The Web of Science (WoS) contains 20,000 reputable academic publications that span 250 different fields worldwide (Zhong and Lin, 2022). Other academic researchers in the field of bibliography have used the WoS as the most trustworthy data source for data extraction in bibliometric analysis (Dong et al., 2022).

We conducted a comprehensive literature search using the Web of Science Core Collection (WoSCC) database from 1 January 2004 to 11 August 2022. To obtain as comprehensive and accurate results as possible, the search strategies we used were TS=(stroke OR apoplexy OR “cerebrovascular accident” OR “cerebral hemorrhage” OR hematencephalic OR encephalorrhagia OR “cerebral ischemia”) AND TS=(“Stem Cells” OR “Cell, Stem” OR “Cells, Stem” OR “Stem Cell” OR “Progenitor Cells” OR “Cell, Progenitor” OR “Cells, Progenitor” OR “Progenitor Cell” OR “Mother Cells” OR “Cell, Mother” OR “Cells, Mother” OR “Mother Cell” OR “Colony-Forming Unit” OR “Colony Forming Unit” OR “Colony-Forming Units” OR “Colony Forming Units”). Only articles and reviews were included. Furthermore, letters, commentaries, meeting abstracts, and other types of documents were excluded. Finally, 6,703 records were included for analysis. The specific literature screening process is presented in Figure 1.

The original data downloaded from the WoSCC were firstly imported into Microsoft Excel 2016, and then two authors (QZh and YZ) independently screened the final included articles and collected all data from the final papers that were included, such as titles, authors, keywords, institutions, countries/regions, citations, journals, and publication dates. Subsequently, the processed data was imported to VOSviewer (version 1.6.15), CiteSpace (version 5.8), and R package “bibliometrix” for bibliometric analysis.

CiteSpace is a bibliometric software that enables the analysis and visualization of trends and patterns in a research area (Pan et al., 2018; J; Zhang and Lin, 2022). It also creates a knowledge map of connected fields, clearly presents the panoramic information of a particular knowledge field, and identifies the critical studies, hot research, tendency, and frontiers of a specific scientific field using a variety of dynamic network analysis techniques (Godfrey et al., 2018; Y; Chen, Lin, and Zhuang, 2022). CiteSpace was used in this study to conduct co-occurrence and cluster analyses of authors, research institutions, nations, and discipline features. The parameters of CiteSpace were set as follows: in the Time Slicing column time settings 2004.01–2022.08, each year is a time slice.

The Leiden University Center for Science and Technology Studies (CWTS) created VOSviewer, a software for creating and analyzing bibliometric networks (Netherlands). VOSviewer can extract bibliographic networks (co-authorship, co-occurrence, and citation-based) from bibliographic data (Lin, Chen, and Chen, 2020; Moral-Muñoz et al., 2020; Luo and Lin, 2021). In this study, co-occurrence and cluster analyses of authors, research institutions, countries, and discipline features were conducted using CiteSpace.

Bibliometrix (https://www.bibliometrix.org) is an open-source R package developed by Dr. Massimo Aria and Corrado Cuccurullo from Naples University in Italy. It is capable of conducting comprehensive bibliometric and scientometric analyses (Moral-Muñoz et al., 2020). In this study, bibliometrix was used to create a global distribution network of articles on stem cells in stroke and to analyze the thematic evolution of those publications (Aria and Cuccurullo, 2017).

As can be seen from Figure 2, the histogram and curves exhibit two trends: the total number of papers published and citations per year. Both trends grow throughout time, illustrating the direction in which research in this area is moving. The number of citations significantly and rapidly increased between 2004 and 2021, suggesting that research on stem cells in stroke has attracted interest. From 2004 to 2007, the number of literature grew rapidly, and in 2009, it dramatically increased. However, the number of articles remained relatively stable from 2010 to 2017. The year 2020 had the most number of publications in recent years, which peaked at 515. Although the data for 2022 have not yet been completed, it is predicted that they will exhibit a moderate trend compared with those in the previous year.

