Bibliometrics of artificial intelligence applications in hepatobiliary surgery from 2014 to 2024

Abstract

BACKGROUND

In recent years, the rapid development of artificial intelligence (AI) in hepatobiliary surgery research has led to an increase in articles exploring its benefits. We performed a bibliometric analysis of AI applications in hepatobiliary surgery to better delineate the contemporary state of AI application in hepatobiliary surgery and potential future trajectories.

AIM

To provide clinical practitioners with a reliable reference point. It offers a detailed overview of the development of AI in hepatobiliary surgery by systematically examining the contributions of authors, countries, institutions, journals, and keywords in this domain over the last 10 years.

METHODS

The academic resources utilized in this study were obtained from the Web of Science Core Collection database. The search results were subsequently integrated and imported into CiteSpace and VOSviewer software for the purpose of visual analysis.

RESULTS

The study analyzed 2552 publications during 2014–2024. These publications collectively garnered 32 628 citations, averaging 15.66 citations per paper. The top contributor to this field was China. The USA had the highest citation count. The author with the highest citation count was Summers RM. In terms of the number of articles published, the leading journals were Medical Physics. Excluding the subject search terms, the most frequently used keywords included “classification”, “CT and “diagnosis”.

CONCLUSION

This bibliometric analysis indicates that research on AI in hepatobiliary surgery has entered a period of rapid development, particularly in the domain of disease imaging diagnostics.

Key Words: Artificial intelligence; Hepatobiliary surgery; Bibliometrics; Deep learning; Computed tomography

Core Tip: We performed a bibliometric analysis to better delineate the contemporary state of artificial intelligence (AI) application in hepatobiliary surgery, its evolution over recent years, and potential future trajectories. At the same time, we used visual maps to list in detail the current developmental trends and research status of AI, to provide practitioners with research hot spots and give a better understanding of the current situation in this field.

INTRODUCTION

An increasing number of innovative diagnostic and therapeutic technologies have been developed for both benign and malignant diseases of the liver and bile ducts[1-6]. However, surgical intervention remains the cornerstone of treatment in hepatobiliary surgery[7,8]. The main challenges in liver surgery involve accurately assessing and diagnosing the nature and location of lesions appropriately[9], and optimizing liver parenchymal preservation during surgery, while ensuring complete removal of pathological tissues[10,11]. Therefore, accurate preoperative diagnosis and localization of liver lesions are critical for guiding surgical strategy, improving surgical outcome, and reducing postoperative complications. Currently, the diagnosis of uncertain liver and bile duct masses, along with related pathologies, primarily depends on imaging and pathological evaluations, such as computed tomography (CT)[12,13], magnetic resonance imaging (MRI)[13,14], ultrasound angiography[15], and pathological biopsy[16]. However, certain liver lesions are located deep within the liver, making biopsy challenging, and atypical presentations can lead to diagnostic uncertainty, complicating lesion localization. Thus, there is an urgent need for more efficient and precise methods to assist surgeons in making informed clinical decisions.

In recent years, rapid advancements in artificial intelligence (AI) technology have fostered the development of various computer techniques, including deep learning (DL) and convolutional neural networks (CNNs)[17,18]. These technologies are increasingly applied across numerous areas within the medical field. For example, they assist radiologists in analyzing medical images[19], help internal medicine physicians with disease classification, risk prediction, and treatment recommendation[20], and support surgeons in preoperative planning, intraoperative lesion localization, and prognosis prediction for patients with malignant tumors[21]. Similarly, the use of AI in hepatobiliary surgery research is expanding significantly. For instance, researchers are using DL algorithms to autonomously analyze liver CT and MR images, facilitating the precise identification of lesion areas and assisting physicians in formulating more accurate treatment plans. AI offers distinct advantages in analyzing large-scale clinical data, with some researchers developing models to predict the prognosis of liver cancer patients[22]. The ultimate goal of these efforts is to achieve individualized precision treatment.

