was determined using receiver operating characteristics analyses, and multivariate linear regression analyses were performed to identify variables significantly associated with the NLV values. Results The number of patients in each degree of fatty liver and hepatic fibrosis was 118/ 37/26/13 and 81/68/24/ 6/14 for none/mild/moderate/severe steatosis and F0 / F1/F2 / F3/ F4 fibrosis on histopathological examinations, respectively. The area under the receiver operating characteristics curve and optimal cut-off NLV value for detecting fatty liver of varying degrees were 0.911 and 1.095 for = S1, 0.974 and 1.055 for = S2, and 0.954 and 1.025 for = S3, respectively. Multivariate analyses revealed that not fibrosis or inflammation but rather the degree of steatosis was associated with the NLV value. Conclusion The NLV value demonstrated excellent diagnostic performance for detecting varying degrees of fatty liver, and the degree of steatosis on histopathological examinations was the only significant factor affecting the NLV value.

Automated summary

The normalized local variance (NLV) ultrasound technique showed excellent diagnostic performance in detecting various degrees of fatty liver, with the degree of steatosis being the only significant factor affecting the NLV value.