Send Orders for Reprints to reprints@benthamscience.

net
921

Current Medical Imaging, 2023, 19, 921-930

RESEARCH ARTICLE
ISSN: 1573-4056
eISSN: 1875-6603

Current
Medical Imaging

Medical Students’ Perspectives on Artificial Intelligence in Radiology:

The Current Understanding and Impact on Radiology as a Future
Specialty Choice
BENTHAM
SCIENCE

Ali Alamer1,*

1
Department of Radiology, College of Medicine, Qassim University, Buraidah 6655-51452, Saudi Arabia

Abstract: Background: Medical students' career choices and motivations might be significantly im-
pacted by the rapid advances in artificial intelligence (AI) and the recent hype around it.
Objective: This study aimed to assess the impact of AI on medical students’ preferences for radiology
as a future specialty choice.
Methods: A cross-sectional study was conducted between October and December 2021 among all
ARTICLE H I S T O R Y
medical students in the three regional medical colleges in Al-Qassim Province, Saudi Arabia.
Results: The survey resulted in 319 complete responses. Among the respondents, 26.96% considered
Received: June 28, 2022
Revised: July 24, 2022 radiology to be one of their top three future specialty choices. Only a minority of the respondents
Accepted: August 02, 2022
(23.2%) believed that radiologists would be replaced by AI during their lifetime. The misperceptions
DOI: of the potential impact of AI led 22.26% of the students to be less likely to consider a career in radiol-
10.2174/1573405618666220907111422

ogy. Students with an interest in radiology were less influenced by such misperceptions (p=.01). Based
on self-reported confidence measures, the basic understanding of AI was higher among students with
an interest in radiology and students with prior exposure to AI (p<.05).
  icalImaging

Conclusion: The students' preferences for radiology as a future specialty choice were influenced by
their misperceptions of the potential impact of AI on the discipline. Students' interest in radiology and
prior exposure to AI helped them grasp AI and eliminate the hype around it.

Keywords: Artificial intelligence, radiology, medical imaging, machine learning, undergraduate, medical students.

1. INTRODUCTION AI applications, as imaging analysis and interpretation are the

fundamental cognitive tasks of an interpreting radiologist [1].
The terms artificial intelligence (AI), machine learning
Several DL solutions have been proposed for the interpreta-
(ML), and deep learning (DL) are sometimes used inter-
tion of different radiological modalities, including radiog-
changeably in the field of information technology to de-
raphy, ultrasound, computed tomography (CT), and magnetic
scribe software that behaves intelligently. AI is a broad term
resonance imaging. The accuracy of such algorithms has been
that refers to a variety of algorithms that enable computers
compared to interpreting radiologists through different peer-
to accomplish tasks that would normally require a human's reviewed publications. For the reference, a DL algorithm has
cognitive ability [1]. ML is a subfield of AI in which the
been tested in the detection of abnormalities on routine chest
algorithms are trained through data rather than by explicit
X-rays [6]. A DL model has also been used to assess mam-
programming [2]. DL is a type of ML that relies on multiple
mographic breast density [7]. Furthermore, the effectiveness
processing layers to learn and make intelligent decisions
of DL algorithms has been evaluated in the assessment of
regarding the given data [3]. Such innovative techniques
acute neurological events on CT, such as the detection of
have dramatically improved a variety of complex and con- acute intracranial haemorrhage [8-10]. The AI solutions in
figurable tasks, including speech recognition, visual object
radiology are not limited to image interpretation and include
detection, and image classification [3, 4].
worklist management, image enhancement, automated meas-
The recent advancements in AI technology have drawn urements, boundary tracking, standard plane recognition, and
much attention for its application in medicine. Although in departmental operations [11-13].
its early stage, it is expected that almost all medical special- Such rapid technological developments have introduced
ties will be using AI technology in the near future [5]. Radi-
AI as a hot topic for discussion in radiology concerning its
ology, in particular, is a potential and attractive field for
potential impact on the future of the field [14, 15]. The dis-
cussion is sometimes based on the recent hype around AI
*Address correspondence to this author at the Department of Radiology, [16]. Many people, including some health care practitioners,
College of Medicine, Qassim University, Buraidah 6655-51452, Saudi believe that AI will dominate the medical field in the near
Arabia; E-mail: ali.alamer@qu.edu.sa

