Imaging Diagnosis of Mild Cognitive Impairment: A Bibliometric Analysis (1999–2024)

This bibliometric analysis utilized data from the Web of Science Core Collection (WoSCC, Clarivate Analytics), covering publications from January 1, 1999, to December 31, 2024. The WoSCC provides access to high‐impact journals across disciplines, forming a robust foundation for analyzing trends in imaging diagnosis for MCI. To ensure specificity and minimize noise, the search strategy included both inclusion and exclusion criteria. The inclusion criteria were (1) English‐language publications, (2) indexed in the SCIE or SSCI databases, and (3) publication type restricted to articles. The exclusion criteria were (1) non‐English publications, (2) proceeding papers, retracted publications, book chapters, data papers, and (3) studies unrelated to MCI imaging (e.g., animal or in vitro studies, materials science, or unrelated imaging modalities). The revised search formula was TS = (((“mild cognitive impairment” or “amnestic MCI” or “aMCI” or “non‐amnestic MCI” or “naMCI”) and (MRI or PET or CT or SPECT or EEG or fMRI or “Abeta PET” or “amyloid PET” or “Tau PET” or “neuroimaging”) and (diagnos* or predic* or prognostic* or screen* or detect*))) not (animal or cell or mouse or rat or in vitro or materials science or physics or engineering or dichroism). The final search was conducted on October 14, 2025, and retrieved 8898 publications. After applying the inclusion and exclusion criteria, 7568 eligible publications were retained for analysis, as illustrated in the updated data screening flowchart (Figure 1).

The data extracted from the selected publications included titles, keywords, publication years, countries/regions, authors, institutions, journals, citation counts, and h‐index (Gao et al. 2017; Hirsch 2005; Synnestvedt et al. 2005). Data organization and basic statistical analyses were performed using Microsoft Excel. For visualizing collaboration networks, keyword co‐occurrence, and citation patterns, we employed three bibliometric tools: VOSviewer (version 1.6.20), CiteSpace (version 6.3.R1), and R‐bibliometrix (version 4.5.1). These tools provide complementary insights into research trends and collaboration dynamics within the field.

VOSviewer was used to construct co‐authorship and co‐citation networks, as well as to perform keyword co‐occurrence analyses. The parameters for each visualization are annotated in the respective figures to ensure methodological transparency. Specifically, the first two parameters in VOSviewer figures represent the number of included nodes and the minimum number of publications required for inclusion. The third and fourth parameters, “attraction” and “repulsion,” were adjusted to optimize the visualization layout, ensuring clarity and minimizing overlap between nodes. CiteSpace was employed for keyword burst analysis, with the following parameters: time slicing from 1999 to 2024 (1‐year intervals); node types set to “keywords”; and thresholds for the top 5 nodes per time slice. For pruning, the pathfinder and merged network pruning methods were applied to ensure clarity and relevance. R‐bibliometrix was used for an integrated analysis of bibliometric data, providing insights into publication trends, journal impact, and emerging research hotspots. The choice of parameter thresholds for CiteSpace and VOSviewer was guided by established practices in bibliometric research (Liu et al. 2022) and optimized to balance sensitivity and specificity. Although a formal sensitivity analysis was not performed, the stability of keyword clusters was ensured by iterative testing. Additionally, journal impact factors (IF) were reported based on JCR 2024 quartiles, reflecting the most up‐to‐date metrics at the time of analysis. Key bibliometric indices, including the h‐index and g‐index, were used to evaluate the productivity and citation impact of authors and journals. The h‐index reflects the number of articles (h) that have been cited at least h times, while the g‐index gives more weight to highly cited articles by ensuring the top g articles collectively receive at least g² citations (Gao et al. 2017; Hirsch 2005; Synnestvedt et al. 2005).

