Indian research on generative artificial intelligence in healthcare imaging: A comprehensive bibliometric analysis

INTRODUCTION

The healthcare industry is undergoing a transformative shift fueled by rapid advancements in artificial intelligence (AI), particularly in the domain of generative AI (GAI).^[1,2] AI is increasingly being recognized for its ability to detect patterns, anomalies, and risks, offering capabilities that range from administrative automation to clinical decision support, with the potential to improve patient outcomes, reduce healthcare costs, and accelerate medical discoveries.^[1] Among AI technologies, generative models, such as generative adversarial networks (GANs) and large language models, stand out due to their ability to synthesize new data, imagery, text, and other content with humanlike creativity and nuance.^[2]

Medical imaging remains a cornerstone of modern clinical practice, offering critical insights for diagnosing, treating, and monitoring a wide spectrum of conditions, including tumors, infections, and cardiovascular diseases.^[3] However, the high volume and complexity of data generated by imaging modalities such as X-rays, magnetic resonance imaging (MRI), and computed tomography (CT) scans continue to challenge traditional analysis methods.^[4,5] GAI models offer a compelling solution by enabling more precise, efficient, and automated interpretation of these complex datasets.

Generative AI techniques such as GANs and Variational Autoencoders are at the forefront of this revolution and are used for various tasks.^[6-9] The applications of GAI in healthcare, like medical imaging, span numerous specialties, including radiology, pathology, oncology, and cardiology, where they assist in tasks ranging from early cancer detection to predicting patient outcomes and even aiding in drug discovery and development.^[10,11] While GAI promises to enhance healthcare quality and accessibility while potentially reducing costs, it also introduces challenges related to data privacy, algorithmic bias, and model interpretability that must be addressed for responsible implementation.^[12,13]

To gain a comprehensive understanding of the research landscape in this domain, bibliometric analysis is a valuable tool.^[1,14] This quantitative method examines scholarly literature to identify research trends, influential authors, key institutions, and collaborative networks.^[15] While previous bibliometric studies have explored global trends in the application of GAI to Medical Imaging,^[6,7,16] a specific and comprehensive analysis of the “Generative AI in Healthcare Imaging,” “Application of GAI to MIA” within the Indian context has been lacking.^[16]

This study aims to fill said gap by examining the growth and citation impact of Indian research, evaluating the productivity of leading organizations and authors, and mapping the co-authorship and keyword co-occurrence patterns to uncover key themes and research gaps from 2017 to 2025. The main objective of the study is to understand bibliometric outcomes, including publication trends, citation impact, collaborative networks, and thematic patterns (rather than technical imaging applications).

MATERIAL AND METHODS

The Scopus database has been selected to identify, retrieve, and download the relevant literature in the field, using a comprehensive search strategy developed using a set of keywords related to “Generational AI” and “Medical Imaging” and limiting the search to the affiliation country “India” and period from 2017 to 2025.

The search strategy has been indicated below:

(KEY (“generative AI” OR “generative artificial intelligence” OR “generative adversarial network”) OR KEY (“ChatGPT” OR “art artificial intelligence” OR “artificial intelligence art”) OR KEY (“deep generative models” OR “intelligence amplication”) OR KEY (“foundation models” OR “artificial general intelligence” OR “hyperscale artificial intelligence” OR “Large Language Models” OR “foundational technique” OR “Generative predetermined transformer” OR “Gemini AI”)) AND KEY (medical image)) AND (LIMIT-TO (AFFILCOUNTRY, “India”)).

The above search was performed on 20.8.2025, leading to the download of 383 Indian records. The paper presents a detailed bibliometric analysis of the subject. From each record, the following information is extracted: author names, publication titles, publication years, affiliations, keywords, citation count, and funding. However, no medical images or algorithms were analyzed; only bibliographic metadata from Scopus (titles, authors, keywords, affiliations, and citations) were examined.

We have generated some figures for this biometric study by VOSviewer (https://www.vosviewer.com/), as it provides a visual representation of bibliometric relationships, enabling the identification of research clusters, collaboration networks, and emerging themes. These maps enhance the interpretability of complex data and highlight the structure and evolution of research within the field.