1875-6603/23 $65.00+.00 © 2023 Bentham Science Publishers

922 Current Medical Imaging, 2023, Vol. 19, No. 8 Ali Alamer

future and may replace radiologists or other clinicians. A reminder for participation was sent to the group leaders two
growing body of literature has also emerged on the percep- weeks after the initial invitation. Survey participation was
tions and attitudes towards AI among radiologists whose closed on December 19, 2021. SurveyMonkey (SVMK Inc.,
profession is already influenced by AI [11, 17]. Some radi- San Mateo, CA, United States) was used to design the web-
ologists are uncertain about their future professional careers based survey. All medical students in the preclinical (1st to
given the advancement of AI [17]. In addition, it is not un- 3rd years) and clinical phases (4th to 5th years) at the three
common for practicing radiologists to be approached by colleges were approached. The new group of interns who
medical students with doubts about the potential impact of had spent less than three months in the internship was also
AI on the profession. Hence, it remains unclear whether included. Participation was voluntary and anonymous, with-
medical students are generally concerned about the rapid out incentives or rewards. The main purpose of the study
advancements in AI and the hype around it, which might was explained to the participants. A request for consent to
significantly impact their motives and career choices. participate was also provided at the beginning of the survey.
Unfortunately, only a few studies have thus far explored
2.2. Survey Items
medical students’ perceptions of AI in the field of radiology
and its impact on their future career preferences. According The survey items were previously validated by Sit et al.
to prior initial surveys, 15.2% to 29.3% of medical students in their study evaluating UK medical students’ perceptions
believed that AI would replace radiologists in the foreseea- of AI in the field of radiology [20]. Additional items related
ble future [18, 19]. Such rumours and myths surrounding AI to students’ career interests were also utilized from Gong et
and the radiology field increased anxiety for students about al’s study and used as a variable to assess the impact of AI
choosing radiology as a career [19]. A multicentre study in on students’ preferences for radiology as a future specialty
the United Kingdom (UK) reported that almost half of the choice [19]. The newly arranged survey has undergone fur-
medical students (49.2%) were less interested in pursuing a ther pre-testing with students who represent the target par-
career in radiology solely because of AI [20]. Such results ticipants. The results of the pre-test were not incorporated
were confirmed by recent trends of sharp declines in Cana- into the final results. The final survey includes 21 items that
dian students applying to residency programs, with radiolo- were rearranged into five sections.
gy being their first or only choice [21]. The study concluded
The first section consisted of demographic-related data,
that the fear of replacement by AI in the future was one of
including medical college, year level, and preference for
the contributing factors to this decline. Since the impact of radiology as a future specialty choice, both with and without
AI on students’ preferences for radiology as a future special-
regard to the potential impact of AI. The second, third, and
ty choice remains unclear, particularly in an area with lim-
fourth sections consisted of five-point Likert scale items to
ited AI applications in clinical practice and a lack of dedi-
assess the students’ overall perceptions of AI, impact of AI
cated training in medical curricula, it is crucial to fill this
on students’ preferences for radiology as a future specialty
knowledge gap. This study aimed to assess the impact of AI
choice, current understanding of AI, and future aspirations.
on medical students’ preferences for radiology as a special- The final section addressed prior exposure to AI.
ty, regardless of their college year level, future career pref-
erence, and prior exposure to AI. 2.3. Data Collection and Statistical Analysis
2. MATERIALS AND METHODS The data collected from the survey items were recorded
in a Microsoft Excel spreadsheet (Microsoft Corporation,
2.1. Study Design Redmond, Washington). The descriptive and inferential
A quantitative-based cross-sectional study was ethically statistics were calculated using STATA 16 SE (StataCorp
approved by the Committee of Health Research Ethics in the LLC, College Station, TX, USA). Tableau V.2021.3 (Tab-
Deanship of Scientific Research at Qassim University, Saudi leau, Seattle, WA, USA) and Excel were also utilized for
Arabia (Reference number 21-03-07). The STROBE guide- graphics. Students’ perceptions were correlated with their
lines have been implemented. All medical students in the college year level, preference for radiology as a future spe-
three regional medical colleges in Al-Qassim Province were cialty choice, and prior training in AI. For the analysis of the
invited to participate in the study, including two governmen- Likert responses, strongly agree and agree replies were
tal colleges and one private college. The three colleges are merged into one group. Another group was formed by com-
adopting a five-year medical degree program. Radiology is a bining the strongly disagree and disagree replies. A chi-
compulsory clerkship for fourth-year medical students and is square test of independence was used to determine if a sta-
taught either as a separate two-week clerkship or weekly 2- tistically significant relationship existed between two cate-
hour lectures throughout a whole semester. Moreover, the gorical variables. An unpaired two-tailed Wilcoxon rank-
students were also exposed to radiology even prior to this sum (Mann-Whitney) test was also conducted for compari-
clerkship, where it was integrated with other courses during sons between different groups. A p value of ≤ .05 was con-
the preclinical phase (1st to 3rd years) to teach radiological sidered to be statistically significant.
anatomy and basic X-ray and CT interpretation skills.
3. RESULTS
Between October and December 2021, the group lead-
ers for all three colleges were contacted, and the online sur- 3.1 Respondents’ Characteristics
vey was sent to them. The leaders then volunteered to for-
Over an eight-week period, medical students from all
ward the survey to their corresponding student groups
three regional medical colleges submitted a total of 319
through various methods, including their official emails. A
Medical Students’ Perspectives on Artificial Intelligence in Radiology Current Medical Imaging, 2023, Vol. 19, No. 8 923