Bibliometrics, which utilizes citation analysis and other methodologies, is a valuable tool for evaluating the impact of research within a specific discipline over a defined period and can also help predict emerging trends[23]. With the increasing integration of AI into hepatobiliary surgery, there is a pressing need for a comprehensive analysis to assess and synthesize the current advancements in this field. Notably, there is a lack of bibliometric studies focusing on the application of AI in hepatobiliary surgery. By conducting a bibliometric analysis of AI applications in hepatobiliary surgery over the past decade, this study aimed to provide clinical practitioners with a reliable reference point. It offered a detailed overview of the development of AI in hepatobiliary surgery by systematically examining the contributions of authors, countries, institutions, journals, and keywords in this domain over the last ten years.

MATERIALS AND METHODS

Databases

The resources utilized in this study were obtained from the SCI-E database within the Web of Science (WoS) Core Collection.

Search terms

The following search terms were used: (TS = (“artificial intelligence”) OR TS = (“Deep Learning”) OR TS = (“Convolutional Neural Networks”) OR TS = (CNN)) AND ((TS = (“Hepatobiliary Surgery”) OR TS = (liver) OR TS = (“liver surgery”) OR TS = (“Hepatobiliary operation”) OR TS = (“liver operation”) OR TS = (Hepatobiliary) OR TS = (gallbladder) OR TS = (“bile duct”) OR TS = (hepatic))), country/region = unlimited, language: English. The temporal scope for document acquisition spanned from January 1 to June 1, 2024. The type of documents was confined to articles and reviews. The literature search was finalized on June 12, 2024, to mitigate the potential for data updates or bias.

Data analysis

The complete search results, including Excel tables, texts, and images, were exported and imported into CiteSpace and VOSviewer software for further analysis. This facilitated the creation of visual representations, such as the distribution and collaboration networks of authors, countries/regions, and institutions. These visualizations provided insights into the volume of published literature in the field over the past decade, the distribution of influential journals, and the prevalence of key terms.

Statistical analysis

This research systematically categorized and analyzed the collected data, presenting it in a coherent framework to illustrate the current state of AI application in hepatobiliary surgery, their development over recent years, and potential future directions.

RESULTS

Evolution over the period 2014–2024

The research development from 2014 to 2024 can be traced through the number of papers published. A total of 2552 publications were identified, reflecting significant research activity. These publications have collectively received 32 628 citations, resulting in an average of 15.66 citations per paper and an H-index of 82[24]. The number of publications showed a consistent upward trend from 2014 to 2024, highlighting the rapid growth and increasing interest in this field. There has been a notable annual increase in publications, substantiating the progressive development of the field (Figure 1).

Figure 1  Number of publications from 2014 to 2024.

Countries/regions and funding analysis

Fourteen countries were identified as the leading publishers of articles in this field, with China at the top with 891 publications, followed by the USA with 702, Germany with 193, India with 192, and Japan with 180 (Table 1). In terms of citations, the USA ranked first with 15 043, followed by China with 11 648, and Germany with 3505. The literature retrieved was analyzed using CiteSpace and VOSviewer software[25], which generated a visual map representing the network of cooperation between countries and regions (Figure 2). This encompassed the spatial distribution of hotspots and their temporal distribution. China, Germany, and Japan frequently collaborated with other nations. Additionally, we constructed a global distribution map of the existing literature to provide readers with a comprehensive understanding of the current focal regions/countries (Figure 3). The five countries/regions exhibiting the highest burst intensity were Israel (2017–2018), Turkey (2019–2020), Austria (2018–2019), Cyprus (2019–2021), and New Zealand (2019–2021) (Figure 4). These countries demonstrated significant potential for further research, suggesting a likely increase in future publications.

Figure 2 Collaboration network maps of countries/regions. A: Hot spot distribution; B: Time distribution.

Figure 3  World distribution of published papers in this field.

Figure 4  Burst figure of countries/regions.

Table 1 Major countries/regions of origin for the artificial intelligence in hepatobiliary surgery.