RESULTS

Overall picture

The research output from 2017 to 2025 demonstrates a significant upward trajectory in global publishing landscapes, including India [Figure 1]. The global output had 2,761 papers, of which India contributed 383, reflecting a robust overall Compound Annual Growth Rate of 1.329. India’s contribution to the global total expanded dramatically, rising from a mere 0.26% share (1 paper) in 2017 to a peak of 34.46% (132 papers) in 2024. However, while the volume of publications surged, the impact metrics showed an inverse trend; the Citations Per Paper (CPP) for Indian output was highest at the inception (71.00 in 2017) and steadily declined to 2.90 in 2024 and 2.67 in 2025, with the Relative Citation Index similarly (RCI) dropping from 10.847 to 0.409 [Table 1].

Close

Figure 1:

Rising trend of Indian publications in generative artificial intelligence in Healthcare (N.B.: data of *2025 is of an incomplete year, until August 20, 2025).

Export to PPT

Table 1: Bibliometric data of the growth of global and Indian output related to healthcare imaging (2017–2025).

Period	Global	India
	TP	TP	TP%	TC	CPP	RCI	CAGR
2017	6	1	0.26	71	71.00	10.847	0.000
2018	44	1	0.26	18	18.00	2.750	0.000
2019	102	6	1.57	99	16.50	2.521	1.4495
2020	153	10	2.61	116	11.60	1.772	1.1544
2021	279	21	5.48	301	14.33	2.190	1.1407
2022	368	39	10.18	535	13.72	2.096	1.0807
2023	499	87	22.72	754	8.67	1.324	1.105
2024	750	132	34.46	383	2.90	0.443	1.0088
2025	560	86	22.45	230	2.67	0.409	0.8738
Total	2761	383	100.00	2507	6.55	1.000	1.329

TP: Total papers, TC: Total citations, CPP: Citations per paper, CAGR: Compound annual growth rate, RCI: Relative citation index

The United States (US) with 596 papers (21.59%). India ranked third with 383 publications, demonstrating a significant and growing presence. While China and India were high in productivity, their research impact, as measured by CPP, was comparatively lower than that of many Western countries. The US had the highest CPP (34.76), indicating high-impact research. Other countries such as Canada (34.19 CPP), Spain (24.93 CPP), and the United Kingdom (UK) (24.77 CPP), also performed well in terms of research impact despite having fewer publications. India’s CPP was 6.55, and China’s CPP was 16.07, both below the global average of 21.71 [Table 2]. India’s research output exhibits a stark contrast between its high volume of publications and its low research impact. Ranking 3^rd globally in total papers (TP) with 383, India contributes a significant 13.87% to the total output of these top countries. While India is highly productive in publishing, its research is cited less frequently and, consequently, has a substantially lower average influence or scholarly impact compared to its major global counterparts such as the US and Canada [Table 2].

Table 2: Contribution of the top 12 countries in global output, related to healthcare imaging, comparing the research metrics with India.

S. No.	Name of the country	TP	TC	CPP	TP%
1	China	903	14510	16.07	32.71
2	United States	596	20,718	34.76	21.59
3	India	383	2507	6.55	13.87
4	United Kingdom	169	4186	24.77	6.12
5	Germany	134	2575	19.22	4.85
6	Canada	111	3795	34.19	4.02
7	South Korea	109	1539	14.12	3.95
8	Japan	105	1464	13.94	3.80
9	Australia	94	1948	20.72	3.40
10	France	66	832	12.61	2.39
11	Italy	60	1008	16.8	2.17
12	Saudi Arabia	59	1107	18.76	2.14
13	Spain	54	1346	24.93	1.96
14	Hong Kong	52	530	10.19	1.88
15	Switzerland	45	1116	24.8	1.63
		2761	59181	21.71	100.00

TP: Total papers, TC: Total citations, CPP: Citations per paper

This analysis shows a clear divide: certain countries, particularly in the West, prioritize quality and impact, while high-volume contributors such as China and India still need to improve their citation rates to match the global average.

Indian research

Indian research has been expanded and clarified to explain publication trends, institutional performance, and reasons for lower citation impact more coherently.

