Is Artificial Intelligence a Threat to Radiologists? Perception of Radiologists in Saudi Arabia

Artificial Intelligence (AI), described as any program, mechanism or tool capable of identifying modifications in its environment to accomplish specific objectives, has profoundly influenced not only medicine but also multiple facets of modern life [1]. The foundational concept, introduced by Alan Turing in 1950, proposed that computers could emulate human intelligence [2]. In healthcare, AI has demonstrated promising utility across various specialties by predicting prognoses, supporting clinical decision-making and performing complex image analyses-particularly in radiation oncology.

In radiology, AI applications are increasingly being integrated into workflows for tasks such as mammography screening, report generation, cancer staging and detection of incidental findings [3]. These tools enhance efficiency by reducing radiologists’ workload and improving reporting accuracy and turnaround time [4,5].

Several international studies have evaluated radiologists’ awareness and perceptions of AI. For example, a study from the United Arab Emirates highlighted a lack of knowledge and underappreciation of AI’s potential, which may have limited its utility [6]. Similarly, research in Ghana revealed that although 97% of radiologists had heard of AI, their knowledge remained generally average [7]. In Saudi Arabia, a previous study reported that 69% of radiologists had basic knowledge of AI and only 29% viewed it as a potential threat [8]. In contrast, 90% of Indian radiologists expressed concern about job loss due to AI integration [9]. Nonetheless, a promising trend has emerged: a majority of radiologists, including 97% in one study, are interested in learning about and integrating AI into their practice [10].

Despite global interest, the local context in Saudi Arabia remains underexplored. Understanding the current knowledge, attitudes and training needs of Saudi radiologists is crucial, especially given the rapid technological evolution in medical imaging. This study addresses that gap by assessing radiologists’ perceptions of AI across various regions in the Kingdom.

The objective of this study is to evaluate the perception of AI among radiologists in Saudi Arabia, contribute to national data and support the formulation of guidelines for structured in-service training. Such training can enhance radiologists’ ability to leverage AI technologies for improved reporting efficiency, diagnostic accuracy and patient management.

Study Design and Setting

This cross-sectional study was conducted from March 2024 to February 2025 after obtaining ethical approval from the Local Committee of Bioethics at Northern Border University, Arar, Saudi Arabia (approval letter no. 5/24/H, dated 31/01/2024). The study aimed to assess the perceptions, knowledge and training needs related to Artificial Intelligence (AI) among radiologists practicing in Saudi Arabia.

Questionnaire Development

The survey instrument was adapted from a previously validated study by Alghamdi and Alashban [8] and further refined based on an extensive review of relevant literature obtained from databases such as Google Scholar, PubMed and ScienceDirect. The questionnaire was translated into an online format using Google Forms for ease of distribution and completion.

To ensure content validity, the revised tool was reviewed and validated by two experts in community medicine following a pilot study. The finalized questionnaire consisted of six sections with a total of 31 questions, covering the following domains:

• Demographics (e.g., age, gender, specialization, years of experience, workplace and region)
• AI Education History (binary yes/no questions)
• Knowledge of AI Concepts (e.g., machine learning, deep learning)
• Attitudes toward AI in Radiology (measured via Likert scale)
• AI-related Practices in the Workplace
• Open-ended Comments and Suggestions

Sample Size and Sampling Technique

The required sample size was calculated as 103 radiologists based on a 90% confidence level, 7.5% margin of error and an expected prevalence of 69% having basic knowledge of AI [8]. The formula used was:

A non-probability consecutive sampling method was employed. All radiologists currently practicing in Saudi Arabia who provided informed consent were eligible for inclusion. Those not in active service or who declined to participate were excluded. Consent was implied by the voluntary completion of the online survey.

Data Collection Procedure

The survey link was disseminated through professional radiology WhatsApp groups and email lists across all regions of Saudi Arabia. The principal investigator actively promoted participation using professional networks to enhance reach and ensure regional representation. Participation was voluntary and no incentives were provided.

Ethical Considerations

Ethical approval was secured prior to data collection. While explicit written consent was not obtained due to the online nature of the survey, informed consent was implied through the act of completing the questionnaire. Anonymity and confidentiality of all responses were assured. No identifying information was collected. Ethical aspects related to the use of online platforms and data protection were considered during the design of the study.

