Global Research Trends of Artificial Intelligence on Histopathological Images: A 20-Year Bibliometric Analysis

Abstract

Cancer has become a major threat to global health care. With the development of computer science, artificial intelligence (AI) has been widely applied in histopathological images (HI) analysis. This study analyzed the publications of AI in HI from 2001 to 2021 by bibliometrics, exploring the research status and the potential popular directions in the future. A total of 2844 publications from the Web of Science Core Collection were included in the bibliometric analysis. The country/region, institution, author, journal, keyword, and references were analyzed by using VOSviewer and CiteSpace. The results showed that the number of publications has grown rapidly in the last five years. The USA is the most productive and influential country with 937 publications and 23,010 citations, and most of the authors and institutions with higher numbers of publications and citations are from the USA. Keyword analysis showed that breast cancer, prostate cancer, colorectal cancer, and lung cancer are the tumor types of greatest concern. Co-citation analysis showed that classification and nucleus segmentation are the main research directions of AI-based HI studies. Transfer learning and self-supervised learning in HI is on the rise. This study performed the first bibliometric analysis of AI in HI from multiple indicators, providing insights for researchers to identify key cancer types and understand the research trends of AI application in HI.

Keywords: CiteSpace; VOSviewer; artificial intelligence; bibliometrics; histopathological images.

Figure 1

Global Publication Trends on AI in HI from 2001 to 2021.

Figure 2

Country production map based on the total publications. The color indicates the number of publications.

Figure 3

The network of countries/regions’ co-authorship analysis. Node color and size indicate average publication year and the number of publications, respectively; thickness of lines indicates the strength of the relationship. Each country/region in this map has at least 5 collaboration publications. The average year of publications in different countries are from 2017 to 2020.

Figure 4

The network of institutions’ co-authorship analysis. Node color and size indicate the average publication year and the number of publications respectively; thickness of lines indicates the strength of the relationship. Each institution in this map has at least 10 collaborative publications. The average years of publications in the institutions are from 2018 to 2020.

Figure 5

The network of authors’ co-authorship analysis. Node color and size indicate the average publication year and the number of publications, respectively; thickness of lines indicates the strength of the relationship. Each author in this map has at least 7 collaborative publications. The average years of publications of authors were from 2016 to 2020.

Figure 6

Dual-map overlay of journals on the application of AI in HI. The width of the paths is proportional to the z-score-scaled citation frequency. From top to bottom, the content in the main paths are: MATHEMATICS, SYSTEMS, MATHEMATICAL $\to$ MOLECULAR, BIOLOGY, GENETICS (z = 1.84, f = 986); MOLECULAR, BIOLOGY, IMMUNOLOGY $\to$ MOLECULAR, BIOLOGY, GENETICS (z = 4.01, f = 1937); MOLECULAR, BIOLOGY, IMMUNOLOGY $\to$ HEALTH, NURSING, MEDICINE (z = 2.60, f = 1321); MEDICINE, MEDICAL, CLINICAL $\to$ MOLECULAR, BIOLOGY, GENETICS (z = 4.66, f = 2222); and MEDICINE, MEDICAL, CLINICAL $\to$ HEALTH, NURSING, MEDICINE (z = 5.56, f = 2617).

Figure 7

The network of author keywords’ co-occurrence analysis. Node color and size indicate average publication year and the number of occurrences, respectively; thickness of lines indicates the strength of the relationship. The occurrences of each keyword in this map were at least 11 times. The average years of the publications with the keywords were from 2017 to 2020.

Figure 8

(A) The co-citation map of reference. (B) Timeline visualization of co-citation map. The label selection method was the log-likelihood ratio. Lines indicate reference relations, and color of lines and circles from yellow to green represent the years from 2001 to 2021. Red circles indicate the burst citation, which means that the number of citations to the publication increased rapidly, lasting for multiple years or a single year. The vertical direction from top to bottom represents the cluster from large to small, the largest cluster is shown on the top with label “#0”. The horizontal direction represents the timeline from 2001 to 2021. A node indicates a publication. The larger node size, the more times co-cited.