complete responses. Among the respondents were junior ents (n=245, 76.8%) agreed that receiving teaching in AI
(n=150, 47.02%) and senior (n=169, 52.98%) medical stu- would be useful for their future careers. For that reason,
dents. Junior medical students included 1st-year (n=43, almost two-thirds of the respondents (n=212, 66.46%)
13.48%), 2nd-year (n=39, 12.23%), and 3rd-year (n=68, agreed that all medical students should receive teaching in
21.32%) students. Senior medical students included 4th-year AI. Such agreement was higher among junior students
students (n=81, 25.39%), 5th-year students (n=47, 14.73%), (72.67%) and students who received AI training (84.61%)
and interns (n=41, 12.85%). There was no statistically sig- than among senior students (60.95%) and students who did
nificant difference between junior and senior medical stu- not receive AI training (65.69%), which was statistically
dents among the respondents (p=.29). significant (p=.02 and p=.04, respectively). There was no
significant difference found between students’ overall per-
The vast majority of the respondents (n=306, 95.92%)
did not receive any dedicated teaching or training in AI. ceptions of AI and their preferences for radiology as a future
specialty choice.
Among the 13 respondents who had received prior training
in AI, 69.23% were senior medical students, 84.62% report-
3.3. Impact of AI on Students’ Preferences for Radiology
ed that the training was not a compulsory part of their medi-
as a Future Specialty Choice
cal degree, and 92.31% found it to be useful.
Radiology was among the top three future specialty More than half of the respondents (n=176, 55.17%)
agreed that some specialties would be eventually replaced
choices in 26.96% (n=86) of the respondents; 4.7% (n=15)
by AI during their lifetime (Fig. 3). For the field of radiolo-
ranked radiology as their first specialty choice, 6.58%
gy, less than one-quarter of the respondents (n=74, 23.2%)
(n=21) ranked it as their second choice, and 15.67% (n=50)
agreed that radiologists would be replaced by AI during
ranked it as their third choice. However, radiology was
their lifetime. In contrast, the majority of the respondents
ranked below the third choice in 34.80% (n=111) of the re-
spondents, and 38.24% (n=122) were not interested in radi- (n=257, 80.56%) agreed that AI would augment radiolo-
gists’ capabilities and make radiologists more efficient.
ology. Table 1 provides a detailed breakdown of the re-
spondents’ characteristics. Less than one-quarter of the respondents (n=71,
22.26%) agreed that they were less likely to consider a ca-
The preference for radiology as one of the top three fu-
reer in radiology given the advancement of AI, while
ture specialty choices was higher among senior medical stu-
dents (17.55%) than among junior medical students (9.40%), 41.69% (n=133) and 36.05% (n=155) were neutral and dis-
agreed with this statement, respectively. Such advancements
which was statistically significant (p<.001), as shown in Fig.
in AI and its uncertain impact on the field of radiology made
(1). As all medical students from all three regional medical
34.17% of the respondents (n=109) worried about choosing
colleges were approached (n=1413), the approximate conven-
radiology as a career. Among the respondents who chose
tional response rate was calculated to be 22.58%. Based on
radiology as one of their top three future specialty choices,
the number of respondents who chose radiology as one of
their top three future specialty choices (n=86), the relevant 48.84% disagreed with the potential impact of AI on radiol-
ogy career selections compared to 31.33% who were not
response rate was calculated to be 26.96%.
interested in radiology or radiology was below their third
3.2 Students’ Overall Perceptions of AI choice (p=.01), as shown in Fig. (4). There were no signifi-
cant differences found between students’ perceptions of the
The majority of respondents (n=281, 88.09%) agreed potential impact of AI on radiology and their college year
that AI would play a significant role in the future of level or prior exposure to AI.
healthcare (Fig. 2). Furthermore, the majority of respond-

