Artificial intelligence in neurosurgery: A patent bibliometric analysis of the Lens database

Abstract

Background:

Artificial intelligence (AI) has rapidly expanded across neurosurgery with applications in neuroimaging analysis, surgical planning, intraoperative guidance, neuromodulation, and outcome prediction. Bibliometric analyses of scholarly works cited by patents provide a unique method for identifying research that has influenced technological development. This study aimed to identify and characterize the most patent-cited scholarly publications related to AI in neurosurgery.

Methods:

A bibliometric analysis was performed using the Lens database, which links global patent records with scholarly literature. A systematic search using predefined neurosurgical and machine learning terms yielded 72,719 publications, filtered to 4663 works cited by patents spanning 1962–2025. Publications were ranked by patent citation count as a proxy for technological influence. The 50 most patent-cited publications were analyzed by publication year, journal, institutional affiliation, and geographic origin.

Results:

The top 50 publications were published between 1994 and 2020. Most (41 of 50; 82%) were published between 2010 and 2020, reflecting rapid growth in AI-related neurosurgical research. Earlier influential works focused on image-guided surgery and augmented reality visualization, whereas more recent publications increasingly involved machine learning-based neuroimaging analysis, intraoperative diagnostics, neuromodulation optimization, and spine surgery applications.

Conclusion:

AI-related innovation in neurosurgery has accelerated substantially over the past decade. Patent citation analysis highlights the importance of interdisciplinary collaboration in translating computational advances into neurosurgical technologies and provides insight into research most closely associated with technological innovation in the field.

INTRODUCTION

Artificial intelligence (AI) and machine learning technologies have rapidly expanded across multiple domains of medicine, showing increasing applications in diagnostics and clinical decision support. In neurosurgery, AI-driven approaches have been applied to a wide range of problems, including neuroimaging analysis, prediction of clinical outcomes, intraoperative guidance, neuromodulation optimization, and robotic-assisted surgery.[7,9,10] These computational techniques have the potential to enhance surgical precision, improve diagnostic accuracy, and facilitate data-driven decision-making across neurosurgical subspecialties such as neuro-oncology, spine surgery, cerebrovascular surgery, and functional neurosurgery.[7,9,10]

As the volume of research in this field has grown, bibliometric analyses have become an increasingly valuable tool for evaluating scientific impact and identifying influential research trends. Bibliometric studies examine patterns in scientific publications, citations, and authorship to better understand how knowledge evolves within a particular field. Conventionally, bibliometric analyses have focused on scholarly citation counts as a measure of academic influence. However, an alternative and complementary approach involves examining the relationship between scientific publications and patents.[2] Because patents often cite foundational scientific studies that contribute to technological development, analysis of scholarly works cited by patents can provide unique insight into research that has influenced real-world technological innovation.

Within neurosurgery, bibliometric analyses have been performed to identify influential publications across various subspecialties. However, relatively few studies have specifically examined the intersection of neurosurgical research and technological innovation as reflected in patent literature.[3,6] This gap is particularly relevant in the context of AI, where many advances are closely tied to the development of new computational tools, diagnostic systems, and surgical technologies. Identifying the scholarly works that have most strongly influenced patented technologies may provide insight into the types of research that drive innovation within the field.

The objective of the present study was, therefore, to perform a bibliometric analysis of scholarly publications related to AI in neurosurgery that has been cited by patents. Using the Lens database, we identified and analyzed the 50 most patent-cited scholarly works in this domain. By examining temporal trends, journal and institutional contributions, and geographic distribution of influential publications, this study aims to characterize the academic research that has most strongly contributed to technological innovation in neurosurgery.