As for the geographical distribution, 6,703 documents were published from 94 different countries and regions; Table 1 presents the top 10 countries/regions in this category. As can be seen from Figure 3B and Supplementary Figure S1, we also used VOSviewer for the visual analysis of countries or regions. The USA published the most papers (1,555papers, 25.61%), followed by China (738 papers, 12.15%) and then Japan (714papers, 11.76%), indicating that these three countries play a crucial role in this field. The quantity and connections among publications in each nation were then used to create a collaborative network (Figure 3A). A country collaboration analysis was conducted on the 30 countries with the highest number of publications in this area (Figure 3C). According to the total link strength, the top five countries/regions were the USA, China, Japan, Germany, and South Korea.

In Table 2, the top 10 institutions with the highest productivity are ranked by their productivity. China and the USA were the only two countries where the highest productivity were located. The most prolific institution was the University of South Florida (164 publications, 2.70%), followed by the Oakland University (151 publications, 2.49%) and then the Shanghai Jiao Tong University (129 publications, 2.12%). The clustering analysis of institutions is presented in Figure 4A. A tight and continuous interaction between institutions can also be observed. Among them, the institutions with more collaborations were Oakland University, Henry Ford Hospital, and Henry Ford Health Science Center, followed by Sapporo Medical University and Yale University. In Figure 4B, we analyzed the data of articles published in the last 5 years using VOSviewer. For instance, Harvard University started research in the field earlier and had published significantly more articles in the past than it had recently. Contrarily, Capital Medical University entered the field later and has recently published a higher number of articles. As can be seen from the cluster analysis figure, the red and light blue circles indicate mainly Chinese institutions. Combined with the visual timeline, it can be seen that Chinese institutions are predominantly yellow, indicating that they entered the field late or have recently published a high number of articles.

Figure 4C presents the publication trend in this field by different institutions over time. The University of South Florida accounted for more than half publications from 2004 to 2006 and then gradually declined; however, the University of South Florida remained in the leading position, demonstrating the remarkable contributions of this institution to this sector. The rest of the institutions exhibited an upward trend in the publication quantity. Interestingly, Shanghai Jiao Tong University has made the most significant progress. Since 2018, Shanghai Jiao Tong University has, unsurprisingly, ranked first among the 10 institutions in terms of publication ratio. We also used CiteSpace for the visual analysis of institution clusters and marked them with keywords (Figure 4D). Consequently, the Johns Hopkins University, Massachusetts General Hospital, and National University of Singapore exhibited high centrality. The largest cluster in Supplementary Figure S1 was designated “ischemic stroke” (cluster #0), indicating that various institutions are most concerned with this word. It was followed by “clinical trials” (cluster #1), “primate” (cluster #2), and “extracellular matrix” (cluster #3), respectively. Other important clusters were “mesenchymal stem cell,” “neurological function,” and “hypothermia.”

Approximately 174 writers contributed to a total of 6,703 articles. The most prolific author was Chopp, M. who produced 160 articles (2.63%), closely followed by Borlongan, Cesario, V who produced 143 publications (2.35%). Three authors published 50 and more articles (Zhang Zheng Gang, Yang Guo-Yuan, and Kokaia Zaal), and five authors published 40 and more articles (Hermann Dirk M. Chen Jieli, Lindvall Olle, Sanberg P. R. and Kaneko Yuji) (Table 3). Information on co-authors and co-cited authors was also analyzed using VOSviewer (Figures 5A, B). Figure 5A shows that there are primarily two research teams involved in author collaboration, led by Chopp, M. and Borlongan, Cesario, V, who frequently and closely collaborate with other authors. Co-cited authors are those who have had two or more of their names concurrently mentioned in one or more subsequent articles and who are therefore considered to have a co-citation connection. A total of more than 1000 citations have been received by the top six authors among the top 10 co-cited writers (Table 3). The most frequently referenced author was Chen Jl. (n = 2,214), followed by Jin, Kl (n = 1,473), Li, Y (n = 1,367), Zhang, RI (n = 1,313), Borlongan, Cv (n = 1,144), and Avidsson, A (n = 1,003).