Table 1. Characteristics of the respondents.

. Number of Responses Percentages -

Student year level:
1st-year 43 13.48% Junior
2nd-year 39 12.23% (n=150);
3rd-year 68 21.32% 47.02
4th-year 81 25.39% Senior
5th-year 47 14.73% (n=169);
Intern 41 12.85% 52.98%
Prior exposure to AI:
Yes 13 4.08% -
No 306 95.92%
Radiology as a future specialty choice:
One of the top three choices 86 26.96% -
Below the third choice or not interested 233 73.04%
924 Current Medical Imaging, 2023, Vol. 19, No. 8 Ali Alamer

Junior Senior Radiology as one of their top 3 future specialty choices

No
% of students considering Radiology as one of their top three future specialty choices 40% 120 Yes
37.62%
113
35.42%
35%

30%

25%

20% 56
17.55%

15%

30
10% 9.40%

5%

0%
No Yes No Yes

Fig. (1). Students’ preferences for radiology as a future specialty choice based on their year level. (A higher resolution / colour version of this
figure is available in the electronic copy of the article).

Fig. (2). Students’ overall perceptions of AI. (A higher resolution / colour version of this figure is available in the electronic copy of the arti-
cle).
Medical Students’ Perspectives on Artificial Intelligence in Radiology Current Medical Imaging, 2023, Vol. 19, No. 8 925

Fig. (3). Impact of AI on students’ preferences for radiology as a future specialty choice. (A higher resolution / colour version of this figure is
available in the electronic copy of the article).