MATERIALS AND METHODS

Study design and data source

This study was designed as a bibliometric analysis of scholarly publications related to AI in neurosurgery that has been cited by patents. Patent and scholarly publication data were obtained using the Lens platform (Canberra, Australian Capital Territory, Australia), which is an integrated research database that links global patent records with scholarly literature.[5] The Lens platform aggregates patent documents from major international patent authorities, including the United States Patent and Trademark Office, European Patent Office, and the World Intellectual Property Organization, along with additional national patent offices worldwide.[5]

These patent records are cross-referenced with peer-reviewed scholarly publications indexed across multiple academic databases, including PubMed, CrossRef, and ORCID. Both active and expired patents are included, allowing a comprehensive evaluation of technological development over time.

The Lens platform provides detailed metadata for both patents and scholarly works, including citation relationships between patents and academic publications, patent-to-patent citations, authorship information, inventors, applicants, publishing journals, and institutional affiliations. This integrated structure enables both traditional scholarly bibliometric analyses and analyses examining the relationship between scientific research and technological innovation. The Lens database has previously been used to investigate influential patents in cranial neurosurgery as well as scholarly works cited by patents in spine surgery.[3,6]

Search strategy

A systematic search was conducted within the Lens platform to identify scholarly publications related to applications of AI in neurosurgery. The final search query was:

([“Neurosurgery” OR “neurosurgical” OR “spine surgery” OR “spinal surgery” OR “cranial surgery” OR “brain surgery” OR “intracranial”]

AND [“artificial intelligence” OR “AI” OR “machine learning” OR “deep learning” OR “neural network” OR “convolutional neural network” OR “CNN” OR “computer vision” OR “large language model” OR “LLM” OR “predictive model” OR “automated detection” OR “radiomics”]).

This search strategy was designed to capture a broad range of AI applications within neurosurgery while minimizing inclusion of unrelated publications that might result from overly general terms such as “algorithm” or “neural network.”

Study selection

The initial search yielded 72,719 journal articles. These results were subsequently filtered within the Lens platform to include only scholarly works cited by patents, resulting in 4663 publications. Publication years of the resulting papers ranged from 1962 to 2025.

For each paper, bibliometric data were extracted, including title, authors, publication year, journal, institutional affiliation, country of origin, total scholarly citation count, and patent citation count. The resulting publications were ranked in descending order according to the number of patent citations, which served as a proxy measure of technological influence.

From this ranked list, the 50 scholarly works with the highest number of patent citations were identified. Final inclusion of these publications was determined by three independent reviewers. Studies were included if they explicitly involved AI or machine learning-based methodologies applied to neurosurgical problems, including but not limited to surgical planning, intraoperative guidance, neuromodulation, surgical robotics, neuroimaging analysis, and clinical outcome prediction. Publications focused solely on general radiology, nonsurgical neurology, hardware development without algorithmic components, or nonclinical AI applications were excluded.

Bibliometric analysis

The top 50 most patent-cited publications in neurosurgery and AI were analyzed according to multiple bibliometric characteristics, including publishing journal, institutional affiliation, and geographic origin of the publishing institution (both by U.S. state and by country). These groupings allowed comparative analysis of both patent citation counts and scholarly citation counts across journals, institutions, and geographic regions.

To evaluate temporal trends, publications within the top 50 list were stratified by decade. For each decade, the total number of publications, the average number of patent citations per year, and the normalized average citation rate were calculated. Stratification by decade also enabled the identification of the three scholarly works with the highest patent citation counts within each decade.

In addition, journals represented in the top 50 most patent-cited list were ranked according to (1) the total number of publications appearing in the list and (2) the cumulative number of scholarly citations received by those publications. Similar analyses were performed for publishing institutions, with the top ten institutions ranked according to both publication count and total scholarly citations.

Finally, geographic analyses were conducted to determine the distribution of influential publications by region. For the top 50 most patent-cited works, the total number of publications originating from each geographic region was calculated, along with the total number of scholarly citations received by those publications. These analyses were performed at both the global level and across regions within the United States.

Statistical analysis and data visualization

All data cleaning, statistical analyses, and graphical visualizations were performed in R (version 4.5.2). The following packages were used: Dplyr for data manipulation; tidyr for data reshaping; ggplot2 for figure generation; knitr and kableExtra for table formatting; maps, sf, rnaturalearth, stringr, and mapproj for geographic mapping and spatial visualization; and htmltools, webshot2, and base64enc for HyperText Markup Language (HTML) rendering and image export.