We found that 252 journals published 6,703 papers regarding stem cells in stroke. As can be seen from Table 4, it is clear that the journal Cell Transplantation has the most papers (172, 2.83%), followed by Brain Research (160, 2.63%). Among the top 10 journals, Stroke (10.17) has the greatest impact factor (IF). The number of times the top 10 most co-cited journals are cited determines their influence. As presented in Table 4, the publication with the most citations is Stroke (19,776), indicating that it has a significant impact in this category, followed by the Journal of Neuroscience (14,061) and Proceedings of the National Academy of Sciences of the United States of America (11,769).

Using VOSviewer, we conducted a visual analysis of the published journals and obtained details about journal collaboration through Figures 6A, B. We could see that the journals of Stroke, Archives of Physical Medicine and Rehabilitation, and Neurorehabilitation and Neural Repair had more times of co-citation and greater influence. We also conducted comparative analysis of the journals’ popularity, as presented in Figure 6C. Through this heat map, we can understand the change in the research direction and emphasis in this field and grasp the development trends. We found that in recent years, the popularity of NEUROSURGERY, CURRENT NEUROVASCULAR RESEARCH, and PANS had gradually decreased, whereas that of CELLS, INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES, and FRONTIERS IN CELLULAR NEUROSCIENCE had gradually increased. One of the interesting things about STEM CELL REVIEWS AND REPORTS is that its popularity had declined year by year, but in the last 2 years, it had returned to its level in 2004. Furthermore, on the dual-map overlay of journal publishing research (Figure 6D), we found four citation paths (colors orange, pink, and green), demonstrating that the studies published in molecular/biology/genetics journals and health/nursing/medicine journals were mainly cited by the studies published in molecular/biology/immunology, medicine/medical/clinical, and neurology/sports/ophthalmology journals.

The top 15 documents that were cited the most often out of the 6703 retrieved are listed in Table 5. Mesenchymal stem cells as trophic mediators was the most frequently cited (2,082), which is a review of studies on the applications of adult marrow-derived MCSs. It was followed by Concise review: Mesenchymal stem cells: Their phenotype, differentiation capacity, immunological features, and potential for homing (1,725) and then Adult mesenchymal stem cells for tissue engineering versus regenerative medicine (1,378).

When two or more articles are simultaneously cited in the same article, the relationship between the simultaneously cited articles is referred to as co-citation. As presented in Figure 7A, we also used the CiteSpace clustering function to create a visual map to cluster the co-citation literature, and the collected literature was separated into 11 clusters via cluster analysis. Each cluster was intimately connected to the others and worked together in specific areas. The weighted mean silhouette and modularity Q were 0.8694 and 0.691, respectively, demonstrating the stability, believability, and persuasiveness of the clustering structure. Figure 7A illustrates the time dimension by the color of circle changing from purple to yellow, which also shows the change in the direction and concentration of the research. While the “hippocampus” cluster had received more attention in the past, researchers recently turned their attention to “exosomes,” “hydrogels,” and “ischemic stroke.” The name of the biggest cluster was “subventricular zone” (cluster #0), which was followed by “transplantation” (cluster #1), “ischemic stroke” (cluster #2), and “hydrogel” (cluster #3). The “hippocampus,” “neurogenesis,” “bone marrow stromal cells (BMSCs),” and “exosomes” clusters were other significant groups that may have represented a turning point in some sense.

As shown in Figure 7B, we created a visual map to cluster the cited literature over the last 3 years and divided them into 21 clusters through cluster analysis using CiteSpace. The weighted mean silhouette and modularity Q were 0.8729 and 0.7518, respectively, demonstrating the stability, believability, and persuasiveness of the clustering structure. The Figure 7B can help us get the latest research hotspots. The largest cluster was “cell treatment” (cluster #0), which was followed by “exosomes” (cluster #1), “biomaterials” (cluster #2), and “ischemic stroke” (cluster #3). Additional significant clusters were “adult neurogenesis,” “microglia,” “intracerebral hemorrhage,” and “pericytes.” Combined with Figure 7A, the “hippocampus” cluster has received more attention in the past, researchers have recently turned their attention to “exosomes,” “hydrogels,” and “ischemic stroke.” In Figure 7C, the top 25 references are listed in chronological order, which have the greatest burst intensity. References that receive several citations over a period of time are known as “citation burst” references. We set the time period in CiteSpace to 2004–2022 and still kept references with a burst termination date of 2022. Neuronal replacement from endogenous antecedents in the adult brain following stroke by Andreas et al. was published in Nature Medicine in 2002, and it had the strongest burstiness (strength = 81.35), occurring from 2004 to 2007 (Arvidsson et al., 2002). The advantages and disadvantages of the top 25 articles with the strongest citation bursts are summarized in Supplementary Table S1.

