The radiography profession is rapidly evolving with the introduction of artificial intelligence (AI) technologies¹. As these changes and advancements emerge, it is important to understand student radiographers' thoughts on AI and its potential impact on their future development. More importantly, do students perceive AI as a benefit to their professional development or a potential hindrance that might influence their career choices, such as deterring them from working in reporting radiography? In 2022, the Clinical Radiology Workforce Census² reported that "81% of radiology departments used reporting radiographers to report images". The possibility of a decline in interest in reporting radiography could raise concerns about future staffing.

Aim

To evaluate student radiographers' attitudes towards AI and its potential impact on their career choices, specifically in the context of selecting future modalities, such as reporting radiography.

Objectives

To analyse survey data to determine whether there is a link between the development of AI and its influence on (either encouraging or deterring) the professional development of radiography students.

Literature review

In response to the ongoing expansion of AI in radiography and radiology practice, a growing body of research has emerged examining practitioners' perspectives. Examining these recent articles is critical for identifying trends in practitioners' opinions and understanding the perspectives of student radiographers. As AI continues to shape daily practice, radiographers' roles may change. This is especially important for student radiographers, who may lack a baseline for comparison in the face of such dynamic changes. The goal of this literature analysis is to explore current practitioners' perspectives on AI and to discuss how these perspectives may impact and shape the outlook of student radiographers.

Radiographers and reporting with AI

Coakley et al.³ conducted a study involving 96 participants to explore radiographers' opinions on AI in radiography. They found that 78 participants believed it "unlikely that AI would replace radiographers," which might be attributed to the patient care aspect of the role³. However, 69 participants felt that "AI would affect the interpretation of images", though the study did not clarify whether this impact was perceived as positive or negative in relation to the role of reporting radiographers³. In addition, 70 participants believed that the implementation of AI would contribute to the expansion of the radiographer's role³.

In 2022, Rainey et al.⁴ published a study on whether reporting radiographers would trust AI interpretation of images. It reported a mean trust score of 5.28 (out of 10) in AI, which indicated a level of caution about AI⁴. However, 69.8% of participants indicated they would seek a second opinion when AI disagreed with their interpretation⁴. Both the Rainey et al.⁴ and the Coakley et al.³ studies show that AI has the potential to influence the decision-making of radiographers^3,4.

Orešković et al.⁵ published a study about radiologists' attitudes towards AI, reporting a mean "confidence in AI" score of 3.5 out of five (five being "strongly agree") and showing a similar level of caution to that expressed by the radiographers in the Rainey et al.⁴ study⁵. However, "more than 70% of respondents believed AI could improve the quality of radiological examinations, enabling their more reliable interpretation" (Orešković et al.⁵). This suggests AI might help them with their job role and corresponds with the high percentage of radiographers who would seek a second opinion in Rainey et al.⁴ However, the Orešković study focused on radiologists rather than radiographers and considered the situation in Croatia and Slovenia, so it might not reflect the opinions of radiographers in the UK⁵. Also, these studies have the limitation that students were not included. However, from the available information, it can be concluded that AI will influence the way radiographers practise because it can act as a way to expand the role of the radiographer or make certain tasks easier.

Radiographers and general practice with AI

Rainey et al.⁶ previously published a study in 2021 that looked into radiographers' attitudes towards AI. It surveyed 411 participants, including diagnostic students (19.6%) and radiotherapy students (13.2%)⁶. It found that 57% (diagnostic) and 49% (radiotherapy) of the participants said they did not feel adequately trained to implement AI in a clinical setting and "furthermore 52% and 64%, respectively, said they have not developed any skill in AI"⁶. This shows a sense of anxiety that some radiographers have around AI and this might influence students⁶. Students might avoid modalities that use AI or could be unlikely to use AI in the future if they are not confident they are implanted into their practice. Coakley et al.³ highlighted similar concerns as 55 out of 96 participants were concerned about the ethical issues around AI integration³. However, there is some contrast as 79 participants were excited about AI in radiography³. Even though Coakley et al.³ did not have student participants and Rainey et al.⁶ did not have a large number of student participants, they do provide some insight into the opinions of students^3,6. This is because radiographers will influence the perspectives of the students they mentor.