Data Analysis

Data were entered and analyzed using IBM SPSS Statistics for Windows, Version 20 (IBM Corp., Armonk, NY, USA). Descriptive statistics were used to summarize demographic and categorical variables in terms of frequencies and percentages. Continuous variables were presented as Means±standard deviation.

Comparative analyses were conducted using chi-square tests for categorical variables and independent-samples t-tests for continuous variables. A p-value ≤0.05 was considered statistically significant. Stratification techniques were applied to control for potential confounding variables, such as age, experience and type of institution.

The online survey was completed by 103 radiologists from all over the Kingdom of Saudi Arabia. Males constituted 59.2% (n = 61) while 40.8% (n = 42) were female radiologists. The majority of the radiologists, 43.7% (n = 45) belonged to the age group 30 to 40 years. The majority of the participants (58%, n = 60) were FCPS, the rest had other forms of specialization. The maximum number 79.6% (n = 82) were working in public hospitals. The majority 66% (n = 68) had less than 5 years of experience. The maximum number of radiologists responded from the western region (52.4%, n = 54) (Table 1).

The radiologists who had heard about AI were 85.4% (n = 88). Out of all, 72.8% (n = 75) had basic information about AI. More than three-quarters of the radiologists knew about deep learning and machine learning. Ninety-four percent (n = 97) of the participants did not receive any education about AI during their degree and 79.6% (n = 82) did not receive any professional training in AI use. The 69.9% (n = 72) radiologists demanded that AI should be taught as continuous professional development (Table 2).

Out of all the respondents, 97(94.2%) described themselves as people who enjoy technology. The majority of the participants (69.9%, n = 72) were rather comfortable in explaining an algorithm. More than three-quarters (78.6%, n = 81) radiologists were of ready to adapt to the novel technology of AI in their field. Only 15.5% (n = 16) of radiologists were confident enough to explain the difference between AI, machine learning and deep learning. Only 29.3% (n = 30) of radiologists were apprehensive about the introduction of AI into radiology. About 73% (n = 75) of radiologists were excited about the introduction of AI into radiology. About 66% (n = 68) of radiologists thought that AI already played a role in radiology. About 82% of radiologists disagreed with the notion that AI will replace them. About 65% of radiologists thought that AI played an important role in radiology (Table 3).

A significant relationship was found between age and experience with the knowledge of AI.

Table 1: Demographics of participants

Table 2: Knowledge about artificial intelligence

Table 3: Attitude and practices towards AI

Table 4: Association of Knowledge of AI with age

Table 5: Association of knowledge of AI with experience

Almost 93% (n = 42) of the radiologists less than 40 years of age had heard of AI (p<0.001). However, 55% of radiologists above 60 years of age had heard of AI. Similarly, the younger radiologists having lesser experience had more heard of AI than their seniors (p<0.001) (Table 4, 5).

The integration of artificial intelligence (AI) into medical imaging has significantly enhanced diagnostic accuracy and operational efficiency. However, the effective utilization of AI tools in radiology hinges on the professionals’ comprehension of its functions and clinical applications [11]. As AI continues to evolve, discussions around its potential to disrupt or replace radiologists have intensified [12]. These ongoing debates underline the importance of understanding radiologists' perceptions, particularly within specific local contexts such as Saudi Arabia.

In this study, 85.4% of radiologists reported having heard of AI and 72.8% acknowledged having basic knowledge of its use in radiology. This level of awareness is considerably higher than that reported in several previous international studies. For instance, Huisman et al. [13] found that only 21% of surveyed radiologists had prior exposure to AI and just 30% had basic knowledge of its applications. Similarly, a French study reported that only 23% of radiologists had foundational knowledge of AI [14]. When compared to earlier local data from Saudi Arabia, which showed that 69% of radiologists had basic knowledge [8], our study demonstrates a positive upward trend.

Despite this improvement, advanced understanding remains limited-only 9.7% of respondents in our study reported having advanced knowledge of AI, a figure closely aligned with Huisman et al. [13] finding of 11%. This indicates an urgent need for structured education and training initiatives to deepen radiologists’ understanding of AI’s practical uses in clinical workflows [16].