Indian contribution and citation impact

India’s contribution to this research field amounted to 383 publications, accounting for a 13.87% share of the global output. This output grew significantly from just one paper in 2017 to 86 in 2025, with a maximum output (132) in 2024. Despite this high publication volume, the citation impact was comparatively low, with an average of 6.55 CPP, which is well below the global average of 21.71 CPP. This lower impact is largely attributed to the dominance of conference papers, which constituted 58.49% of the total publications but had a very low CPP of 2.92. In contrast, journal articles (11.29 CPP) and review papers (14.84 CPP) garnered significantly more citations, highlighting the importance of publishing in high-impact, peer-reviewed journals to enhance research visibility and influence.

International collaborative papers

Of the 383 Indian publications, 57 papers (14.88%) involved international collaboration, which received a total of 520 citations, averaging 9.2 CPP. The US (17 papers) and the UK (nine papers) were India’s primary partners, reflecting strong academic ties. These were followed by China (five papers), Vietnam, the United Arab Emirates (UAE), Saudi Arabia, Norway, Laos, and Japan (four papers each), South Korea, the Russian Federation, Malaysia, Iraq, Germany, Canada, Botswana, and Australia (three papers).

While the geographical span of collaborations was diverse, the intensity of collaboration was low, with most countries contributing fewer than five papers [Figure 2]. The limited scale of these partnerships suggests a need for India to forge deeper collaborations with high-impact countries to boost its overall research visibility and influence.

Close

Figure 2:

International collaboration network in Indian generative artificial intelligence (AI) research on healthcare imaging: This network visualization depicts India’s collaborative landscape in generative AI research applied to healthcare imaging. Each node represents a collaborating country, with node size proportional to the number of joint publications and colors indicating clusters of interconnected nations. The United States and the United Kingdom emerge as India’s primary collaborators, followed by China, Vietnam, the United Arab Emirates, and Saudi Arabia. (Data Source: Scopus [2017–2025]; Visualized by VOSviewer [v1.6.19]).

Export to PPT

External funding

Only 32 (8.35%) Indian papers received external funding, which collectively garnered 205 citations, averaging 6.41 CPP. The primary funding agencies were national and international, with the Ministry of Electronics and Information Technology, India, being the most frequent contributor (three papers). Other key agencies included the University Grants Commission and the Department of Science and Technology, India, each supporting two papers.

Leading organizations

The top five organizations collectively contributed 200 papers (13.63%) and accounted for a significant 51.57% of total citations. In terms of productivity, SRM Institute of Science and Technology, Chennai (26 papers), and Vellore Institute of Technology, Vellore (13 papers) were the most prolific. For citation impact (CPP), GLA University, Mathura, stood out with a high CPP of 49.0, followed by Chandigarh University, Punjab (21.47 CPP). The analysis also highlights strong collaboration networks, with KLE Deemed to be University, Belgaum, showing the highest total link strength (TLS) (51 TLS), indicating its role as a key collaborator. The top six most productive and impactful organizations have been presented in Table 3.

Table 3: Six most productive and impactful Indian organizations contributing to the publications related to healthcare imaging.

S. No.	Name of the organization	TP	TP%	TC	CPP	TLS
Six most productive organizations
1	SRM Institute of Science and Technology, Chennai	26	6.79	59	2.27	17
2	Vellore Institute of Technology, Vellore	13	3.39	56	4.31	3
3	KLE Deemed to be University, Belgaum	12	3.13	50	4.17	51
4	Chandigarh University, Punjab	11	2.87	234	21.27	11
5	Delhi Technological University, New Delhi	10	2.61	116	11.60	8
6	JAIN (Deemed to be a university), Bangalore	9	2.35	1	0.11	26
Six most impactful organizations
1	GLA University, Mathura	5	1.31	245	49.00	13
2	Chandigarh University, Punjab	11	2.87	234	21.27	11
3	Indian Institute of Technology Madras	5	1.31	98	19.60	3
4	Delhi Technological University, New Delhi	10	2.61	116	11.60	8
5	National Institute of Technology, Patna	7	1.83	61	8.71	12
6	Lovely Professional University	5	1.31	43	8.60	10

TP: Total papers, TC: Total citations, CPP: Citations per paper, TLS: Total link strength

This concentration of research output and impact among a few institutions suggests a need to strengthen partnerships to enhance India’s collective research presence.