Rank	Country	Articles	Total citations	Average citation	H-index
1	China	891	11 648	14.18	50
2	America	702	15 043	22.54	64
3	Germany	193	3505	18.95	33
4	India	192	2983	16.31	30
5	Japan	180	2394	14.16	21
6	South Korea	175	2136	12.82	22
7	Italy	164	3055	19.35	29
8	England	144	3498	24.78	33
9	France	115	2567	22.92	28
10	Canada	86	2213	26.1	25
11	Switzerland	70	1161	16.77	18
12	Netherlands	66	1269	19.52	21
13	Spain	62	1267	20.77	19
14	Saudi Arabia	54	778	14.59	15

Table 2 outlines the top 10 sources of funding in this field, with the National Natural Science Foundation of China providing the most substantial funding at 15.909%, followed by the US Department of Health and Human Services at 10.031%, US National Institutes of Health at 9.953%, and the National Research Foundation of Korea at 2.351%. These data indicate that national governmental departments constitute the predominant source of research funding in this field.

Table 2 Top 10 funding sources in terms of number of supports.

Rank	Source of fundings	No. of grants	Percent of total funding
1	National Natural Science Foundation of China	406	15.909%
2	United States Department of Health Human Services	256	10.031%
3	National Institutes of Health	254	9.953%
4	National Research Foundation of Korea	60	2.351%
5	National Institutes of Health National Cancer Institute	60	2.351%
6	Ministry of Education Culture Sports Science and Technology Japan MEXT	54	2.116%
7	Japan Society for the Promotion of Science	51	1.998%
8	Grants in Aid for Scientific Research Kakenhi	48	1.881%
9	National Key Research and Development Program of China	41	1.607%
10	National Key RD Program of China	34	1.332%

Authors and institutions analysis

In this field of research, 13 434 authors have contributed. The five most prolific authors were primarily of Chinese origin. Wang Y led with 28 publications, followed by Liu Y and Wang K, each with 22 publications, Wang J with 21, and Zhang L and Zhang Y, each with 20 (Table 3). The authors with the highest citation counts were Summers RM (1103), Pickhardt PJ (864), and Saba L (745). A visual representation of the author collaboration network (Figure 5) illustrates the connections among authors from various countries and regions, highlighting that Julius Chapiro has the highest frequency of collaboration with other authors. Figure 5A illustrates a concentration of hotspots predominantly associated with authors such as Wang Yi and Wang Wei. Figure 6 reveals the authors with the most significant burst intensity: Tian Jie (2020–2021), Saba Luca (2021–2022), Yoon Jeong Hee (2022–2024), Summers RM (2019–2021), and Liu Zhi-Chao (2021–2022), indicating their significant potential for future contributions to this field. Table 4 shows that the institutions with the highest number of publications include the Chinese Academy of Sciences (84 publications), University of California System (74), Zhejiang University (68), Sun Yat-sen University (66), and Harvard University (62). In terms of references, the University of California System led with 2074, closely followed by the Chinese Academy of Sciences with 1976. Figure 7 presents a contact map illustrating the interaction between various institutions. Figure 7A illustrates that Sun Yat-sen University was situated at the epicenter of the hotspot. The institutions with the highest burst intensity in recent years included Korea University (2011–2013), NIC Clinic Cancer (2019≠2021), the University System of Ohio (2021–2022), Southeast (2020–2021), and the Institute of Automation (2020–2021) (Figure 8). These institutions have been demonstrated rapid research growth over the past two years.

Figure 5 Collaboration network figures of authors. A: Hot spot distribution; B: Time distribution.

Figure 6  Burst figure of authors.

Figure 7 Collaboration network maps of institutions. A: Hot spot distribution; B: Time distribution.

Figure 8  Burst figure of institutions.

Table 3 Top 15 authors in terms of number of publications.