Interestingly, only 9.7% of our respondents expressed concern about AI replacing radiologists, a stark contrast to findings by Tajaldeen and Alghamdi [17], where 73% reported such fears. This shift may reflect increasing familiarity with AI and a growing realization that it is more likely to augment radiological practice rather than replace professionals. A recent international study also supports this perspective, showing that 88.9% of radiologists believe their roles will remain essential despite AI advancements [18]. As the technology matures, continuous professional education and evidence-based communication will be crucial in dispelling myths and reinforcing AI’s role as a collaborative tool in radiology.

The enthusiasm for learning AI was another key highlight. Approximately 70% of the participants expressed interest in participating in continuous professional development programs focused on AI. This is consistent with another study, which showed that up to 90% of radiologists favored in-service education on AI [19]. This strong interest underscores the need for national and institutional strategies to introduce AI into radiology curricula, professional workshops and certification programs. Such steps are particularly relevant in a rapidly developing healthcare system like Saudi Arabia, which aims to remain aligned with international technological standards [20].

Subgroup Differences: Age and Experience

Our findings revealed significant associations between AI knowledge and both age and experience. Radiologists under 40 years of age (93%) and those aged 41-50 years (100%) were significantly more likely to have heard of AI than their older counterparts (p<0.001). While 71% of radiologists under 40 years and 70% of those aged 41-50 reported basic knowledge of AI, this proportion dropped in the older age groups (p = 0.135). Similar trends were noted in studies assessing information and communication technology (ICT) knowledge, where younger professionals demonstrated better proficiency [21]. This generational divide likely reflects differences in training exposure and technological engagement, with newer residency programs increasingly incorporating AI-related content.

However, not all literature supports this trend. A Norwegian study found that older radiologists could possess knowledge levels comparable to their junior colleagues, with 70% of those above 50 years and 24% below 50 showing average AI knowledge (p = 0.077) [22]. These discrepancies highlight the need for universal training initiatives, regardless of experience level.

Experience-wise, radiologists with fewer than five years in practice had significantly higher levels of AI awareness-95.6% had heard of AI, compared to only 50% of those with over 15 years of experience (p<0.001). Knowledge appeared to decline progressively with years of experience. This pattern may be due to the fact that recently trained radiologists are more likely to have been exposed to AI during their medical education. In line with our findings, a study across Arab countries also reported that professionals with less experience had significantly greater AI knowledge (p = 0.024), attributing this to more current academic exposure (1.486±0.5).

In summary, while the perception of AI among radiologists in Saudi Arabia is generally positive-with low levels of fear and high levels of interest in further learning-there are clear disparities in knowledge based on age and experience. Addressing these gaps through structured training and curriculum reforms is essential to ensure equitable AI adoption across all levels of practice.

Future Research and Recommendation

Given the rapidly evolving nature of AI in medicine, future research should prioritize multi-centric, larger-scale studies to strengthen generalizability. Equal regional representation is essential to understand localized barriers and facilitators of AI integration. Comparative international studies could also offer valuable global insights.

Subsequent investigations should evaluate the practical, clinical, ethical and legal aspects of AI use in radiology, as these remain underexplored [23]. There is also a pressing need for randomized controlled trials to assess the effectiveness of structured, in-service AI training programs and their impact on clinical outcomes.

In addition, research should extend beyond radiology to include other medical specialties such as surgery, pathology and emergency medicine, as AI is increasingly becoming relevant across various disciplines. Cross-specialty research can help identify common challenges and interdisciplinary strategies for AI adoption. Finally, studies should also investigate ethical considerations, including patient data protection, bias in algorithms and transparency of AI decision-making processes, to ensure responsible and equitable use of AI in healthcare [24].

This study concludes that radiologists in Saudi Arabia largely view artificial intelligence not as a threat but as an evolving tool that can enhance their professional roles. While the majority of respondents reported basic awareness of AI, advanced knowledge remains limited, particularly among more experienced practitioners. Encouragingly, a substantial proportion of radiologists expressed strong interest in continuous education on AI.