In the co-authorship network of leading Indian institutions, the VOSviewer software revealed a dual structure of collaboration [Figure 3]. While 17 institutions have formed collaborative links, creating distinct clusters, five have remained isolated, highlighting an uneven distribution of engagement. The analysis identified several key clusters, including Cluster 1, led by SRM Institute of Science and Technology, Kattankulathur (TLS = 17), occupying a central and dominant position in the network, serving as a key hub in Indian research collaboration and represented by its large node size and multiple connecting links. The Vellore Institute of Technology (TLS = 51), KLE Deemed to be University, and Manipal University also show significant activity, with exceptional collaborative intensity (TLS = 51), reflecting extensive partnerships across institutions despite relatively moderate publication volume. The diverse interconnections suggest healthy interdisciplinary and inter-institutional engagements among India’s leading technological and higher education institutions. Indicating its role as a key collaborator.

Close

Figure 3:

Co-authorship network among top Indian organizations in generative artificial intelligence and medical imaging: This VOSviewer-based map illustrates inter-institutional collaboration patterns across leading Indian organizations. Node size corresponds to publication productivity, while colors mark distinct collaboration clusters and link thickness reflects total link strength. The analysis identified several key clusters, including Cluster 1, led by SRM Institute of Science and Technology, Kattankulathur, which occupies a central and dominant position, serving as a key hub in Indian research collaboration. (Data Source: Scopus [2017–2025]; Visualized by VOSviewer [v1.6.19]).

Export to PPT

These clusters showcase a mix of highly productive and emerging institutions, underscoring that some universities are actively building collaborative networks while others operate independently. Strengthening these inter-organizational partnerships is therefore crucial for enhancing India’s collective research presence in this domain.

Leading authors

Of 392 authors involved in the Indian research on GAI and medical image processing (MIP), a small group of 18 authors was highly productive, collectively contributing 61 papers and receiving 441 citations Among these top authors, S.N. Bhat and R.M. Pai from Manipal Academy of Higher Education, and G. Singh from Chandigarh University, were the most productive, each with four papers. However, the highest citation impact (CPP) was achieved by S.P. Awate and U. Upadhyay from the Indian Institute of Technology, Bombay (17.67 CPP each). The analysis also highlighted S.N. Bhat and R.M. Pai as having the strongest co-authorship networks, with a TLS of 12, indicating their central roles in collaboration [Table 4]. This shows that a few key researchers are driving the field, with some excelling in productivity and others in research impact and collaboration. The top six most productive and impactful authors have been mentioned in Table 4.

Table 4: Six most productive and impactful Indian authors contributing to the publications related to healthcare imaging.

S. No.	Name of the author	Name of the Organization	TP	TP%	TC	CPP	TLS
Six most productive authors
1	S. N. Bhat	Manipal Academy of Higher Education, Manipal	4	1.04	17	4.25	12
2	R.M. Pai	Manipal Academy of Higher Education, Manipal	4	1.04	17	4.25	12
3	G. Singh	Chandigarh University, Punjab	4	1.04	0	0.00	12
4	S.P. Awate	Indian Institute of Technology, Bombay	3	0.78	53	17.67	4
5	V.Wiwanitkit	Saveetha Institute of Medical and Technical Sciences, Chennai	3	0.78	0	0.00	3
6	U.Upadhyay	Indian Institute of Technology, Bombay	3	0.78	53	17.67	4
Six most impactful authors
1	S.P. Awate	Indian Institute of Technology, Bombay	3	0.78	53	17.67	4
2	U.Upadhyay	Indian Institute of Technology, Bombay	3	0.78	53	17.67	4
3	A.K. Singh	National Institute of Technology, Patna	3	0.78	47	15.67	11
4	V. Sarada	SRM Institute of Science and Technology, Chennai	3	0.78	31	10.33	6
5	S.P. Porkodi	SRM Institute of Science and Technology, Chennai	3	0.78	31	10.33	6
6	S. N. Bhat	Manipal Academy of Higher Education, Manipal	4	1.04	17	4.25	12

TP: Total papers, TC: Total citations, CPP: Citations per paper, TLS: Total link strength

The co-authorship analysis of the top 18 authors highlights the collaborative dynamics of a research network, Indian research on GAI and MIP, as shown in Figure 4.

