Bibliometric analysis of the top 100 cited articles in breast radiology

Objective: Bibliometrics analysis is a widely used approach that enables influential research within specific fields to be identified To identify the 100 most-cited articles in breast radiology and analyse the trend in breast imaging research. Methods and materials: A systematic search was conducted using the Thomson Rheuters Web of Science database. The results were ranked according to citation count and screened to create a single database. Data including first author, year of publication, journal, country of origin, primary institution, number of citations and average number of citations per year were extracted, as well as the impact factor and the 5-year impact factor of journals publishing the articles. Results: The systematic search yielded a total of 114,426 articles, after filters were applied to include papers that were available in English only. Citations for the 100 most-cited articles ranged from 515 to 3660. Half of the articles on the list were published between 2001 and 2010. Radiology has the most number of publications (n = 17), followed by JAMA-Journal of The American Medical Association (n = 9). CA-A Cancer Journal For Clinicians had the highest impact factor of 286.13. Mammogram (n = 49) was the most commonly studied modality, followed by Magnetic Resonance (n = 26). The most common topic of publication was diagnosis (n = 83). Conclusion: This research serves as a guide to the most influential articles on the topic of breast radiology.


INTRODUCTION
Bibliometrics is a widely used approach that enables influential research within specific fields to be identified. 1 It is based upon statistical methods, which allows the reviewer to evaluate published papers and assess core parameters such as research performance, impact, and the productivity of key authors in the field. 2 In so doing, this style of research can be effective in identifying past trends within an area, and in enabling potential future directions to be identified in the process. 3 Because evidence-based practice is such an underpinning force within the field of healthcare provision, 4 the use of bibliometrics can therefore be both influential and impactful within the delivery of care to patients that is both safe and effective. 5 The use of bibliometrics has a strong precedence in the field of radiology, with bibliometric analysis having been applied both in the field of radiology as a whole 6 and within more specific subfields of the profession. 7 The recent bibliometric analysis conducted by Oo and Chu, 8 which focused on radiology of the head and neck was effective at producing a novel set of results that could have both clinical and research-related impact, and this work shall provide the framework for how this current study is conducted and reported. There is, to the best of this author's knowledge, no bibliometric analysis that has been conducted on the issue of breast radiology, and so by providing a comprehensive assessment of this field, this current study shall seek to provide novel insights into this important area.

METHODS AND MATERIALS
A bibliometric analysis of the most highly cited articles in breast radiology was carried out in September 2022. A search was conducted within Thomson Reuters Web of Science (WOS) using the following key terms: This process returned a total of 145,396 articles. Filters were then applied to include full manuscripts that were available in English only (n = 114,426). The results were then sorted by the number of citations from the most-cited to the least-cited. This method was developed by Paladugu et. al 9 and was used in several other studies. Each article was manually screened for inclusion. Articles were included if they focused on diagnostic imaging, interpretation, imaging technique, utility and role of different imaging modalities or trends in breast radiology. Articles that focused on benign breast pathologies other than breast cancer were also included. Articles are excluded if they are not related to breast radiology, or if the primary focus was not on the breast. Articles that focus primarily on therapeutic radiology were excluded.
Data such as WOS citations, year published, first author, last author, primary institution, country of origin, journal, journal impact factor, title, study design, study focus and modality were collected for each article and assembled into a single database (As shown in Figure 1).

RESULTS
The top 100 most-cited articles has a mean citation number of 899.41 and a median of 772, as shown in Table 1. The number of citations ranged from 515 to 3660.

Citations per year
Citations per year ranged from 13.92 to 1830 with a mean of 84.74 and a median of 45.40 per year. The article by Rueckert et al 10

Most common first author
There was a total of 92 first authors on the top 100 list. Amongst them, Berg has the highest number of articles, with four articles. There were five authors who were first authors of more than one article to the top 100 list.

Journals
The top 100 articles were published across 41 journals ( Table 2). Radiology has the greatest number of publications (n = 17), followed by JAMA-Journal of The American Medical Association (n = 9). CA-A Cancer Journal For Clinicians had the highest impact factor of 286.13, the highest 5 year impact factor of 334.26 and the highest journal citation factor of 76.09, and it contributed to three articles on the list.

Country and institution of origin
The United States (USA) has the greatest number of publications on the list (n = 61), followed by the United Kingdom (UK) (n = 12). The results are displayed in Figure 3. In terms of the affiliated academic institutions of the first authors on the list, Stanford University and the University of Washington both have the highest number of articles, with four each.