These findings underscore the urgent need for structured national guidelines and training programs to equip radiologists with the skills necessary to engage meaningfully with AI technologies. As radiology continues to evolve in the age of digital transformation, proactive efforts in education and policy-making are essential to ensure that professionals are prepared to leverage AI safely and effectively.

Limitations

This study has several limitations that should be acknowledged. First, the sample size was relatively small and may not fully represent the diversity of radiologists across Saudi Arabia. Second, the use of a non-probability sampling method and reliance on digital platforms such as WhatsApp may have introduced selection bias, potentially excluding radiologists with limited access to technology or those in rural areas. Additionally, although participants were drawn from multiple regions, regional representation was unequal. Future studies should aim for balanced geographic sampling to identify region-specific variations in AI adoption and preparedness.

Moreover, this study focused primarily on perceptions, knowledge and self-reported practices. It did not assess the practical application, clinical integration or real-world use of AI tools. Furthermore, critical domains such as ethical concerns, data privacy, legal implications and algorithmic accountability were not explored in depth. These areas are crucial for the safe and effective deployment of AI in radiology and warrant inclusion in future research designs.

Acknowledgement

The authors extend their sincere appreciation to the Deanship of Scientific Research at Northern Border University, Arar, Saudi Arabia, for funding this research work through the project number “NBU-FFR-2025-2709-02”.

Conflicts of Interest

The authors declare no conflict of interest related to this study.

1. Syed, Ali B. and Adam C. Zoga. “Artificial intelligence in radiology: current technology and future directions.” Seminars in Musculoskeletal Radiology, vol. 22, no. 5, 2018, pp. 540-545. https://www.thieme-connect.com/ products/ejournals/abstract/10.1055/s-0038-1673383.

2. Malik, Paras et al. “Overview of artificial intelligence in medicine.” Journal of Family Medicine and Primary Care, vol. 8, no. 7, July 2019, pp. 2328-2331. https://journals.lww.com/ jfmpc/fulltext/2019/08070/Overview_of_artificial_intelligence_in_medicine.27.aspx.

3. Khafaji, Mawya A. et al. “Artificial intelligence in radiology: are Saudi residents ready, prepared and knowledgeable?.” Saudi Medical Journal, vol. 43, no. 1, January 2022, pp. 53-60. https://pmc.ncbi.nlm.nih.gov/ articles/PMC9280560/.

4. Qurashi, Abdulaziz A. et al. “Saudi radiology personnel’s perceptions of artificial intelligence implementation: a cross-sectional study.” Journal of Multidisciplinary Healthcare, vol. 14, November 2021, pp. 3225-3231. https://www.tandfonline. com/doi/abs/10.2147/JMDH.S340786.

5. Dikici, Engin et al. “Integrating AI into radiology workflow: levels of research, production and feedback maturity.” Journal of Medical Imaging, vol. 7, no. 1, February 2020, pp. 016502-016502.

6. Abuzaid, Mohamed M. “Assessment of the willingness of radiologists and radiographers to accept the integration of artificial intelligence into radiology practice.” Academic Radiology, vol. 29, no. 1, January 2022, pp. 87-94. https://www.Science direct.com/science/article/pii/S1076633220305535.

7. Edzie, Emmanuel Kobina Mesi et al. “Perspectives of radiologists in Ghana about the emerging role of artificial intelligence in radiology.” Heliyon, vol. 9, no. 5, May 2023. https://www. cell.com/heliyon/fulltext/S2405-8440(23)02765-2.

8. Alghamdi, Sami A. and Yazeed Alashban. “Knowledge, attitudes and practices towards artificial intelligence (AI) among radiologists in Saudi Arabia.” Journal of Radiation Research and Applied Sciences, vol. 16, no. 2, June 2023. https://www.sciencedirect.com/science/article/pii/S168785072300047X.

9. Coppola, Francesca et al. “Artificial intelligence: radiologists’ expectations and opinions gleaned from a nationwide online survey.” La Radiologia Medica, vol. 126, April 2020, pp. 63-71. https://link.springer.com/article/10.1007/s11547-020-012 05-y.

10. Murugaiah, Viswanathan and Dharan Venkatesh K.A. “Radiologist knowledge of artificial intelligence on diagnostic radiology: A survey-based study.” International Journal of Medical Reviews and Case Reports, vol. 6, no. 16, November 2022, pp. 70-74. https://mdpub.net/?mno=125783.