Close

Figure 4:

Co-authorship network of the top 18 Indian authors in generative artificial intelligence for healthcare imaging: Nodes denote individual researchers sized by publication output, with colors representing collaborative clusters and connecting lines indicating co-authorship strength. The visualization reveals Bhat, Pai, and Sindhura of Manipal Academy of Higher Education (in red cluster) showing the strongest internal collaboration (total link strength = 12). (Data Source: Scopus [2017–2025]; Visualized by VOSviewer [v1.6.19]).

Export to PPT

The visualization reveals 11 clusters, with the largest and most cohesive cluster (Red) consisting of Shaymasunder N. Bhat, Manohara R.M. Pai, and D.N. Sindhura, all from the Manipal Academy of Higher Education, Manipal. Their strong interlinkages (high TLS = 12) indicate a well-established internal collaboration network and a consistent research partnership within the institutions.

The Green cluster, comprising Uddeshya Upadhyay and Suyash P. Avate from the Indian Institute of Technology, Bombay, demonstrates close bilateral collaboration (TLS = 4) with a remarkably high citation impact (CPP = 17.67), reflecting the quality and visibility of their research output. The blue cluster connects Anita Murmu and Piyush Kumar from the National Institute of Technology, Patna, showing moderate collaboration strength (TLS-8) and consistent productivity. Another smaller but distinct yellow cluster links S.T. Vijaya Sarada and S.P. Porkodi from the SRM Institute of Science and Technology, Chennai, indicating a tight co-authorship bond (TLS = 6) and moderates citation performance (CPP = 10.33).

In contrast, Viroj Wiwanitkit and Hinpetc Daungsupawong from the Saveetha Institute of Medical and Technical Science, Chennai, from an isolated purple cluster, showing limited but institutionally consistent collaboration (TLS = 3). Scattered nodes, such as A.K. Singh (NIT Patna), S. Mehta (Chitkara University), and Vijaypal Singh Dhaka (Chandigarh University), represent less connected or independent authors with low co-authorship density, despite having reasonable productivity.

Leading publishing journals

The Indian research on GAI and MIP has been primarily published in conference proceedings, which accounted for the largest share of papers (179 papers, or 46.74%). However, these conference papers had a lower citation impact. In contrast, journal articles (149 papers, or 38.9%) and book series (50 papers, or 13.05%) had a much higher impact.

The papers were distributed across 88 journals, but a small number of top journals published a significant portion of the work. The top 20 journals published 82 papers, accounting for over 55% of the total journal papers and more than 70% of their citations. Journals such as Multimedia Tools and Applications (15 papers) were the most productive, while others, such as Information Fusion (IF) (114.0 CPP) and Current Medical Imaging (38.75 CPP), had the highest citation impact.

Keywords analysis

Keywords provide valuable insights into research priorities and emerging trends in the field of GAI and MIP in India. Out of 3,190 keywords identified across the 383 papers, a shortlist of 47 significant keywords was created. The most frequently occurring keywords, which define the core of the research, were “generative adversarial networks” (285 occurrences), “medical imaging” (256 occurrences), and “deep learning” (188 occurrences).

Using VOSviewer software for co-occurrence analysis, the keywords were organized into six distinct clusters, each representing a thematic orientation. The red cluster focused on the application of AI in image analysis and diagnostics. The green cluster highlighted key technical applications, centered on generative models and image quality, featuring terms such as “generative adversarial networks,” “image enhancement,” and “image denoising.” The blue cluster was defined by deep learning and data-related tasks, including “data augmentation,” “image classification,” and “transfer learning” [Figure 5].

Close

Figure 5:

Keyword co-occurrence network in Indian publications on generative artificial intelligence in healthcare imaging (2017–2025): This map organizes 47 high-frequency keywords from 383 Indian papers into six thematic clusters. Node size reflects keyword frequency, colors identify thematic groups, and link thickness indicates co-occurrence strength. Dominant terms included “generative adversarial networks,” “medical imaging,” and “deep learning.” (Data Source: Scopus [2017–2025]; Visualized by VOSviewer [v1.6.19]).