Rank	Authors	No. of publications	Affiliated organization	Citation counts
1	Wang Y	28	College of Information and Engineering, Wenzhou Medical University, Wenzhou, China	275
2	Liu Y	22	Calico Life Sciences LLC, South San Francisco, United States	191
3	Wang K	22	Guangdong Key Laboratory of Liver Disease Research, Department of Medical Ultrasound, The Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China; CAS Key Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing, China	502
4	Wang J	21	Department of Radiology, The Third Affiliated Hospital, Sun Yat-Sen University, Guangzhou, China	129
5	Zhang L	20	School of Computer Science, and Technology, Xidian University, Xi’an, China	175
6	Zhang Y	20	Electrical and Computer Engineering Department, University of California, Los Angeles, United States; Bioengineering Department, University of California, Los Angeles, United States; California NanoSystems Institute, University of California, Los Angeles, United States	466
7	Lee JM	19	Department of Radiology, Seoul National University Hospital, Seoul, South Korea; Department of Radiology, Seoul National University College of Medicine, Seoul, South Korea; Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, South Korea	58
8	Lee S	19	Department of Radiology, Seoul National University Hospital, Seoul, South Korea	137
9	Liu Z	19	School of Computer Science and Telecommunication, Jiangsu University, Zhenjiang, Jiangsu, China	242
10	Pickhardt PJ	19	Department of Radiology, University of Wisconsin School of Medicine and Public Health, Madison, United States	864
11	Liu J	18	Department of Gastroenterology, Wuhan Union Hospital, Huazhong University of Science and Technology, Wuhan, China	65
12	Chen Y	17	School of Software, Nanchang Hangkong University, Nanchang, China	94
13	Saba L	17	Department of Radiology, Policlinico Universitario, Cagliari, Italy	745
14	Summers RM	17	Imaging Biomarkers and Computer-Aided Diagnosis Laboratory, Radiology and Imaging Sciences, National Institutes of Health Clinical Center, Bethesda, United States	1103
15	Yoon JH	17	Department of Internal Medicine and Liver Research Institute, Seoul National University College of Medicine, Seoul, South Korea	126

Table 4 Top publishing institutions.

Rank	Institution	Articles	Country	Total citations	Average citation	H-index
1	Seoul National University	97	South Korea	807	8.77	15
2	Chinese Academy of Sciences	84	China	1976	23.87	22
3	University of California System	74	America	2074	28.2	23
4	Zhejiang University	68	China	1573	23.54	18
5	Sun Yat-sen University	66	China	1048	16.23	17
6	Harvard University	62	America	1444	23.42	20
7	University of Texas System	62	America	865	14.39	15
8	Mayo Clinic	58	America	951	16.72	17
9	Fudan University	57	China	653	11.54	14
10	Institut National De La Sante et de la Recherche Medicale Inserm	47	France	1370	29.55	17
11	Harvard Medical School	44	America	685	15.7	14
12	Shanghai Jiao Tong University	43	China	302	7.14	10
13	University of London	42	England	1294	30.88	15
14	Assistance Publique Hopitaux Paris APHP	40	France	891	22.95	13

Journals analysis

A total of 744 journals were analyzed. The top 10 journals collectively published 509 articles, accounting for 68.41% of the total. Among these, each of the top 10 journals published > 30 articles (Table 5). The journals with the highest citation counts were European Radiology (1596 citations), Scientific Reports (1213), and Medical Physics (925). In terms of the number of articles published, the leading journals were Medical Physics (75 articles), European Radiology (70), and Scientific Reports (61). For practitioners in hepatobiliary surgery, these journals represent key sources of relevant research and should be closely monitored for the latest developments in the field.

Table 5 Journal distribution of the artificial intelligence in hepatobiliary surgery.

Rank	Journal	Articles	IF as of 2023	Total citations	Average citations/article	JCR	H-index
1	Medical Physics	75	3.2	925	12.56	Q1	18
2	European Radiology	70	4.7	1595	23.91	Q1	19
3	Scientific Reports	61	3.8	1213	20.1	Q1	15
4	Frontiers in Oncology	56	3.5	454	8.25	Q2	11
5	Diagnostics	54	3.0	358	6.76	Q1	12
6	IEEE Access	45	3.4	819	18.36	Q2	15
7	Cancers	39	4.5	362	9.38	Q1	12
8	Computers in Biology and Medicine	39	7.0	754	19.64	Q1	15
9	Abdominal Radiology	35	2.3	455	13.17	Q2	11
10	International Journal of Computer Assisted Radiology and Surgery	35	2.3	837	24.14	Q2	13