11. Yu, Feiyang et al. “Heterogeneity and predictors of the effects of AI assistance on radiologists.” Nature Medicine, vol. 30, no. 3, March 2024, pp. 837-849. https://www.nature.com/articles/s41591-024-02850-w.

12. Strohm, Lea et al. “Implementation of artificial intelligence (AI) applications in radiology: hindering and facilitating factors.” European Radiology, vol. 30, May 2020, pp. 5525-5532. https://link.springer.com/article/10.1007/s00330-020-06946-y.

13. Huisman, Merel et al. “An international survey on AI in radiology in 1,041 radiologists and radiology residents part 1: fear of replacement, knowledge and attitude.” European Radiology, vol. 31, March 2021, pp. 7058-7066. https:// link.springer.com/article/10.1007/s00330-021-07781-5.

14. Waymel, Quentin et al. “Impact of the rise of artificial intelligence in radiology: what do radiologists think?.” Diagnostic and Interventional Imaging, vol. 100, no. 6, June 2019, pp. 327-336. https://www.sciencedirect.com/ science/article/pii/S2211568419300907.

15. Alsharif, Walaa et al. “A qualitative study to explore opinions of Saudi Arabian radiologists concerning AI-based applications and their impact on the future of the radiology.” BJR Open. vol. 4, no. 1, January 2022. https:// academic.oup.com/bjro/article-abstract/4/1/20210029/724031 4.

16. Schuur, Floor et al. “Training opportunities of artificial intelligence (AI) in radiology: a systematic review.” European Radiology, vol. 31, February 2021, pp. 6021-6029. https:// link.springer.com/article/10.1007/s00330-020-07621-y.

17. Tajaldeen, Abdulrahman and Salem Alghamdi. “Evaluation of radiologist’s knowledge about the Artificial Intelligence in diagnostic radiology: a survey-based study.” Acta Radiologica Open, vol. 9, no. 7, July 2020. https://journals.sagepub.com/ doi/full/10.1177/2058460120945320.

18. Alelyani, Magbool et al. “Radiology community attitude in Saudi Arabia about the applications of artificial intelligence in radiology.” Healthcare, vol. 9, no. 7, July 2021. https://www. mdpi.com/2227-9032/9/7/834.

19. Hedderich, Dennis M. et al. “AI for doctors-a course to educate medical professionals in artificial intelligence for medical imaging.” Healthcare, vol. 9, no. 10, September 2021. https://www.mdpi.com/2227-9032/9/10/1278.

20. Baig, Mansoor Ali et al. “Immunostimulating influence of herbal biomedicines on nonspecific immunity in grouper Epinephelus tauvina juvenile against Vibrio harveyi infection..” The Importance of Health Informatics in Public Health during a Pandemic, edited by Mantas, John et al., Washington, DC, IOS Press, 2020, pp. 253-256. https://ebooks.iospress.nl/doi/10.3233/SHTI200542.

21. Catanese, A. et al. “Radiologists and trainees’ perspectives on artificial intelligence.” Radiología, vol. 67, no. 3, June 2025, pp. 287-298. https://www.sciencedirect.com/science/article/ pii/S217351072400048X.

22. Lippert, Karoline Asama Nyberg. Artificial Intelligence in Radiology: Promises and Pitfalls. A cross-sectional study of Norwegian radiologists’ knowledge and attitudes. https://www.duo.uio.no/bitstream/handle/10852/80322/44/Innlevert-masteroppgave-Juni-2020---KANL.pdf.

23. Chan, Kai Siang and Nabil Zary. “Applications and challenges of implementing artificial intelligence in medical education: integrative review.” JMIR Medical Education, vol. 5, no. 1, April 2019. https://mededu.jmir.org/2019/1/e13930.

24. Eltawil, Fatma A. et al. “Analyzing barriers and enablers for the acceptance of artificial intelligence innovations into radiology practice: a scoping review.” Tomography, vol. 9, no. 4, July 2023, pp. 1443-1455. https://www.mdpi.com/2379-139X/9/4/115.