Bar charts were generated using the geom_col() function. Geographic choropleth maps were created using geom_ polygon() for United States regional maps and geom_sf() for global maps. Individual figure panels were exported using ggsave() at 300 dpi, and multipanel figures were assembled as HTML layouts and exported as combined PNG files at 300 dpi through webshot2. Tables were formatted using knitr and kableExtra and exported as Microsoft Word documents (.docx) for manuscript preparation.

RESULTS

Overall search results and top 50 selection

The initial search of the Lens database identified 72,719 scholarly publications related to AI and neurosurgery. When filtered to include only scholarly works cited by patents, 4663 publications remained. These publications spanned from 1962 to 2025, reflecting several decades of technological development at the intersection of AI and neurosurgical practice.

All retrieved publications were ranked according to the total number of patent citations, which served as a proxy measure of technological influence. After ranking the dataset in descending order of patent citations, the 50 most patent-cited scholarly works were identified. Following an independent review by three investigators to confirm relevance to AI – based neurosurgical applications, these publications were included for further bibliometric analysis. The resulting 50 most patent-cited scholarly works included for bibliometric analysis are presented in Table 1.

Table 1:

Top 50 Scholarly Works Cited in Patents – Neurosurgery and AI. Categorized by the year, Number of Citations in Patents. The Works are Ranked by the Number of Citations in Patents and the Number of Citations in Patents per year Published.

Subsequent analyses examined patterns among the top 50 most patent-cited works, including temporal publication trends, journal and institutional contributions, and geographic distribution of publishing institutions. These analyses aimed to characterize the academic research most strongly associated with technological innovation in neurosurgery.

Temporal trends in patent-centered AI neurosurgery publications

To evaluate temporal patterns in influential AI research within neurosurgery, the top 50 most patent-cited publications were stratified by decade. Publication years of these works ranged from 1994 to 2020, with a marked increase in influential publications in more recent decades.

Among the top 50 works, 41 publications (82%) were published between 2010 and 2020, compared with 5 publications (10%) between 1994 and 1999 and 4 publications (8%) between 2000 and 2009 [Table 2]. When adjusted for the number of years within each period, the 2010–2020 decade demonstrated the highest normalized average number of patent-cited publications per year, reflecting the rapid expansion of AI applications in neurosurgery during the past decade.

Table 2:

Distribution of the top 50 most patent-cited neurosurgery and artificial intelligence publications across decades from 1994 to 2020, including the average number of top 50 publications per year per decade. 2010–2020 contained the greatest number of the top 50 most patent-cited publications. After normalizing by the total number of years since the initial publication through 2026, the 2010–2020 decade also had the highest normalized average.

Analysis of the most patent-cited publications within each decade further illustrates the evolving technological priorities of the field. The 1990s were characterized by foundational work in image-guided surgery and augmented reality systems, including the highly influential review by Azuma (1997), which described early frameworks for augmented reality and has since been cited extensively in patents related to surgical visualization and navigation technologies. Other influential works from this period focused on early image registration and computer-assisted surgical guidance systems.

During the 2000–2009 decade, influential publications increasingly focused on advanced imaging registration, correction of brain shift during surgery, and early applications of machine learning for clinical outcome prediction. These studies helped establish computational frameworks for integrating imaging data into neurosurgical decision-making and intraoperative navigation systems.

The 2010–2020 decade demonstrated a substantial expansion in both the number and diversity of influential publications. Highly cited works during this period included studies involving AI applications in spine surgery research, optimization of deep brain stimulation programming, intraoperative tumor detection using optical and spectroscopic techniques, and deep learning-based neuroimaging analysis. These publications reflect the growing role of machine learning and deep learning methods across multiple neurosurgical subspecialties, including neuro-oncology, functional neurosurgery, spine surgery, and cerebrovascular imaging. The three most patent-cited publications within each decade are summarized in Table 3.