In this bibliometric study, we conducted a comprehensive analysis of the scientific literature on MCI, with a focus on imaging diagnostics (Figure 1). A total of 7568 scholarly articles published between 1999 and 2024 were included after screening. These studies were authored by 26,863 researchers affiliated with 784 institutions across 104 countries and regions. Collectively, these articles were published in 784 sources and cited a total of 131,726 references, demonstrating the depth of collaboration and research in this field (Figure 2a). The dataset also includes a total of 9,448 distinct keywords, reflecting the diversity of topics explored. The annual growth rate of publications in this field was 22.27%, indicating a significant increase in research activity over time (Figure 2b). The number of annual publications remained relatively low before 2007, with fewer than 100 articles published yearly. However, a sharp increase was observed starting in 2008, with 165 articles, and this upward trend continued in subsequent years. Notable peaks include 548 publications in 2015, followed by a significant surge to a record high of 762 articles in 2024, marking a culmination of the field's growth. A slightly lower publication count of 616 articles in 2023 was followed by this unprecedented peak in 2024 with 7568 articles, reflecting the field's strong momentum. This growth trajectory, along with a high average citation rate of 49.97 citations per document, underscores the field's academic and clinical importance.

In the present study, we identified the most influential literature within the field by examining highly cited articles. Notably, the most influential is the article titled “Current concepts in mild cognitive impairment” published in Arch Neurol (Petersen et al. 2001), garnering an impressive 3,838 citations. Following this, the article “Lecanemab in Early Alzheimer's Disease”, from N Engl J Med, has also made a substantial contribution, amassing 3,073 citations (van Dyck et al. 2023; Jack et al. 2018). Additionally, “Neuropathological alterations in Alzheimer disease”, published in Cold Spring Harb Perspect Med, has been cited 2505 times (Serrano‐Pozo et al. 2011).

The updated analysis of publication and citation profiles for leading countries in the field of MCI imaging diagnostics (Table 1 and Figure 3a) underscores substantial contributions from several nations. The USA remains the most prolific country, with 1,961 articles, accounting for 25.9% of the total output. It ranks first in both total publications (TP = 12,535) and total citations (TC = 158,671), with an impressive average of 80.9 citations per article, reflecting its strong academic influence and productivity. The USA also has the highest share of single‐country publications (SCP = 1472) and demonstrates notable international collaboration (MCP = 489), with a multiple‐country publication ratio (MCP Ratio) of 0.249. China follows as the second‐leading contributor, with 1,316 articles (17.4%), 5,583 total publications, and 32,876 citations, ranking second in both total publications and citations. However, its average citations per publication (25) remain significantly lower than that of the USA, indicating room for growth in academic influence. China's MCP ratio of 0.277 highlights its moderate engagement in international collaborations. Italy ranks third, with 492 articles (6.5%), 2,992 total publications, and 17,892 citations, securing fifth place in total citations. Italy's average citations per publication are 36.4, and it demonstrates a strong focus on international collaboration, as reflected by an MCP ratio of 0.376. In the realm of international research collaborations on MCI imaging diagnostics (Figure 3b), the USA has the highest total link strength (2398), followed by the UK (2150) and Sweden (1,372).

The analysis of the most relevant affiliations in the field of MCI imaging diagnostics, as depicted in Figure 3c, highlights the significant contributions from top institutions. The University of California System emerges as the leading institution, with a total of 1762 articles, underscoring its dominant role in advancing research output and innovation in this field. Following closely, Vrije Universiteit Amsterdam ranks second, contributing 821 articles, which reflects its prominent position in MCI imaging research. The University of London is also a major contributor, publishing 784 articles, further emphasizing its influence in the global research landscape. In the landscape of international research collaborations (Figure 3d), with a threshold of at least 50 articles, San Francisco takes the lead in collaborative efforts, engaging in 876 total link strength with institutions across the globe. The Karolinska institution comes next with 683 collaborations, and University College London, with a notable 627 collaborations, follows closely behind.

A total of 26,863 authors have contributed to this research field. The analysis of high‐impact authors in the field of MCI imaging diagnostics, as detailed in Table 2, highlights several leading researchers whose work has significantly influenced the scientific community. The top‐ranked author is Clifford R. Jack Jr., with an h‐index of 80 and a g‐index of 159. He has published 192 total papers since 1999, which have collectively garnered 25,779 total citations (TC), making him one of the most impactful contributors in this field. His m‐index of 2.963 reflects his sustained high‐quality research output over time. Ronald C. Petersen follows closely, with an h‐index of 76 and a g‐index of 140. He has published 140 papers, accumulating the highest total citation count (27,786). Petersen's m‐index of 2.815 further emphasizes the consistent influence of his work since 1999. In the third position is Philip Scheltens, who has an h‐index of 65 and a g‐index of 123. Scheltens has contributed 160 papers to the field, which have received 15,620 citations, reflecting his significant impact. His m‐index of 2.407 highlights his steady research contributions since 1999.