Fig. (4). Considering radiology as a career given the advancement of AI based on students' career preferences. (A higher resolution / colour
version of this figure is available in the electronic copy of the article).

who received prior training in AI, 53.85% reported that they

3.4. Students’ Current Understanding of AI and their
Future Aspirations understood the basic computational principles of AI,
61.54% were comfortable with its nomenclature, and
Fewer than one-third of the respondents agreed that they 76.92% understood its limitations compared to the 41.83%,
had an understanding of the basic computational principles 39.87%, and 51.96% of respondents who did not receive
of AI (n=86, 26.96%) and were comfortable with the no- prior AI training, respectively (p<.05). Furthermore, among
menclature related to AI (n=79, 24.77%). Fewer than half of the students who chose radiology as one of their top three
the respondents (n=130, 40.75%) reported that they under- future specialty choices, 40.70% reported that they under-
stood the limitations of AI (Fig. 5). Among the respondents stood the basic computational principles of AI, 39.53% were
926 Current Medical Imaging, 2023, Vol. 19, No. 8 Ali Alamer

Fig. (5). Students’ current understanding of AI and their future aspirations. (A higher resolution / colour version of this figure is available in
the electronic copy of the article).

comfortable with its nomenclature, and 51.17% understood [5, 22]. Radiology, in particular, as a digital specialty, has
its limitations compared to the 21.89%, 19.31%, and experienced dramatic revolutionary changes over the past
36.91%, respectively, of students who were not interested in decades that were driven by technological innovations, mak-
radiology or radiology was below their third choice (p<.05). ing it a rich environment for AI applications. Currently, AI
There was no significant difference found between students’ is a hot and evolving topic in the field of radiology [23]. The
perceptions of the level of understanding of AI and their primary driving force behind the development of AI in radi-
college year level. ology has been the urgent demand for improved clinical
efficacy and efficiency in managing the growing radiolo-
However, the students were more positive about their
gist's workload [24]. The number of AI-related publications
future aspirations in AI. Nearly two-thirds of the respond-
in radiology has dramatically increased [25]. As a result of
ents (n=210, 65.83%) and nearly half of the respondents
these research projects, a variety of AI applications in radi-
(n=138, 49.53%) believed that at the end of their medical
ology have been developed to enhance many aspects of ra-
degree, they would be confident utilizing basic AI tools in
healthcare if needed and that they would have a better un- diologists' daily practice, from workflow management to
image interpretation and structured reporting [26]. The re-
derstanding of the techniques used to evaluate the execution
spondents in this study were generally aware of the im-
of healthcare AI algorithms, respectively. Furthermore,
portance of AI and positively perceived its value in the fu-
more than half of the respondents (n=163, 51.1%) felt they
ture of healthcare, which is comparable to prior reports [18,
would have the knowledge required to operate AI in clinical
20].
practice at the end of their medical degree. Among the re-
spondents who received prior training in AI, 84.61% be- On the other hand, such technological advancements in
lieved that at the end of their medical degree they would be AI introduced much hype around it, including the assertion
confident utilizing basic AI tools in healthcare, and 84.61% of the replacement of radiologists and that DL will be better
believed they would have a better understanding of the than radiologists in the foreseeable future [16, 27]. Despite
techniques used to evaluate the execution of healthcare AI the fact that just a few AI applications are now used in clini-
algorithms compared to the respondents who did not receive cal practice, ongoing AI research initiatives promise to
prior AI training (65.03% and 48.03%, respectively; p<.05). adopt diverse AI tools when medical students begin their
Furthermore, junior medical students were more positive future professions [16, 28]. Over 75% of the students in one
that at the end of their medical degree, they would be confi- study believed that AI would have a significant impact on
dent utilizing basic AI tools in healthcare (71.33%) and that their careers, and 66% of them chose radiology as a special-
they would have the knowledge required to operate AI in ty that would be the first and most impacted, which again
clinical practice (60%) compared to senior medical students confirms that AI is a hot topic and these students are likely
(60.95% and 43.20%, respectively; p<.05). impacted by the discussions regarding AI [29]. Hence, this
study hypothesized that medical students are generally con-
4. DISCUSSION cerned about the impact of AI on the field of radiology and
that this concern might alter their preferences for radiology
Recent advancements in AI technology can potentially
as a future specialty choice. A review of the literature re-
improve various aspects of the current healthcare services
Medical Students’ Perspectives on Artificial Intelligence in Radiology Current Medical Imaging, 2023, Vol. 19, No. 8 927