Table 3:

The topmost scholarly works cited by patents stratified by time periods in the neurosurgery and artificial intelligence literature from 1994 to 2020. The works are ranked by the total number of citations per decade. The 1994–1999 decade contained the highest-cited individual papers, led by Azuma’s augmented reality survey with 1,148 patent citations.

Overall, these findings demonstrate a clear acceleration in the influence of AI-related research on neurosurgical technological development, particularly during the past decade. The increasing number of scholarly works cited by patents suggests that recent advances in machine learning, computational imaging, and intraoperative diagnostic technologies are playing a progressively larger role in shaping innovation within the field.

Journal and institutional contributions

The distribution of influential publications across journals and academic institutions was evaluated among the 50 most patent-cited scholarly works related to AI in neurosurgery [Figure 1].

Figure 1:

(a) Institutions ranked by number of unique citing patents. (b) Institutions ranked by the number of unique citing scholarly works. Data show the broader research and commercial impact of contributions from leading academic and medical centers.

Several journals contributed multiple publications to the top 50 list, reflecting the interdisciplinary nature of research at the intersection of AI, biomedical engineering, and neurosurgery. Journals spanning neurosurgery, medical imaging, computer science, and biomedical engineering were represented, highlighting the cross-disciplinary collaboration that characterizes the development of AI-based neurosurgical technologies. Among these, a small number of journals accounted for a disproportionately large share of the most patent-cited publications, suggesting that certain venues serve as major platforms for research that ultimately influences technological innovation.

When journals were ranked according to the total number of scholarly citations accumulated by their publications within the top 50 list, several high-impact journals emerged as particularly influential. These journals frequently publish foundational studies in medical imaging, machine learning, and computational neuroscience, which are often subsequently translated into technological developments reflected in patent literature.

Institutional analysis demonstrated that a limited number of academic centers contributed multiple publications to the top 50 most patent-cited works. These institutions represent leading research hubs in neurosurgery, biomedical engineering, and computational medicine. When ranked by the total scholarly citations associated with their publications, several institutions demonstrated particularly strong influence, reflecting both high publication impact and substantial downstream technological translation.

Taken together, these findings highlight the important role of interdisciplinary academic environments in advancing AI applications within neurosurgery. Institutions with strong collaborations between neurosurgeons, engineers, and computer scientists appear to contribute disproportionately to research that ultimately influences technological development in the field.

Geographic distribution

The geographic distribution of institutions contributing to the 50 most patent-cited scholarly works related to AI in neurosurgery was analyzed at both the global level and within the United States [Figure 2]. Globally, influential publications were concentrated within a limited number of countries. The United States contributed the largest proportion of publications within the top 50 list. Additional contributions originated from institutions across Europe and Asia, reflecting the increasingly international nature of AI research applied to neurosurgical practice. Several countries demonstrated multiple highly cited publications, highlighting the presence of established research ecosystems integrating neurosurgery, biomedical engineering, and computer science.

Figure 2:

(a) A map of the United States depicting the regional breakdown of all published scholarly works related to artificial intelligence in neurosurgery. (b) A map of the United States depicting the regional breakdown of the number of times that the top 50 scholarly works (most cited by patents) from that region were cited by other scholarly works. (c) A map of the world depicting the regional breakdown of all published scholarly works related to artificial intelligence in neurosurgery. (d) A map of the world depicting the regional breakdown of the number of times that the top 50 scholarly works (most cited by patents) from that region were cited by other scholarly works.

Within the United States, influential publications were similarly concentrated in specific academic regions. Institutions located in major academic and technology centers accounted for the majority of contributions to the top 50 list. These regions are home to universities and medical centers with strong interdisciplinary collaboration between clinical neurosurgery departments and computational research groups.

When evaluated according to the total number of scholarly citations associated with publications from each region, similar geographic patterns were observed. Regions contributing a greater number of publications to the top 50 list also tended to accumulate higher overall scholarly citation counts, suggesting that these academic environments produce research with both strong scientific influence and downstream technological impact.