Export to PPT

These clusters demonstrate that Indian research is concentrated on both fundamental technical aspects and their practical clinical applications, confirming a robust and focused research landscape.

Highly cited papers (HCPs)

The top 20 HCPs collectively received 1,280 citations, averaging 64.0 CPP. Only four of these papers received more than 100 citations. The majority of these HCPs were published in 2023 (seven papers) and 2022 (four papers), with the rest distributed between 2021, 2024, and 2025.

In terms of collaboration, 16 out of the 20 HCPs involved two or more organizations, with an equal number of papers resulting from national and international collaborations. The international collaborations included authors from eight different countries, with the US and Japan being the most frequent partners. The distribution of these top papers across various high-quality journals indicates the high standard of research being conducted.

DISCUSSION

The bibliometric analysis of research on GAI in healthcare reveals a dynamic and rapidly evolving landscape, particularly in the Indian context.^[1,14,15] From 2017 to 2025, a total of 2,761 papers were published globally, with an average of 306.78 papers per year and a notable annual average growth rate of 133.2%. Over 65% of these publications appeared in the past 3 years (2023–2025), indicating a surge in research activity in this field. The global output collectively received 59,181 citations, averaging 21.71 CPP.^[2,3]

India has emerged as a significant contributor, ranking third globally with 383 publications, which accounts for 13.87% of the total global output. This places India behind China and the US, which lead with 903 and 596 publications, respectively. While China’s output is substantial, its CPP (16.07) is relatively low, whereas the US boasts the highest CPP (34.76), signifying a high impact and visibility of its research. India’s CPP stands at 6.55, which is also below the global average benchmark of 21.71, suggesting that while the country’s research output is growing, its overall impact per publication is still limited compared to its Western counterparts.^[4,5] Another Asian country, South Korea, although it has a lower TP of 109 but has a higher CPP of 14.12 compared to India.

India is rapidly advancing in AI applications for medical imaging through public-private partnerships, academic collaborations at institutions such as IITs, AIIMS, and CDAC, and innovative startups such as Niramai and Qure. ai, focusing on early detection of cancers, tuberculosis, diabetic retinopathy, and other diseases through analysis of X-rays, CT scans, MRIs, ultrasounds, and positron-emission tomography images.^[17] However, there is a potential translational gap in the engineering-driven versus clinical research imbalance, suggesting a need for increased collaboration between engineering institutions and clinical centers/AIIMS to move from technical GAI models to validated clinical applications. These efforts enhance image processing by reducing noise, improving segmentation and pattern recognition, and boosting diagnostic accuracy, such as AI outperforming radiologists in thoracic disease detection on chest radiographs,^[18] while enabling quantitative analysis, faster interventions, cost reductions, and expanded access in rural areas amid a radiologist shortage and projected market growth to over $7 billion by 2033. Despite challenges such as data privacy breaches, algorithmic biases, interpretability issues, and integration hurdles, future trends, including federated learning, personalized medicine through multimodal data integration, edge AI, and AI-augmented radiology reporting – where vision-language models generate reports preferred over human ones in over 60% of cases,^[19] promise transformative, ethical deployment to improve patient outcomes nationwide. AI also shows promise in ultrasound for 83.8% accurate gestational age prediction,^[20] highlighting its role in scalable, precise healthcare delivery.

A key finding related to India’s output is the dominance of conference papers, which constitute 58.49% of the 383 Indian publications. These papers, however, have a low average CPP of 2.92, indicating less overall research impact compared to original research articles and review papers. Journal articles (31.85% of TPs) received a much stronger CPP of 11.29, and review papers had an even higher CPP of 14.84, highlighting the crucial role of high-quality, peer-reviewed journals in gaining visibility and impact.^[6,7] The low CPP of Indian publications is primarily due to the dominance of conference papers with low CPP over higher-impact original articles. It also links lower impact to limited international collaboration (14.88% of papers) and a low rate of external funding (8.35% of papers). This points to a strategic imperative for Indian researchers to prioritize publishing in reputable journals to enhance their global visibility.

