Abstract: Objective To investigate imaging evaluation of hepatocellular carcinoma (HCC) using convolutional neural network model in PubMed database based on bibliometric analysis, and provide reference for clinicians. Methods PubMed database was systematically used to search for relevant literatures. The data were collected since the database established to March 3, 2021. Key information in the literature was extracted and quantitatively analyzed. The keyword frequency was analyzed through online software. Results A total of 38 relevant studies were retrieved from the database with publication date ranging from 2017 to 2021, and the maximum number of publication was in 2020 (18, 45%). The highest number of single citations was 229, and the highest average number of citations was the articles published in 2018 (nearly 80). The results of word frequency analysis showed that artificial intelligence, magnetic resonance imaging (MRI), computerized tomography (CT) and liver tumor were keywords of research hotspot. There were 8 literature with high citations, 4 were from Chinese (including Hong Kong) authors and 4 were from foreign authors. Conclusion Convolutional neural network model has been developed to maturation in recent years. More and more researchers use it for imaging evaluation of HCC. However, a visualization and easy-to-operate application system of the model should be further developed due to the complexity.

Key words: Convolutional neural network, Hepatocellular carcinoma, Imaging assessment