Hot spot analysis

This study used keyword co-occurrence analysis to identify the main research hotspots within the field. High-frequency keywords generally reflect the dominant research directions. Co-occurrence analysis facilitated the identification of these research trajectories and aided in addressing potential research gaps. The keyword co-occurrence map, generated using VOSviewer software, identified 8109 keywords, with 325 surpassing the threshold of > 10 co-occurrences (Figure 9). The top 10 most frequently occurring keywords were “deep learning”, “artificial intelligence”, “machine learning”, “liver”, “classification”, “CT”, “diagnosis”, “segmentation”, “cancer” and “hepatocellular carcinoma”. Figure 10 presents a keyword clustering analysis from 2014 to 2024, revealing 15 distinct clusters, primarily focusing on hepatocellular carcinoma (HCC) and image segmentation. Figure 11 provides a timeline view of keywords over the same period. The evolution of research trajectories within this field can be observed across various historical periods. Figure 12 highlights the top 20 keywords with the most significant increase in frequency over the past decade. Notably, the keyword bursts were predominantly concentrated between 2018 and 2020, indicating a critical period phase of rapid development and exploration within this domain.

Figure 9 Keyword frequency and correlation graph. A: Hot spot distribution; B: Time distribution.

Figure 10  Keyword clustering diagram.

Figure 11  Keyword timeline figure.

Figure 12  Keyword burst figure.

DISCUSSION

This study was the first to use bibliometric methods to analyze the current status and trends of AI applications in hepatobiliary surgery. Our search in the WoS database identified 2552 documents. Before 2017, research output in this field was minimal. However, since 2017, the annual number of publications has consistently reached double digits. A notable surge in research activity occurred between 2019 and 2023, with the number of published documents exceeding 500 per year by 2021. Projections suggest that this upward trend will continue, with the number of publications potentially reaching a new peak in 2024. The substantial increase in publications over the past 5 years indicates that research on AI in hepatobiliary surgery has entered a phase of rapid development, highlighting its emergence as a growing field of global significance.

In our study, we found that the majority of the literature consisted of research articles. This trend may be attributed to the current focus of the field on reporting advances in emerging technologies, alongside a comparatively low demand for review documents. Our analysis revealed that the three most frequently cited documents mainly explored the use of AI and CNNs for segmenting and identifying liver lesions, such as tumors and other hepatic imaging features. The current emphasis of AI research in this domain is to supports surgeons in the rapid and accurate detection of lesions, thereby enhancing surgical decision-making and improving the accuracy of imaging recognition. The most frequently cited paper in this field, published in 2018 with a total of 1265 citations, introduced a novel hybrid densely connected UNet (H-DenseUNet). This approach has effectively achieved accurate and automatic segmentation of the liver and tumors by overcoming the limitations of 2D convolution, which does not fully use 3D spatial information, while also mitigating the high computational costs and substantial graphics processing unit memory requirements associated with 3D convolution[26]. This provides the possibility for AI to help surgeons diagnose liver cancer and formulate surgical plans in the future, despite this promise, the current limitations in big data and algorithms necessitate further refinement and rigorous validation before such technologies can be reliably integrated into clinical practice. However, this work has significantly influenced subsequent publications and servers as a benchmark for citations and further research in the industry. Additionally, our findings indicate that CT is the most frequently referenced diagnostic method[27-30], likely due to its ease of use and widespread availability. Beyond image segmentation and recognition, some researchers have used AI DL techniques to predict the survival rates of HCC. For example, a DL model was developed using RNA sequencing, miRNA sequencing, and methylation data from The Cancer Genome Atlas to predict survival outcomes in 360 HCC patients[22]. This model has been validated with external datasets and shows potential for future clinical application. Beyond its application in liver cancer diagnosis, localization, and patient prognosis, AI has been developed to aid pathologists in distinguishing subtypes of primary liver cancer. A notable study by Kiani et al[31], published in 2020, introduced a DL-based model designed to assist pathologists in differentiating subtypes of primary liver cancer using whole-slide images stained with hematoxylin and eosin. The findings indicated that the model significantly enhanced the accuracy of pathologists across varying expertise levels. However, incorrect predictions by the model adversely affected overall predictive accuracy. In 2017, Liu et al[32] introduced a methodology for feature extraction from image segments cropped around the liver capsule utilizing a deep CNN model. Subsequently, a trained support vector machine classifier was used to conduct differential diagnosis on the samples, effectively facilitating the diagnosis of early liver cirrhosis. This technology serves as a complementary tool to liver CT diagnosis and holds significant importance. Nevertheless, due to limitations in the sample size for feature extraction, this approach currently exhibits substantial errors and is not yet suitable for reliable application in clinical practice. Consequently, it is crucial to consider the unpredictability of the model in current applications. As the model undergoes further refinement, AI holds the potential to become an invaluable tool for clinical practitioners in the future.