The use of cluster analysis to cluster the included keywords and summarize the study subjects might be helpful for relevant researchers in identifying popular topics and assisting scholars in better understanding current scientific concerns. We used VOSviewer to cluster the keywords into eight, as presented in Figure 8A: origin and behavior of stem cells (red), pathophysiological process of stroke (green), treatment of stroke and application of stem cells (purple), effects of stem cell therapy after stroke (light blue), cells that make up the central nervous system and the pathophysiological changes (orange), other diseases associated with stem cell therapy (brown), exosomes and mechanism of action (yellow), others (navy blue).

Meanwhile, we performed a series of keyword burst detections. To evaluate the development of stem cells in stroke research, researchers used a method named “keyword burst detection,” which is the recognition of phrases that often occur in a certain period of time. Table 6 demonstrates that terms with a high frequency in this study, aside from “stroke” (1538), include “ischemia” (801), “neurogenesis” (660), “brain ischemia” (553), “mesenchymal stem cell” (549), and “neural stem” (541). Prominent keywords were further analyzed. The top 25 keywords are listed Figure 8B in chronological order, which have the greatest burst intensity. It can be seen from the figure that the keywords that had exploded in the past 5 years mainly focused on “extracellular vesicles,” “inflammation regulation,” and “regulation of extracellular matrix.” A timeline analysis that visually depicts the research hotspots and development paths of stem cells in stroke at various stages from a temporal perspective is presented in Supplementary Figure S2. The keywords that have received extensive attention in this field appear early, for example, cerebral ischemia, regeneration, NSC, endothelial progenitor cells, apoptosis, and cell therapy. Emerging hot words in the past decade include adhesion molecule, mesenchymal stromal cell, extracellular vesicle, pluripotent stem cell, signaling pathway, plasticity, and exosomes.

This study aimed to conduct a bibliometric analysis of the last 18 years’ worth of research on stem cells in stroke. The citation count has been steadily increasing each year, with the 2008–2009 period showing the clearest growth pattern. Prior to 2008, an average of 150 papers per year were published in this field. From 2009 to 2022, the number of publications gradually increased, with an average of 350 papers published annually. The number of articles published in 2020 peaked at 515. These findings indicated that studies on stem cell therapy for stroke have gained increasing attention from researchers from all over the world in recent years, and the research area is currently in a steady developmental stage.

The top three authoritarian countries performing research on stem cells in stroke are the USA, China, and Japan. Four European nations, four Asian-Pacific nations, and two American nations make up the top ten nations. Furthermore, among the top 10 research institutions, five were American, and the remaining five were Chinese. We observed a close coordination between four nations, namely, Germany, Japan, China, and the United States. China also actively works with Japan, Canada, and England. There are some academic institutions that collaborate well with one another, such as Shanghai Jiao Tong University, Capital Medical University, and Harvard University. Although Oakland University ranked second in terms of paper publications, it was found to have limited collaborations with other universities. Such collaborations are detrimental for these universities in terms of long-term scholarly advancement. Consequently, it is indeed necessary that research organizations from many nations work closely together and communicate to collaboratively promote stem cell therapy for stroke.

