Computer-assisted ultrasound quantification analysis in fatty liver model of rats

BACKGROUND: Early detection of nonalcoholic fatty liver disease (NAFLD) shows its significant efficacy in preventing the patients from liver failure. The ultrasonic image quantitative analysis software can assist to diagnose NAFLD in the clinical studies. In this study, we aim to explore new method to evaluate the value of computer-assisted ultrasound in diagnosis and classification of fatty liver via Image J software. METHODS: Forty Sprague-Dawley rats were randomly divided into control group (n=10) and model group (n=30). For model group, the rats received high fat diet and subcutaneous injection of carbon tetrachloride to establish nonalcoholic fatty liver model. Ultrasound and pathological examinations on rats were performed on 4, 8, and 12 weeks. Image J software was used to measure the liver grayscale value (LGV) and renal grayscale value (RGV). The difference between LGV and RGV (LRGV) was analyzed. The diagnostic performance of computer-assisted ultrasound quantification was evaluated by receiver operating characteristic (ROC) analysis. RESULTS: We compared ultrasonic quantization parameters between control and model groups and found that the LGV and LRGV were statistically different between the normal and light fatty livers, light and moderate fatty livers, as well as moderate and severe fatty livers, respectively (P<0.05). There was no significant difference in RGV among these groups (P>0.05). Kappa statistic and Bland-Altman analyses showed the consistency of ultrasonic examination and pathological examination was good in diagnosis of fatty liver. CONCLUSIONS: This study indicated that the computer-assisted ultrasound quantification analysis, with high performance of NAFLD diagnosis like pathological examination, could provide a new and flexible noninvasive method for preclinical pharmacological research and basic research.