Akudjedu et al.⁷ conducted a study of a total of 314 participants from across the world. However, 54.1% of the respondents were from North America and there were only 12 radiography students among the 314 total⁷. This means the study might not reflect the views of UK radiography students well because not many of them took part⁷. However, it did highlight similar concerns to those raised in the work of Coakley et al.³ and Rainey et al.⁶, with only 26.2% of respondents feeling well prepared to implement new AI technologies into their daily practice⁷. Nevertheless, a majority of respondents (more than 67%) in Akudjedu et al.⁷ believed that AI would change daily clinical radiography practice even though "a proportion of the respondents agree (n = 76, 26.2%) or somewhat agree (n = 70, 24.1%) that they feel well prepared to implement new AI technologies and innovations in daily practice"⁷. Akudjedu et al.⁷ also found that only 29.7% of respondents somewhat agreed that the implementation of AI would make the radiography profession more attractive to them, which is significantly different to the 79 out of 96 participants in Coakley et al.³ who were excited about AI^3,7.

Chen et al.⁸ conducted a qualitative study that interviewed 12 radiologists and 14 radiographers in breast units across the National Health Service about their attitudes towards AI⁸. They found that the radiologists viewed AI as an opportunity for professional development because they believed AI could automate routine tasks⁸. Both radiologists and radiographers identified that AI technology could change their roles, a view also expressed by 67.6% of the radiographers in the study by Akudjedu et al.⁷ Still, radiographers also expressed concerns about the impact of AI on their skills and decision-making⁸. This was also highlighted in the Rainey et al.⁴ study, which considered how AI might affect whether they would get a second opinion⁴. An ethical concern raised by participants in the study by Coakley et al.³ was the possibility of diagnostic errors due to AI³. However, this study did not include radiography students and the other studies mentioned either had none or very few.

Medical students' attitudes towards AI

To gain insights into the perspectives of radiography students and the potential impact of AI on their career choices, this review expanded its scope to encompass medical students. This broader perspective was adopted due to the limited research available that specifically focused on radiography students. Sit et al.⁹ conducted a study in the UK with 484 participants, and 49.2% reported they were less likely to consider a career in radiology due to the impact AI may have on the field⁹. It also found that only 10.4% of students felt confident in using AI tools⁹. Radiology trainees in Hashmi et al.¹⁰ expressed similar concerns, with 47.6% of participants worried that AI might replace the job of radiologists¹⁰. Hashmi et al.¹⁰ did find contrasting opinions as their study included 149 trainee radiologists, 83.7% of whom expressed an interest in the use of AI in radiology¹⁰. Some 74.2% believed that AI would enhance the job role of diagnostic radiologists in the future¹⁰. This is similar to what the radiologists said in the study by Chen et al.⁸ Again, a conclusion can be drawn that AI will influence the way radiographers work because it may enhance areas of practice.

Conclusion

Based on the literature review, it is evident that the development of AI is influencing perceptions and raising concerns among radiographers, radiologists and medical students. However, when it comes to the impact of AI on radiography students specifically, there is a noticeable gap in the literature. Many existing studies either did not include radiography students or included too few participants to ensure the generalisability of the findings. Therefore, the research question guiding this study is: "What are diagnostic radiography students' perceptions of AI in radiography?"

Method

Approach

To answer the research question, an online survey consisting of four open-ended questions was created and distributed to diagnostic radiography students at Birmingham City University. This method was considered more convenient for reaching students than paper-based forms (according to Regmi et al.¹¹). The survey collected qualitative data to explore whether AI influenced students and whether they had any concerns regarding its use. The use of open-ended questions allowed for a deeper understanding of diagnostic radiography students' perceptions of AI.

Consent and information

The online survey included a consent form and participant guidance in the form of an information leaflet to ensure that participants understood the purpose of the research and provided informed consent. The participant leaflet outlined the nature of the survey (in accordance with Principle Eight of the Caldicott Principles¹²), any potential risks involved and provided contact details for further inquiries. This process ensured that the participants gave their informed consent as they were made aware of both the benefits and risks of the study in line with Principle 26 of the Declaration of Helsinki.

Results

The overall response rate was 36, which included only the participants who gave consent to take part in the survey. This number of responses was sufficient for qualitative research, providing meaningful insight into students' perspectives on AI in radiography (according to Creswell¹⁴) and allowing for the exploration of relevant concepts.

Thematic analysis

Using the coding table below, key themes were identified regarding participants' views on AI in relation to modalities.