Study design
The majority of the study designs was prospective (n = 60), followed by reviews (n = 28), retrospective (n = 10) and a mix of prospective and retrospective (n = 2) ( Table 3) Topic Imaging modality, primary topic of the article and if the article was breast cancer-related were also evaluated for all of the articles on the list-83 of them focused on the diagnosis of breast pathology and 90 papers are related to breast cancer (Table 3)

DISCUSSION
This bibliometric analysis offers a range of interesting insights into the field of breast radiology. These findings include, as presented above, insights into how this field has developed over the years, and what the core focuses have been over time.
From the analysis conducted as part of this current study, the most-cited article in this field was the work of Rueckert and colleagues, which was published in the journal IEEE Transactionson Medical Imaging in 1999. This article has, at the time of conducting this analysis, received n = 3660 total citations. This,  averaged across the 23 years since publication equates to approximately n = 159 citations being made of this paper every year; clearly indicating therefore that Rueckert and colleagues' work continues to have a lasting and influential impact on the field of breast radiology. It should be noted that this paper, as with the second most-cited article in this field-that of Ophir et al 11was published before the turn of the century. Because of the way that bibliometric analysis works, this can tend to bias earlier published work towards appearing more influential. 12 However, whilst this caveat should be borne in mind, the scale of influence that these papers had is incontrovertible.
The top three journals within this current analysis-Radiology; JAMA-Journal of the American Medical Association; and the New England Journal of Medicine-are not all journals which focus specifically on the field of radiology; indeed, only Radiology, with a total of n = 17 articles within this bibliometric analysis is focussed on this topic. This is a finding of interest because this does not support the findings of Oo and Chu, 8 who noted that in their analysis of head and neck radiology, the top three journals were all radiology-centric in nature. The reasons for the disparity between the findings of Oo and Chu 8 and those of this current bibliometric analysis may be the subject matter, with the issue of breast cancer, in particular, being considered more feasible or of interest to a wider, more general audience.
Next to be discussed is the journal impact factor of the journals within the top 100 list presented above. The impact factor of a journal is a blunt, yet influential approximation of the esteem and influence of that journal, 13 which is calculated by dividing the citations of papers in that journal across the period of a year by the number of articles published in the previous 2 years. 13 The higher the impact factor is, the more influential that journal is considered to be, 14 and it is notable that the impact factor of the above, top three journals ranges considerably. For example, the impact factor for the journal Radiology is 29.1, whilst the secondranked journal of JAMA-Journal of the American Medical  Association (impact factor = 157.3) and third-ranked journal of New England Journal of Medicine (impact factor = 176.1) are far more influential. This again may indicate why breast radiology researchers targeted these journals for publishing their studies, due to the greater reach and impact of such journals.
It is also notable that, regarding the location of study, the 83% of the included papers were published by American or British research teams, with the vast majority of these-n = 61-being based in America. This is indicative of the research culture across clinical teams in these countries, and the resources available to clinical and research teams to develop the evidence base. This finding is highly conducive with those of other bibliometric analyses published in the field of radiology. 15 From the analysis conducted as part of this current study, it is clear that the core focus in the field of breast radiology is breast cancer, with 90 papers within the selected 100 addressing this issue. This can potentially be related to the emotive nature of breast cancer 16 and the impact that this awareness has on research funding decisions and opportunities. 17 Whilst this is important in one sense-clearly it can be argued that the advances made in understanding breast cancer screening, identification and  Figure 3. Country of the first author for 100 top-cited articles in breast radiology Whilst nearly half of studies focussed on a primary modality of mammograms, this is indicative of practice in this field, although it should be noted that other core approaches such as PET or CT scans tend to be underrepresented within the top 100cited articles in this area, and more perhaps needs to be done to address this imbalance. It is also notable that n = 60 of these studies were prospectively designed; this is important to address because unlike in retrospectively designed work, such studies are able to limit the potential of biases and confounding variables to influence findings. 19 The rigour of these selected works is further underlined by the relatively high volume of review studies (n = 28). Well-conducted reviews can take into account important issues such as risk of bias within primary studies, 20 and as such are considered to be near the apex of the evidence-based pyramid within healthcare research. 21 As far as this author is aware, this is the first bibliometric analysis to have focused on the issue of breast radiology. This study has therefore produced a set of novel results that can provide researchers, radiologists, and other allied health professionals with information about the most impactful. Influential papers within this field, which in turn may inform both their approach to practice, and potential directions for future research projects. However, whilst this work is therefore likely to be influential in the above ways, it should be noted that there are a number of limitations associated with this study. It is important that these limitations are discussed in order for there to be a transparent discussion of the implications from this work. For example, it is possible that the search terms used incorporated some form of researcher bias into the search process. 8 It is also of note that articles which focussed on therapeutic radiology were excluded from consideration for this analysis, which may have biased the inclusion of papers within this work; a future bibliometric analysis into therapeutic radiology may help redress this imbalance. Finally, this bibliometric analysis utilised output from WOS, but as noted in the past bibliometric analysis, 8 citation data can differ considerably across core databases such as WOS, Google Scholar, PubMed or Scopus. However, whilst these limitations should be held in mind, it should also be noted that a central bias of this methodology-that towards earlier published paperswas addressed via the use of a citations per year metric, which is indicative of this work being rigorously conducted and transparent in nature.
In conclusion, this study has provided a detailed and comprehensive overview of the top 100 most-cited papers in the field of breast radiology. In so doing, this study can provide core insights for both clinicians and researchers alike, and lead them to understand the most influential papers within this important field. It is anticipated that this may positively impact on the way that this area is addressed in the future, and that the evidence base in this field can be further advanced as a result.

CONTRIBUTORS
Both DJYT and TKK contributed equally in conceptualisation, data curation, formal analysis, investigation and methodology section of the manuscript. Both authors have equal parts in project administration, writing draft, reviewing, and editing the manuscript.