In the realm of international research collaborations within the field of MCI imaging diagnostics, a select group of 108 authors, each with a minimum of 25 articles, stands out for their cooperative efforts (Figure 4). Prominently, Clifford R Jr. Jack leads in the extent of his international collaborations, with a total link strength of 671, indicating a robust network of partnerships across different countries. This is followed by Ronald C. Petersen, whose collaborations have a total link strength of 664, and David S. Knopman, with a link strength of 532.

The bibliometric analysis of high‐impact journals in the field of MCI imaging diagnostics identifies several key publications that have significantly contributed to the advancement of the field (Table 3). Neurology leads with the highest h‐index of 100, reflecting its substantial scientific influence, and boasts an IF of 8.5, placing it in the prestigious Q1 category of the 2024 JCR Quartile. It has published 236 total articles, ranking fifth in terms of TPs, and has accumulated 25,674 TC, ranking second overall. NeuroImage follows closely, with an h‐index of 90 and an IF of 4.5, also in the Q1 category. The journal has published 240 articles (TP), ranking fourth in total publications, and leads the field with 27,663 citations (TC), making it the most influential journal in terms of citation impact. Brain ranks third with an h‐index of 76, an IF of 11.7, and a Q1 ranking. It has published 110 articles, ranking twelfth in total publications, but its 10,906 citations secure it fifth place in total citations, reflecting its strong academic impact despite a smaller number of publications.

The analysis of co‐occurrence networks in the field of MCI imaging diagnostics reveals a robust framework of interconnected journals, with 133 journals having at least 15 occurrences (Figure 5a). The Journal of Alzheimer's Disease stands out with the highest total link strength, amassing 6838 connections. Neuroimage follows as a significant contributor, with a total link strength of 6397. Neurology also exerts considerable influence, evidenced by its total link strength of 5738. In the context of coupling networks, which measure the interconnectedness of journals based on co‐citation relationships, 140 journals are identified with a minimum of 10 connections (Figure 5b). The Journal of Alzheimer's Disease leads with an exceptional total link strength of 2,134,954, underscoring its pivotal position in the research landscape. Neurobiology of Aging and Frontiers in Aging Neuroscience also demonstrate their importance, with total link strengths of 1,050,705 and 925,376, respectively.

The co‐occurrence analysis of keywords in the field of MCI reveals a complex network of interconnected concepts that define the landscape of scientific inquiry (Table 4). The term “mild cognitive impairment” is the most prominent keyword, with 3,550 occurrences and a total link strength of 15,747, underscoring its foundational role in the field. Other significant keywords include “dementia” (2,609 occurrences, link strength 12,696) and “Alzheimer's disease” (2,379 occurrences, link strength 10,246), reflecting the close relationship between these conditions in the research landscape. The network visualization (Figure 6a) identifies four distinct clusters, each representing a thematic area of focus. These clusters, along with their summarized themes, are as follows: Cluster 1 (red): functional and cognitive networks. This cluster, comprising 45 items, focuses on functional and cognitive processes associated with MCI and AD. Key terms such as “connectivity,” “default‐mode network,” “working memory,” and “functional connectivity” highlight the emphasis on brain network dysfunction and cognitive decline. It also incorporates keywords like “age,” “depression,” and “risk factors,” which reflect the role of demographic and psychological variables in MCI progression. Cluster 2 (green): biomarkers and pathological changes. This cluster includes 31 items revolving around biomarkers and pathophysiological changes in MCI. Keywords like “amyloid‐beta,” “CSF biomarkers,” “tau,” and “glucose metabolism” underline the focus on structural and molecular markers. Additionally, terms such as “PET,” “cerebrospinal fluid,” and “frontotemporal dementia” emphasize imaging techniques and differential diagnosis. Cluster 3 (blue): structural imaging and diagnostic tools. This cluster, with 31 items, centers around structural imaging and diagnostic methodologies. Core keywords like “MRI,” “hippocampal atrophy,” “voxel‐based morphometry,” and “cortical thickness” reflect the use of advanced imaging techniques to study brain morphology. Terms such as “classification,” “segmentation,” and “prediction” indicate the integration of computational tools for automated diagnosis and risk assessment. Cluster 4 (yellow): guidelines and recommendations. This smaller cluster of 11 items focuses on clinical guidelines and diagnostic frameworks. Keywords such as “diagnostic guidelines,” “recommendations,” “criteria,” and “national institute” highlight the importance of standardization and consensus in MCI diagnosis.