vealed variability regarding the misperception of the re- ologists’ workloads due to the rapid acceleration in the
placement of radiologists by AI among medical students number of imaging studies, which is responsible for radiol-
[15, 30]. According to an initial study conducted in Germa- ogists’ burnout, and might result in compromising the quali-
ny in 2018, 15.2% of the students believed that AI would ty of radiological reports and subsequently affect patient
ultimately replace radiologists in the future [18]. A more care [35-38]. For instance, during the 2019 coronavirus dis-
recent study performed in Canada showed an even higher ease pandemic, solutions for image analysis were recog-
level of agreement (29.3%) with this perception [19]. The nized as one of the major battlefields for AI in the fight
respondents in this study were in the middle compared to against the rapid acceleration in radiological studies [39].
prior reports, as 23.2% of the students believed that AI Respondents from members of the ESR believed AI can
would eventually replace radiologists. save time and provide stronger interactions with other clini-
The misperception of the impact of AI led 22.26% of the cians and patients who sometimes fail to be included in the
busy daily workload of radiologists [17]. However, this
students in this study to be less likely to consider a career in
speculation has not yet been supported by evidence and can
radiology. Furthermore, 34.17% of the students were even
result in negative strategic decisions, including limiting the
worried about choosing radiology as a career due to AI. Few
number of students who can be enrolled in radiology pro-
studies have thus far explored the potential impact of AI on
grams [40].
students' preferences for radiology as a future specialty
choice. Sit et al. reported that 49.2% of their students were Although there is a lack of available objective criteria
less likely to consider a career in radiology due to AI [20]. for the assessment of an acceptable basic understanding of
AI was also responsible for the reduction in interest and AI, the literature showed variability concerning this matter
enthusiasm for the specialty in 44% of the students in an- [28]. Respondents’ understanding in this study was assessed
other study [29]. A national survey in Canada among medi- based on self-reported confidence, which showed limited
cal students interested in radiology revealed the students had understanding of the basic computational principles, no-
considerable anxiety about AI, and one-sixth of them felt menclature, and limitations of AI compared to the study
discouraged from considering radiology as a career just as a conducted by Sit et al. [20]. However, students in this study
result of the uncertain impact of AI [19]. Another multicen- showed higher self-reported confidence in their expected
tre study in Brazil reported a higher result, as 61.11% of level of understanding of AI at the end of their medical de-
their students who were interested in radiology changed gree. The knowledge of AI was more formally assessed in
their minds as a result of the potential influence of AI [31]. other studies through specific questions rather than students’
Such an impact of AI can partially explain the major issue self-reported confidence in their understanding, which re-
of a recent decline in students choosing radiology as their vealed limited knowledge [19, 41]. Not surprisingly, the
first or only choice when applying to residency programs in students with prior exposure to AI had a higher level of un-
Canada and France [21, 32]. Unfortunately, no similar study derstanding, similar to prior reports [19, 41]. Interestingly,
in Saudi Arabia has assessed the recent trends of applicants there was a significant relationship between the students
to local residency programs for radiology. Moreover, the who chose radiology as one of their top three specialty
concern with the impact of AI extends beyond students, choices and their perceived level of understanding. This
even into radiology residents and other radiology personnel finding can be explained by the fact that students with an
[28, 33, 34]. interest in radiology are positively influenced by the recent
discussions about AI and are more likely to seek more in-
One encouraging result in this study is that the majority
formation about it, which is reflected in their level of under-
of students agreed that AI would augment radiologists’ ca-
pabilities and make radiologists more efficient, which is standing. Moreover, the limited level of understanding and
awareness extends beyond students to radiology residents
comparable to prior reports [19, 22]. This confirms that the
and even other radiology personnel [33, 42, 43].
fear of the replacement of radiologists by AI has started to
fade away and remains far from reality, revealing a shallow Considering the recent advancements in AI, medical
understanding of the applications of AI at that time [16, 27]. education lags behind such technological developments
According to the European Society of Radiology (ESR) [44]. It is hard to believe that AI will be included in the dai-
eHealth and Informatics subcommittee, AI cannot replace ly practice of future physicians without being a part of the
the complex tasks of radiologists [26]. However, the daily medical curricula. Therefore, there is a strong need to incor-
clinical practice of radiologists will certainly change in the porate basic AI training into the medical curricula [18].
era of AI with the assumption of faster and better perfor- Moreover, AI education for medical students may help them
mance [26]. Radiology personnel’s views on AI have indeed develop more positive views regarding the potential impact
evolved beyond the stage of fear of being replaced to active- of AI technology on the field of radiology [15]. Dumić-Čule
ly participating in the development of AI tools. Collado- et al. performed a national survey in Croatia to assess the
Mesa et al. reported that all of their radiologists and the ma- perceptions of radiologists and radiology residents on the
jority of their trainees believed that their jobs would not be need for AI education in medical school curricula [44]. The
replaced by AI, and most of them were willing to help in the vast majority of their respondents (89.6%) agreed that edu-
development of AI tools [33]. Furthermore, respondents cation in AI should be a part of medical school curricula.
from the ESR in another study agreed that radiologists must The stronger agreement in the Dumić-Čule et al’s. study
take the lead in the development and evaluation of AI tools compared to our results (66.46%) can be explained by the
[17]. Participation in the development of AI can, in fact, fact that the practising physicians (radiologists and resi-
open up new job opportunities for radiologists with AI expe- dents) recognized the need even more clearly than the stu-
rience [23]. AI was recently proposed as a solution to radi- dents [44]. On the other hand, AI is a topic that is not yet
928 Current Medical Imaging, 2023, Vol. 19, No. 8 Ali Alamer