Overall, the geographic distribution of the most patent-cited AI neurosurgery publications demonstrates that technological innovation in this field is concentrated within a relatively small number of global research hubs. These hubs appear to provide the collaborative infrastructure necessary to translate advances in AI and computational methods into neurosurgical technologies reflected in the patent literature.

DISCUSSION

Background on bibliometrics and patent citation analysis

Bibliometric analyses are widely used to evaluate scientific influence and identify emerging trends within a research field. Conventionally, these studies rely on scholarly citation counts to assess the academic impact of individual publications. While citation counts provide insight into how frequently a study influences subsequent scientific work, they do not necessarily capture the extent to which research contributes to technological development.[4] Analyses examining scholarly works cited by patents offer a complementary perspective by identifying academic studies that have directly informed the development of new technologies.

This approach is particularly relevant in fields where technological innovation plays a central role in advancing clinical practice. In neurosurgery, many transformative developments, such as surgical navigation systems and neuromodulation technologies, have emerged from interdisciplinary collaboration between clinicians, engineers, and computer scientists. Because patents often cite foundational scientific studies that inform the development of these technologies, bibliometric analysis of patent-cited scholarly works provides a unique opportunity to identify research that contributes to the translation of scientific discovery into practical technological applications. Measures of technological influence are distinct from clinical impacts of scholarly work, which are better assessed by metrics such as the rates of clinical adoption, measures of patient outcomes, and complication rates. As the majority of research and innovation in AI in neurosurgery is recent and is largely validated in retrospective or single-center studies, a gap exists between the technological advancement and its implementation in clinical practice. Acknowledging this gap, an analysis of technological influence through patent citations may therefore offer a more appropriate and relevant measure of research impact at this time. This approach is seldom taken within neurosurgery, with only two previously identified papers examining patent citations.[3,6] Examining these patterns may therefore offer insight into the types of academic contributions most closely associated with innovation in neurosurgical technology.

Evolution of AI in neurosurgery

The temporal distribution of the most patent-cited publications identified in this study demonstrates a clear acceleration of AI-related innovation in neurosurgery during the past decade. Most influential publications were produced between 2010 and 2020, reflecting the rapid maturation of machine learning and deep learning methodologies during this period. The widespread availability of large medical imaging datasets, coupled with increased computational power and advances in algorithm development, has likely contributed to the expansion of AI-driven research across multiple neurosurgical subspecialties.[1]

Earlier influential publications from the 1990s and early 2000s primarily focused on foundational computational technologies that enabled computer-assisted surgery. These studies frequently involved advances in image registration, augmented reality visualization, and image-guided navigation systems, which established the computational frameworks necessary for integrating digital imaging data into surgical workflows. In contrast, influential publications from the 2010 to 2020 decade increasingly reflect the application of machine learning techniques to clinical neurosurgical problems, including automated neuroimaging interpretation, intraoperative tumor detection, neuromodulation optimization, and predictive modeling of surgical outcomes. Together, these findings could suggest that AI innovation in neurosurgery has evolved from early computational infrastructure supporting surgical navigation toward increasingly sophisticated data-driven systems capable of assisting with diagnosis, surgical planning, and intraoperative decision-making.

Interdisciplinary research ecosystems driving innovation

The distribution of the most patent-cited publications across journals and institutions highlights the importance of interdisciplinary collaboration in advancing AI technologies within neurosurgery. Influential publications identified in this study were distributed across journals spanning neurosurgery, biomedical engineering, medical imaging, and computer science. This distribution reflects the inherently multidisciplinary nature of AI-driven innovation, where advances often arise from collaboration between clinicians with domain expertise and researchers specializing in computational methods and engineering.[8]

Institutional analysis similarly demonstrated that a relatively small number of academic centers contributed multiple publications to the most patent-cited list. These institutions are typically characterized by strong collaborations between departments of neurosurgery, engineering, computer science, and biomedical informatics. Such academic environments provide the infrastructure necessary to translate advances in machine learning and computational imaging into clinically relevant technologies. Institutions that integrate clinical data resources with advanced computational research programs may be uniquely positioned to develop and validate AI tools that ultimately influence technological innovation and device development reflected in patent literature. These findings underscore the role of interdisciplinary academic ecosystems in facilitating the translation of AI research into neurosurgical technologies.