The analysis of keywords with the strongest citation bursts (Figure 6b) further demonstrates the complex interrelationships among these keywords. During the late 2000s and early 2010s, the field saw a surge in interest in structural and molecular imaging markers, with strong bursts for keywords such as “brain atrophy” (burst strength: 40.48, 2006–2015), “CSF biomarkers” (37.02, 2010–2016), and “Pittsburgh compound B” (30.82, 2009–2015). These keywords reflect a period of intensive research focused on validating diagnostic criteria and understanding pathological changes in MCI. In recent years, however, there has been a dramatic shift toward computational and artificial intelligence‐related topics. Keywords such as “deep learning” (burst strength: 42.13, 2022–2024), “machine learning” (burst strength: 34.56, 2021–2024), and “convolutional neural network” (burst strength: 16.43, 2020–2024) have emerged as dominant themes.

This bibliometric study provides a thorough analysis of the scientific discourse on MCI, emphasizing imaging diagnostics. The examination underscores a robust collaborative framework, with a global network of authors and institutions contributing to a rich corpus of scholarly articles. The significant upward trend in the annual number of publications on MCI imaging diagnostics, with an annual growth rate of 29.05%, indicates expanding interest and advancements in this field. The peaks in publication counts, particularly in 2022 and 2024, may be attributed to rapid progress in technology, especially the integration of machine learning and deep learning techniques (Chaudhary et al. 2022; Illakiya and Karthik 2023). These technological advancements have significantly enhanced the diagnostic capabilities and research methodologies in MCI imaging diagnostics. The eminence of Neuroimage, Neurology, and JAMA Neurology within the MCI imaging diagnostics field is a testament to the interdisciplinary and profound research that defines this area of study (Barisano et al. 2022; Dean et al. 2014; Liew et al. 2023). These journals serve as pivotal platforms for disseminating knowledge, as evidenced by the extensive publication and citation of their articles. This broad academic engagement and the rigorous peer‐reviewed process they represent not only highlight the multifaceted nature of MCI research but also lay a robust groundwork for the evolution of diagnostic techniques and therapeutic strategies.

The analysis of publication and citation profiles across various countries and institutions reveals a dynamic and collaborative research landscape in the field of MCI imaging diagnostics. The USA, China, and Italy have emerged as leading contributors, reflecting a robust research infrastructure and a strong focus on advancing diagnostic methodologies. Institutionally, the University of California System, Harvard University, and the University of London stand out as major hubs of research activity, each contributing significantly to the collective knowledge base. The roles are further amplified by the collaborative research efforts, such as the meta‐analysis conducted by a network of 89 institutions, which underscores the critical role of cerebral amyloid‐β aggregation in the early stages of AD (Jansen et al. 2015). This comprehensive review, which synthesizes individual participant data from 55 studies, focuses on the prevalence of amyloid pathology as detected by positron emission tomography or cerebrospinal fluid, a key aspect of MCI imaging diagnostics. The analysis reveals a significant correlation between amyloid pathology and factors such as age, APOE genotype, and the severity of cognitive impairment (Jansen et al. 2015). The study's most striking insight is the identification of a 20‐ to 30‐year prodromal period between the initial detection of amyloid positivity and the onset of dementia, a finding that is pivotal for grasping the trajectory of AD and for the strategic planning of preventative interventions (Jansen et al. 2015). These collaborations are crucial for the exchange of ideas, resources, and expertise, which are essential for driving innovation and progress in the field. The integration of such collaborative findings into the broader research landscape, particularly in the context of MCI imaging diagnostics, exemplifies the power of global cooperation in advancing our understanding of neurological conditions and informing the development of effective therapeutic strategies.