In terms of collaboration, 14.88% (57 out of 383) of Indian papers involved international collaboration, with a higher average CPP of 9.2. The US (17 papers) and the UK (nine papers) were India’s most frequent collaborators, reflecting strong academic and scientific ties. While India’s global research network is expanding, the intensity of collaboration is low, with many countries contributing fewer than five collaborative papers.^[8,9] Fostering deeper collaborations with high-impact countries could significantly boost India’s research visibility and impact.

The analysis of leading institutions reveals an uneven distribution of contributions. While 115 organizations participated, a small number of institutions dominate in both productivity and impact.^[10,11] The top five organizations collectively contributed 13.63% of the TPs and an impressive 51.57% of the total citations, indicating a high concentration of influential research. Notably, SRM Institute of Science and Technology, Chennai, and Vellore Institute of Technology, Vellore, were among the most productive, while institutions such as GLA University, Mathura, and Chandigarh University, Punjab, demonstrated a higher citation impact per paper.^[12,13] The co-authorship network further reveals that some institutions are actively building collaborative links, forming clusters, while others work more independently. Strengthening these inter-organizational partnerships is key to enhancing India’s collective presence in the research domain.

Similarly, the author’s analysis shows that a small group of researchers is driving the field.^[14] Among the top 12 authors, S. N. Bhat, R.M. Pai, and G. Singh contributed the most papers, while S.P. Awate and U. Upadhyay from the Indian Institute of Technology, Bombay, had the highest citation impact. The keyword analysis underscores the core research themes within the Indian context.^[15] The most frequently occurring keywords, such as “generative adversarial networks” (occurrences = 285) and “medical imaging” (occurrences = 256), along with “deep learning” (occurrences = 188) and “image enhancement” (occurrences = 154), indicate a strong focus on core technical applications and models. The clustering of keywords into distinct groups reveals thematic orientations, including a focus on image segmentation and analysis, technical aspects of generative models, and specific applications such as data augmentation and classification. This analysis confirms that Indian research is concentrated on key areas of GAI in healthcare, but also suggests opportunities to explore new, emerging sub-fields to further diversify and enhance the research landscape.^[7,21,22] Synthetic imaging datasets (a core application of GAI) and foundation models should act as critical future directions for Indian research. Our findings (e.g., low-impact conference papers, low funding rate, uneven collaboration) suggest actionable guidance for prioritizing funding for high-impact journals and international partnerships to enhance high-impact publications.

The present study is limited by its scope as a bibliometric analysis of publications, citations, collaborations, and keywords, and does not involve the evaluation of medical images, algorithms, or AI models. Data were drawn exclusively from the Scopus database, which may not capture all relevant publications indexed elsewhere. The analysis period was restricted from 2017 to an incomplete year of 2025 (only up to August 20^th, 2025) and hence does not capture the full year, which could distort trend interpretation. Accordingly, the findings should be interpreted as reflecting publication activity rather than technical or experimental insights. While a multi-database approach can be comprehensive, the Scopus database was a deliberate choice for this specific domain. Scopus is widely recognized for its strong coverage of engineering and computer science conference proceedings and journals, which are the primary publication venues for Generative AI and MIP research. Given that the search was centered on technical keywords (e.g., “generative adversarial network,” “deep learning”). Furthermore, mixing bibliometric data from databases such as Scopus and Web of Science creates problems such as data format inconsistency, duplicate records, and inaccurate citation counts due to each database’s unique, closed system. The manual merging process is time-consuming and prone to human error, while automated methods are required to handle different file structures, standardize metadata, and perform accurate deduplication.^[23]

CONCLUSION

This study provides a comprehensive bibliometric overview of Indian research on generative AI in healthcare imaging from 2017 to 2025, and identifies key productive and impactful institutions and authors. Although India ranks third globally in research output, its citation impact remains below the global average, reflecting the need to improve research quality, visibility, and international collaboration. The study’s findings suggest that for enhancing India’s research impact, prioritization of funding for high-impact journals and international partnerships is advisable, as a part of resource allocation.