formally taught in some radiology residency programs [33, CONCLUSION

42]. Although some medical colleges have recently started
Medical students in this study were generally aware of the
delivering AI-dedicated elective courses for their students,
role of AI in the future of healthcare, which is comparable to
the medical curricula are not yet ready to accommodate suf-
prior reports. However, the misperceptions of the potential
ficient educational requirements in AI for different reasons,
including a busy curriculum and lack of experience [16]. impact of AI on the field of radiology, including the fear of
the replacement of radiologists by AI, influenced the medical
Hedderich et al. offered an educational program on AI in
students’ preferences for radiology as a future specialty
radiology for medical professionals, including medical stu-
choice. Students' interest in radiology and prior exposure to
dents [45]. The program was perceived well, improved
AI helped them grasp AI and eliminate the hype around it.
skills, and reduced optimistic perception of AI. They con-
Furthermore, the students' interest in radiology grew as a re-
cluded that such educational activities should be integrated
into medical schools and residency programs. Additionally, sult of their undergraduate radiology exposure. Therefore,
there is a strong need to increase radiology exposure and in-
a recent initiative to practically integrate AI technology in
corporate basic AI training into the medical schools' curricula.
medical education was made by Cheng et al. when they cre-
Radiology educators should address the current and future
ated an AI-based medical image learning system to detect
myths surrounding the specialty with their students, which
hip fracture on pelvic radiograph [46]. They concluded that
might impact their preferences for radiology as a career.
AI is practical for enhancing medical education and can
accelerate the learning curve.
LIST OF ABBREVIATIONS
The results of this study showed that senior students
AI = Artificial Intelligence
with more exposure to radiology have more interest in the
field, which subsequently reduces their negative influence CT = Computed Tomography
by the potential impact of AI. According to a multicentre DL = Deep Learning
study conducted in the United States, AI dramatically de-
creased students' preferences for radiology, and such per- ESR = European Society of Radiology
ception was linked to reduced perceived radiology under- ML = Machine Learning
standing [47]. The recent attempts to increase students’ ex-
posure to radiology in the medical curricula have been UK = United Kingdom
shown to correlate well with decreased anxiety about AI and
increased interest in radiology [19]. Grimm et al. performed ETHICS APPROVAL AND CONSENT TO
a study on students’ perceptions of radiology stereotyping PARTICIPATE
based on different levels of exposure to radiology [48]. The study was approved by the Health Research Ethics
They suggested that radiology exposure in medical educa- Committee in the Deanship of Scientific Research at Qassim
tion eliminated various stereotypes, including those regard- University (Reference number 21-03-07 on October 18, 2021).
ing AI, and they recommended active participation in radi-
ology teaching for undergraduate medical students. Tradi- HUMAN AND ANIMAL RIGHTS
tional teaching methods of radiology should be replaced by
innovative learning techniques, such as case-based ap- No animals were used that are basis of this study. A
proaches, to encourage medical student’s interest in radiolo- request for consent to participate was also provided at the
gy. Furthermore, radiology educators should address the beginning of the survey. The study was conducted in ac-
current and future myths surrounding the specialty with cordance with the Declaration of Helsinki principles.
their students, which might impact their preferences for ra-
diology as a career [49]. CONSENT FOR PUBLICATION