The author analysis within the domain of MCI imaging diagnostics highlights the influential contributions of leading researchers, such as Clifford R Jr. Jack, whose work is exemplified by the study on “White matter hyperintensities are higher among early‐onset AD (EOAD) participants than their cognitively normal and early‐onset non‐AD peers” (Alzheimer's Dementia, 2023) (Eloyan et al. 2023). This research underscores the role of white matter hyperintensities (WMHs) in EOAD, revealing a significant association between WMH burden and cognitive impairment, as well as with amyloid and tau burden (Eloyan et al. 2023). Clifford's involvement in this study, which is part of the Longitudinal Early‐Onset Alzheimer's Disease Study (LEADS), showcases his commitment to advancing the understanding of EOAD and its pathological markers. In addition, the seminal work by Ronald C. Petersen, “Current concepts in mild cognitive impairment” (Arch Neurology, 2001), has been instrumental in shaping the field's understanding of MCI as a transitional state between normal aging and AD (Petersen et al. 2001). This article has been widely cited for its comprehensive summary of the diagnostic criteria, clinical outcomes, and the potential for therapeutic intervention in MCI. Petersen's early insights have been foundational in guiding subsequent research and clinical trials, which continue to explore the complexities of MCI and its progression to AD. Together, the contributions of these authors, as indicated by their high h‐ and g‐indices, reflect their substantial impact on the field. Their work has not only enriched the scientific discourse but also propelled the development of diagnostic tools and therapeutic strategies for MCI.

The co‐occurrence of “dementia” and “Alzheimer's disease” in the scientific literature underscores the well‐established trajectory from mild cognitive impairment (MCI) to advanced cognitive decline. MCI is widely recognized as a prodromal stage of AD, characterized by cognitive decline that exceeds normal aging but does not yet meet dementia criteria (Jongsiriyanyong and Limpawattana 2018; Davis et al. 2018). This highlights the urgency of understanding MCI's progression to AD and developing interventions to delay or halt this transition (Jongsiriyanyong and Limpawattana 2018). Early diagnosis remains critical, as reflected in the prominence of keywords like “MRI,” “biomarkers,” and “diagnostic guidelines,” which emphasize the importance of neuroimaging and molecular markers in identifying MCI and predicting the risk of AD (Feng and Ding 2020; Davis et al. 2018). The multidisciplinary nature of MCI research, involving neurology, psychiatry, geriatrics, and cognitive science, highlights the complexity of the condition and the need for a holistic diagnostic and therapeutic approach (Jongsiriyanyong and Limpawattana 2018).

This bibliometric analysis provides valuable insights into the research landscape of imaging diagnostics for MCI, aiding researchers and clinicians in identifying key contributors, influential publications, and emerging trends. By mapping active research areas, the findings can support the adoption of new diagnostic methodologies and inform clinical practice to improve patient care. Moreover, the integration of AI and advanced imaging technologies reflects the growing emphasis on precision diagnostics, which has the potential to revolutionize how MCI is detected and managed.

However, several limitations of this study should be acknowledged. First, the analysis was restricted to articles indexed in the WoSCC, which may exclude relevant studies from other major databases, such as Scopus or PubMed. This reliance on a single database introduces the risk of database selection bias, potentially leading to an incomplete representation of the field. Second, the focus on English‐language publications may result in language bias, underrepresenting high‐quality research published in other languages. Additionally, journal selection bias may be present, as WoS primarily indexes journals with higher impact factors, potentially overlooking significant contributions from regional or emerging journals. The exclusion of non‐article publication types, such as book chapters and editorials, may omit important perspectives or novel hypotheses. Furthermore, the keyword‐based approach to bibliometric analysis may overlook emerging trends that use alternative terminology or synonyms. The lack of sensitivity analysis in keyword clustering is another limitation, although iterative testing was performed to ensure stability. To address these limitations in future studies, integrating data from multiple databases and including broader language and publication‐type criteria would provide a more comprehensive analysis. Additionally, conducting sensitivity analyses and incorporating emerging terms into keyword searches would enhance the robustness and depth of bibliometric findings.