China led in the number of published papers in this field, followed by the USA and Germany. Although China’s total citation count ranked second only to that of the USA, its average citation rate was lower than that of the USA, Germany, and India, indicating room for improvement in the quality and impact of its research. An analysis of the cooperation network diagram indicates that countries with higher centrality in collaborative efforts included China, Japan, India, Germany, and France. The 14 universities with the highest number of publications were located in China, South Korea, USA, France, and England, with both China and the USA contributing five universities each. The Chinese Academy of Sciences garnered 1976 citations, ranking it second only to the University of California System. The cooperation network analysis revealed that Zhejiang University, Sun Yat-sen University, and the University of Hong Kong had notable collaborative interactions with other institutions, while Korean universities demonstrated the highest research output intensity in recent years. In terms of funding, projects supported by the National Natural Science Foundation of China significantly exceeded those funded by other countries. China had an absolute leading position in this field, which may be due to the fact that China has one of the highest incidences of liver cancer globally[33]. The Chinese Government has provided considerable attention and support to research in this area. Furthermore, China’s recent economic ascent, coupled with its focus on scientific research and innovation, has established a robust foundation for the advancement of AI technology. This includes research and development of AI intelligent systems and the widespread adoption of AI machinery, thereby catalyzing progress in the medical field. While China ranks highest in terms of the quantity of published papers, the average citation count for Chinese publications remains comparatively low. Additionally, the H-index, a metric used to assess the impact of scholarly contributions, is also lower for China than for the USA. This indicates that the quality of research outputs from China may require enhancement. Analysis of the international collaboration network revealed that China’s centrality was not the most prominent, and its level of collaboration with other countries was not the highest. Consequently, it is advisable for China to foster broader cooperation and exchanges with other nations, such as the USA, India, Japan, and Germany, to facilitate the production of higher-quality research. Among the top 15 authors by the number of publications, eight were from China, With Wang Y from Wenzhou Medical University having the highest number of publications. Additionally, Tian Jie has demonstrated significant research impact in recent years and is considered one of the most promising scholars in this field[34,35]. Practitioners should closely follow his research developments.

Chinese researchers have been pioneers in the application of AI for the diagnosis of liver diseases. In 2021, a research team led by Tian Jie published a landmark article in European Radiology. This study included a cohort of 807 patients with chronic liver disease and analyzed 4842 images sourced from three different hospitals. The team developed a DL elastic imaging radiomics model (DLRE2.0) for diagnosing fibrosis in chronic liver disease, demonstrating substantial robustness in its test results and marking a significant advancement in the integration of AI into hepatobiliary surgical research[34]. Tian Jie, an esteemed expert affiliated with the Key Laboratory of Molecular Imaging at the Institute of Automation, Chinese Academy of Sciences, has collaborated extensively with Dan Liu from the Department of Artificial Intelligence Technology at the same institution, Xiao-Yan Xie from the Institute of Ultrasound Diagnosis and Intervention at the First Affiliated Hospital of Sun Yat-sen University, Ming Kuang from the Department of Liver Surgery at the First Affiliated Hospital of Sun Yat-sen University, among others. This collaboration underscores the necessity of multidisciplinary efforts in the application of AI in medicine, integrating foundational algorithmic research with clinical expertise. Within the top 10 journals, Medical Physics had the highest number of published articles, while Computers in Biology and Medicine had the highest impact factor. Among these top 10 journals, five were classified under nuclear medicine or imaging surgery, three within the domain of computer science, and two were related to oncology. European Radiology accrued the most citations, totaling 1595. The advancement of AI is inherently a multidisciplinary endeavor, encompassing the domains of computer science, nuclear medicine, and imaging. Progress in these interconnected fields is essential to propel the future of hepatobiliary surgery, enabling AI to assist clinicians in devising more effective medical strategies.