Table 1. Codes and themes

Discussion

Patient care

In their responses, students emphasised the importance of patient care alongside the limitations of AI compared with radiographers in various aspects of their roles. While acknowledging that AI can improve certain aspects of the job, many students expressed the belief that the role of a radiographer in patient-facing modalities would adapt to AI, rather than being replaced by it. Participant 11 said: "... every patient has different needs, different requirements, so human touch cannot be replaced..." The general consensus from responses was that radiographers cannot be completely replaced in roles involving patient care and safety, which mitigates concerns about the potential impact of AI on career advancement in these fields. Radiographers will still be responsible for maintaining high standards of patient care and safety in accordance with the Health and Care Professions Council (HCPC) Standard 1.4. Similarly, Orešković et al.⁵ found that only 13% of practitioners would determine a patient's treatment plan based on AI recommendations. This finding was consistent with the perceptions of participants, who believed that patient care was ultimately the radiographer's responsibility. In contrast, Coakley et al.³ reported that only 53% of participants believed AI would not impact aspects of communication and patient care. This perspective aligns with findings from Rainey et al.⁶ where only 30% of respondents felt confident explaining AI to patients and carers. As AI technologies are integrated into practice, radiographers will be expected to play a crucial role in explaining the implications to patients, aligning with the patient care and informed consent principles outlined in HCPC Standard 1.4¹⁵.

Enhancement

Many participants expressed the belief that AI would improve various aspects of radiography across all modalities. Furthermore, some suggested that AI could make the job more appealing by simplifying specific tasks with its assistance. Participant 9 said: "AI algorithms can be applied to radiographs and interpret the data, so a broader range of image qualities can still be 'diagnostic'..." Many participants agreed that AI could help radiographers reduce their workload without jeopardising the profession. They noted that AI could perform simple tasks, speed up processes and improve image quality. Similarly, Orešković et al.⁵ found that more than 70% of participants agreed that "AI tools can improve the quality of examinations and enable a more reliable interpretation". Many students shared this belief, emphasising AI's potential to improve image interpretation and quality. This finding aligns with Coakley et al.³, who found that 71% of radiographers believed AI would improve daily work and 73% believed it would contribute to the expansion of radiographer roles. This trend was also observed in Akudjedu et al.⁷, where the majority of respondents (32.8%) agreed that AI would reduce workload. However, concerns were raised as participants acknowledged that AI could pose a threat to the radiography profession. Students also expressed this concern, with many worried that AI could replace certain aspects of the radiographer's role.

Replacement

Concerns were raised about the potential negative impact of AI on modalities. The participants expressed the fear that AI could replace the role of the radiographer in certain aspects. However, some viewed this replacement as beneficial, seeing it as an opportunity to open up new possibilities for them. Participant 17 said: "... AI in reporting might soon replace part of the work of reporting radiographers, and the same goes to the reporting part of the sonographer role..." Several students expressed concerns about reporting and image interpretation, with some claiming that AI's pattern recognition could replace the radiographer's role in reporting. Furthermore, there were suggestions that the introduction of automated positioning could change the role of the radiographer in modalities such as computed tomography and magnetic resonance imaging. This apprehension aligned with findings from Sit et al.⁹, where 49% of medical students reported being less likely to consider a career in radiology due to the fear of being replaced. Interestingly, the same study found that 88.8% of students thought AI would be beneficial to their careers. This sentiment is consistent with Hashmi et al.¹⁰, where 74.2% of trainees agreed that AI would improve the job of radiologists. In conclusion, while participants expressed concern about the potential replacement of certain aspects of their job by AI, many were confident in their ability to adapt to change, with a general belief that the implementation of AI would ultimately benefit the radiographer's role.

"Students expressed the concern that AI could replace certain aspects of the radiographer's role"

Outlier analysis

Participant 24 offered an insightful perspective on AI, expressing concerns that it might exacerbate existing issues of laziness among radiographers, potentially leading to improved image quality for those currently producing poor images. The participant also mentioned the possibility of AI making reporting easier by potentially accelerating the process, but radiographers would rely on incidental findings and simply agree with AI-generated results. This viewpoint could be interpreted as both an enhancement, with the potential for improved image quality and speed, and a replacement, as it suggests that some reporting radiographers may rely heavily on AI-generated reports rather than actively seeking findings themselves. Participant 24 said: "[It] will make radiographers that are already lazy even lazier. Radiographers currently performing poor standard images will produce better ones. [It] will make reporting easier for reporters as they just need to look for incidental findings and agree with AI reports so it could increase speed." Furthermore, this raises concerns about patient care, implying that the implementation of AI may lead to complacency among radiographers, potentially resulting in poor patient care. This dual perspective, which acknowledges both the potential benefits and risks, emphasises the complexities of incorporating AI into radiography and the importance of carefully considering the implications for professional practices and patient outcomes.

Conclusion

The qualitative analysis revealed that students believe AI in radiography has the potential to improve various aspects of the profession, particularly efficiency and image quality. While concerns were raised about the possible replacement of certain elements of the radiographer's role, participants generally expressed confidence that radiographers would be able to adapt to these changes and effectively incorporate the benefits of AI into their practice.

Acknowledgement

Thanks to Kate Chadwick, Senior Lecturer in Radiotherapy at Birmingham City University, for supervising this research.