As for the journals, those listed in Table 4 may be the core journals of the publication of research on stem cells in stroke. The most widely followed journal in this field of study is Cell Transplantation (IF = 4.139, Q3), with the majority of research on stem cell therapy published in this journal. Stroke (IF = 10.17, Q1) has the greatest IF, which also received the most citations (19,776 times). According to the Journal Citation Reports (2021 edition), five journals had IF values between 5 and 10 (Journal of Cerebral Blood Flow And Metabolism, International Journal Of Molecular Sciences, Experimental Neurology, Stem Cell Research and Therapy, and Neural Regeneration Research), four had an IF value between 3 and 5 (Cell Transplantation, Brain Research, Plos One, and Neuroscience), and no journal had an IF value below 3. These results indicated that most studies were published in these high-quality journals, and when researching on this topic, academics should concentrate on the content published in these journals. Most of the co-cited journals are high-impact Q1 journals, as can be seen from the list of co-cited journals. These journals support the investigation of stem cells in stroke and are undoubtedly of high quality. Furthermore, current research on stem cells in stroke is mostly published in journals pertaining to molecular biology, immunology, and medicine, followed by journals about neurology, sports, and ophthalmology. This indicates that basic research still constitutes the focus of the research and that the proportion of clinical studies is relatively low.

As for the authors, Chopp, M. (Oakland University, USA) is the most prolific, followed by Borlongan, Cesario, V (Stanford University, USA). They contributed to so many publications and were leaders in this field. Before 2015, Professor Chopp, M. and his team mainly focused on the role of stem cells in stroke, including BMSCs, neural progenitor cells, and human umbilical cord blood cells. In 2002, they found that after adult mouse suffered focal cerebral ischemia, endothelial progenitor cells generated from bone marrow contributed to brain neovascularization (Z. G. Zhang et al., 2002). Then in 2006, they demonstrated that intracarotid transplantation of BMSCs increased axon-myelin remodeling following stroke (L.H. Shen et al., 2006). Their experimental results in 2010 indicated that following a stroke in mice, MSC-mediated enhanced tPA activation in astrocytes encouraged neurite development (Xin et al., 2010). The next year, they published a study on how the production of astrocytic endogenous glial cell-generated neurotrophic factor was enhanced by BMSC implantation in the ischemic boundary area following stroke in adult rats (Shen et al., 2010). In the same year, they confirmed that the subventricular area has more progenitor cells dividing than normal due to the human umbilical cord tissue-derived cells (hUTCs) (Zhang et al., 2011) and had a neurorestorative effect (Ding et al., 2013). They proposed that the level of proinflammatory factors in the blood can be significantly reduced after hUTC transplantation (Bae et al., 2012). Furthermore, Chopp, M. discovered several chemicals that control stem cell migration, differentiation, and proliferation, such as specific miRNAs (X. S. Liu et al., 2013; Buller et al., 2012), atorvastatin (R. L. Zhang et al., 2012; J; Chen et al., 2008), angiopoietin 2 (X. S. Liu et al., 2009), and erythropoietin (L. Wang et al., 2008). These findings provided a solid theoretical basis for the rapid development of stem cells in stroke. Afterwards, Chopp, M. and his team shifted the attention to MSC-derived exosomes and described how they contribute to immunological reactivity, vascular remodeling, and brain regeneration during stroke recovery. In addition, they provided an overview of the potential and perspectives of stem cells in the fields of stroke and regenerative medicine, the links between stem cells and inflammatory factors for stroke rehabilitation, and the prospects of exosomes in the field of stroke. In the recent 2 years, Chopp et al. gradually focused on and studied the application of extracellular vesicles in neurological diseases, indicating that the extracellular vesicles may perhaps become a new research hotspot in this field.