In bibliometrics, keywords are considered the main topics within a field during a specific time period. In this study, we visualized the evolution of these keywords through clustering and timeline analysis, providing a clearer view of current key topics and potential future research directions. Excluding subject search terms, the most frequently used keywords included “classification”, “CT”, “diagnosis”, and “segmentation”. These were followed by terms such as “cancer”, “hepatocellular carcinoma”, “model”, “radiomics”, “MRI”, and “convolutional neural network”. Liver surgery is recognized as one of the most complex and challenging procedures in contemporary hepatobiliary surgery. Despite years of advances, accurately localizing malignant hepatic tumors remains a significant challenge. The advent of AI offers promising prospects to overcome this hurdle. Researchers are developing models that utilize computational memory functions, segmentation, and recognition techniques to achieve accurate lesion localization, ultimately aiding surgeons in making more informed clinical decisions. An analysis of keyword trends reveals that terms such as “algorithm” and “cirrhosis” began to gain prominence between 2014 and 2015. From 2015 to 2020, there was a notable increase in computer-related keywords, along with a gradual emergence of terms related to fatty liver and liver cancer. By 2024, the predominant themes remained “cancer” and “diagnosis”, as supported by the cluster analysis diagram. The evolution of keywords over the past decade signifies the developmental trajectory of this field, exemplified by the CNN. Initially introduced by Yann LeCun and colleagues in 1989, the advancement of CNNs was initially hindered by limited computational resources. However, the advent of graphics processing unit technology in 2010 markedly enhanced computational capabilities, providing CNNs with unprecedented development opportunities. Post-2020, further advances have been made in CNN optimization. An examination of the timeline reveals that the progression of AI in medicine has been intrinsically linked to the formulation of computer algorithm models. While the establishment of AI models has propelled advancements, the inherent complexity and challenges associated with liver cancer ensure that this topic remains a focal point within the field. In the future, the integration of AI within the medical field will be intrinsically linked to patient safety and health outcomes. It must play an important role in the early screening and diagnosis of liver cancer. To facilitate its application in clinical practice, it is also imperative for researchers to advance the architecture of AI neural networks. This may include the incorporation of optimized and interpretable modules through multimodal learning or visualization technologies, thereby enhancing model transparency. AI-generated outcomes must undergo rigorous evaluation and approval by authoritative figures in the medical domain. Consequently, the progression of this field is fraught with challenges and issues requiring resolution and enhancement. The trajectory of development in this area is extensive and demands sustained effort.

However, this study had several limitations. Firstly, it relied solely on the WoS core database SCIE literatures, known for its high-quality research, which may have resulted in the exclusion of relevant articles from other databases (e.g., PubMed and CNKI). Secondly, the study was limited to English-language publications, potentially overlooking research published in other languages. Thirdly, given that the WoS Core Collection is continually updated and the field is rapidly evolving, this bibliometric study had data timeliness and may need to be updated in the future. Lastly, the article was collaboratively authored by multiple contributors, and efforts were made to maintain methodological consistency; however, the potential for selective bias remains challenging to eliminate entirely. Therefore, in the future we may further improve the methodology and include research that has significant influence and significance in databases other than the WoS database.

CONCLUSION

This field primarily concentrates on tumor segmentation, computer-aided localization diagnosis, and HCC, with China exhibiting a predominant influence. The study suggests that future research will persist in enhancing the precision of early malignant tumor diagnosis and localization. Additionally, the enhancement of the role of AI in the early diagnosis of liver cancer, cirrhosis, and prognosis prediction can be significantly achieved through the optimization of neural network architectures, the robust development of multimodal learning and visualization technologies, and the expansion of network nodes. This bibliometrics research holds substantial significance for both researchers and clinicians, offering a new perspective on integrating AI into hepatobiliary surgical practices.

ACKNOWLEDGMENTS

Thanks to Song Su and Yue Hu for reviewing and revising the article and supporting the project.