Technological domains driving innovation in AI neurosurgery

Analysis of the most patent-cited publications also highlights several technological domains in which AI research appears to be most strongly associated with innovation in neurosurgery. A substantial portion of influential publications involved advanced neuroimaging analysis, including machine learning-based interpretation of magnetic resonance imaging and computed tomography imaging data. These approaches have enabled automated detection of intracranial pathology, improved diagnostic classification of tumors and vascular lesions, and enhanced imaging-based surgical planning.

Another prominent category of influential publications involved intraoperative diagnostic and visualization technologies. Several highly patent-cited studies described optical or spectroscopic techniques capable of identifying tumor tissue during surgery, reflecting growing interest in integrating real-time computational analysis into the operative environment. In addition, publications related to neuromodulation and deep brain stimulation optimization represented an important area of technological development, with machine learning approaches increasingly applied to improve programming and therapeutic targeting.

Finally, AI applications in spine surgery research and predictive modeling of surgical outcomes also appeared among the most influential publications. Together, these patterns suggest that innovation in AI neurosurgery is occurring across multiple subspecialties and shows particular emphasis on technologies that enhance diagnostic precision, intraoperative decision-making, and treatment optimization.

Limitations

Several limitations should be considered when interpreting the findings of this study. First, the analysis relied on the Lens database, and although Lens integrates a large number of patent records and scholarly publications, it may not capture all patents or citation relationships across every international jurisdiction. Second, bibliometric analyses based on citation counts are inherently influenced by temporal effects, as older publications have had more time to accumulate both scholarly and patent citations. While stratification by decade was performed to partially address this limitation, citation dynamics may still favor earlier publications. Third, the use of patent citation counts as a proxy for technological influence does not necessarily reflect clinical impact or adoption in surgical practice. Some technologies may influence neurosurgical care without being associated with patent activity. Finally, the field of AI in neurosurgery is evolving rapidly, and newer publications may not yet have had sufficient time to accumulate patent citations. Consequently, influential recent work may be underrepresented in the present analysis.

CONCLUSION

This bibliometric analysis identified the scholarly publications in AI and neurosurgery that have most strongly influenced technological innovation as reflected in patent literature. The findings demonstrate a marked acceleration of influential AI-related research during the past decade, with the majority of highly patent-cited publications emerging after 2010. Early influential works primarily focused on computational frameworks supporting image-guided surgery and augmented reality visualization, whereas more recent studies increasingly involve machine learning-based neuroimaging analysis, intraoperative diagnostic technologies, neuromodulation optimization, and predictive modeling. The distribution of influential publications across journals, institutions, and geographic regions further highlights the importance of interdisciplinary collaboration in advancing AI-driven neurosurgical technologies. By identifying research that has contributed to technological development, this study provides insight into the academic foundations shaping future innovation in neurosurgery.

Ethical approval:

Institutional review board approval is not required. This bibliometric analysis utilized publicly available data and does not involve human subjects or patient data. No IRB approval, waiver number, or patient consent was required.

Declaration of patient consent:

Patient’s consent is not required as there are no patients in this study.

Financial support and sponsorship:

Nil.

Conflicts of interest:

There are no conflicts of interest.

Use of artificial intelligence (AI)-assisted technology for manuscript preparation:

The authors confirm that there was no use of artificial intelligence (AI)-assisted technology for assisting in the writing or editing of the manuscript, and no images were manipulated using AI.

Disclaimer

The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Journal or its management. The information contained in this article should not be considered to be medical advice; patients should consult their own physicians for advice as to their specific medical needs.