The most commonly co-cited author is Chen, Jl. (citation = 2,214), followed by Jin, K.L. (citation = 1,473), and Li, Y. (citation = 1,367). In 2006, Chen, Jl. investigated the effects of giving human BMSCs (hBMSCs) intravenously after intracerebral hemorrhage (ICH) in rats and found that doing so dramatically improves neurological function (Seyfried et al., 2006). This paper provided the basis for the clinical investigation of BMSCs in ICH. The next year, Chen, Jl. demonstrated neurological recovery in rats intravenously injected with hBMSCs 1 month following a stroke (Shen et al., 2007b) and published the first 1-year follow-up report of BMSC therapy in stroke rats (Shen et al., 2007a). This report proved that BMSCs have an effect on scarring reduction and cell proliferation increase. In 2013, Chen, Jl. concluded that while the effect of multiple injections did not outperform single-injection therapy in terms of functional outcomes and histological evaluations, it significantly improved long-term functional outcomes following stroke (Shehadah et al., 2013). In 2014, in a review article published in Progress in Neurobiology, Chen, Jl. described the application of stem cells from various cell origins in stroke as well as the restorative mechanisms, distribution methods, and imaging methodologies, which also covered the difficulties in stem cell therapy converting to clinical applications (X. Liu et al., 2014). The aforementioned studies generally focus on the mechanisms and therapeutic benefits of stem cell treatment for stroke, indicating that the field is still in the development stage and that additional basic and clinical translational research are still required. Clearly, the achievements of Chen, Jl. have laid a theoretical and experimental basis for research on stem cells in stroke.

A reference is deemed to be co-cited if it is cited in a number of different publications, in which case the co-cited references could be viewed as the foundation of the field’s study. To determine the research foundation for stem cells in stroke, we selected 15 references with the largest number of co-citations for this bibliometric analysis. Among the top 15 co-cited publications, two were written by Caplan et al., the first of which was the most often mentioned study and was published in the Journal of Cellular Biochemistry in 2006. This study firstly showed the trophic of the MSC-secreted bioactive molecules and summarized the application of the MSC trophic effect in injured tissues (Caplan and Dennis, 2006). The biological mechanisms of the in vivo functionality of MSCs during development and aging were outlined in another study (Caplan, 2007). The second most co-cited study was written by Chamberlain, G. et al., in 2011 and published in STEM CELLS. Their discovery that the movement of MSCs from the circulation into tissues may be facilitated by chemokine receptors and adhesion molecules attracted interest in terms of the function of stem cells in immunological regulation. Of the top 15 co-cited references (Abrous, Koehl, and Le Moal, 2005; Ohab et al., 2006), two were about the neural regeneration function of stem cells, outlining the molecular mechanisms between stem cells and neurogenesis and suggesting that stem cell therapy may be a therapeutic strategy for nervous system diseases. Three references (Schmidt-Lucke et al., 2005; Falk, 2006; Dutta et al., 2012) demonstrated that stem cells can repair blood vessels and promote neoangiogenesis. Overall, the biological function, transplantation, components, and targeted delivery of stem cells are the main subjects of discussion in the top 15 co-cited references, which represent the research foundation of stem cells in stroke.

Citation bursts refer to references that have received a lot of recent citations from other scholars and highlight emerging themes within a specific research field. In accordance with the key research topics of the references with the strongest citation bursts (Figure 7C), the potential mechanism of directed migration and neurogenesis of NSCs as well as the therapeutic effects of stem cells in stroke are currently the main areas of study for research on stem cells in stroke. In addition to references with citation bursts, keywords can facilitate swift capturing of the distribution and development of hotspots in the realm of research on stem cells in stroke. Combining the citation bursts and keywords, we prepare to divide the hotspots and frontiers into two main research areas, namely, the mechanistic research hotspots and the clinical research hotspots, to discuss the distribution of their respective hot spots according to Table 6 and Figures 8A, B.

This study has several advantages. First, based on research published from 2004 to 2022, this bibliometric analysis is the first investigation of patterns and contentious topics relating to stem cells in stroke. Second, we used three bibliometric methods simultaneously for the survey and analysis in this study, which significantly increased the likelihood that our data analysis process is impartial; VOSviewer and CiteSpace were extensively used in this study (C. Chen et al., 2012). This study also involved a thorough analysis of the number and growth tendency of annual publications; relationships among journals, authors, nations, and institutions; and various references, citations, and keywords.

This study also has limitations that need to be acknowledged. First, because this study only included articles in English, non-English writings may have been underrepresented. Second, only data obtained from the WoSCC database were used in this study; therefore, some pertinent studies from other databases may have been missed. Furthermore, insufficient data prevented full inclusion of articles in 2022.