Objective: Hepatic arteriovenous fistula (HAVF) can be caused by trauma, hepatic biopsy, bile duct radiology, etc. Small intrahepatic HAVF can be found in lesions of carcinoma and hepatocirrhosis. Accurate detection of HAVF was magnitude in the process to take appropriate treatment in clinic. The aim of this article is to evaluate the imaging diagnostic value on HAVF and to study the imaging character of HAVF in patients with hepatocellular carcinoma (HCC), and to evaluate the role of ultrasonographic and radiological techniques in the diagnosis and management of developmental intrahepatic shunts so as to assess the imaging diagnostic evaluation in a follow‐up study.

Methods: Seventy‐eight patients diagnosed with HCC were enrolled in this study, and retrospective analysis of ultrasonographic and radiological data was carried out on all 78 patients, and 25 patients suspected of having HAVF were selected. The results from ultrasonography were compared with that from digital subtraction angiography (DSA) as a gold standard. The portal and hepatic veins, hepatic arteries and vessels around and inside the tumour patients were detected and the haemodynamic indices were recorded with ultrasonography. Ten patients with HAVF were followed up after the therapy of arterial embolization and the reversal effect of the therapy was observed. Associations of HAVF with clinical and ultrasonographic features were evaluated by stepwise logistic regression analysis.

Results: Twenty‐five of 78 HAVF patients were detected by ultrasonography and other imaging methods. Ultrasonographic parameters made excellent predictions for the patients with HAVF; sensitivity (SE) 83.3%, specificity (SP) 90.7%, positive predictive value (pPV, 80.0%), negative predictive value (nPV) 92.5% and accuracy 88.5%. Among the 25 HAVF patients, 16 were central hepatic artery–portal vein fistulae, seven were peri‐hepatic artery–portal vein fistulae and two were hepatic artery–vein shunts. Characteristic ultrasonographic methods of hepatic artery–portal vein fistulae were as follows: 10 patients with hepatic artery–portal vein fistula were followed up after embolization. Compared with that of preembolization, seven cases returned to normal and in three patients abnormalities were still detected.

Conclusion: Ultrasonographic evaluation of HAVF is easy to perform, reproducible and, when present, gives a high degree of certainty for the diagnosis of HAVF. Ultrasonography is a valuable method for the diagnoses HAVF; it can offer imaging evidence after the treatment of hepatic cell cancer. HAVF in patients with HCC can be detected by ultrasonographic methods, which are characterized by changes of hepatic arteries and veins involved in fistulas. These can be used for diagnosing HAVF and evaluating its arterial embolization effect in patients with HCC.