There are intrinsic limitations in the study. The potential Informed consent was obtained from all individual par-
impact of nonresponse bias based on the voluntary nature of ticipants included in the study.
the data collection cannot be totally eliminated. However,
the relatively acceptable response rate (22.58%) from the STANDARDS OF REPORTING
total population promises to overcome such limitation. An- STROBE guidelines were followed.
other limitation source includes the fact that such results
pertain to regional institutions, limiting the generalization of AVAILABILITY OF DATA AND MATERIALS
the results. A national survey is recommended as future
work to capture medical students’ perceptions of AI, regard- The data that support the findings of this study are avail-
less of their difference in exposure to AI in medical curricu- able from the corresponding author [A.A.], on special re-
la. Furthermore, gender was not addressed in this study, quest.
which can potentially influence career preference [50]. Stu-
FUNDING
dents' preferences for radiology as a future specialty may be
influenced by factors other than AI, such as work-life bal- None.
ance or job prospects, which may affect the results of this
study [51]. Finally, the students’ levels of understanding CONFLICT OF INTEREST
were based on self-reported confidence rather than objective
The author declares no conflict of interest, financial or
assessment.
otherwise.
Medical Students’ Perspectives on Artificial Intelligence in Radiology Current Medical Imaging, 2023, Vol. 19, No. 8 929

ACKNOWLEDGEMENTS [17] European Society of Radiology (ESR). Impact of artificial intelli-

gence on radiology: A EuroAIM survey among members of the euro-
Researcher would like to thank the Deanship of Scien- pean society of radiology. Insights Imaging 2019; 10(1): 105.
tific Research, Qassim University for the support of this http://dx.doi.org/10.1186/s13244-019-0798-3 PMID: 31673823
project. The researcher would also like to acknowledge the [18] Pinto Dos Santos D, Giese D, Brodehl S, et al. Medical students’
attitude towards artificial intelligence: A multicentre survey. Eur Ra-
group leaders who voluntarily helped in distributing the diol 2019; 29(4): 1640-6.
survey to their corresponding student groups. http://dx.doi.org/10.1007/s00330-018-5601-1 PMID: 29980928
[19] Gong B, Nugent JP, Guest W, et al. Influence of artificial intelligence
REFERENCES on Canadian medical students’ preference for radiology specialty: A
national survey study. Acad Radiol 2019; 26(4): 566-77.
[1] Chartrand G, Cheng PM, Vorontsov E, et al. Deep learning: A primer http://dx.doi.org/10.1016/j.acra.2018.10.007 PMID: 30424998
for radiologists. Radiographics 2017; 37(7): 2113-31. [20] Sit C, Srinivasan R, Amlani A, et al. Attitudes and perceptions of UK
http://dx.doi.org/10.1148/rg.2017170077 PMID: 29131760 medical students towards artificial intelligence and radiology: A mul-
[2] Erickson BJ, Korfiatis P, Akkus Z, Kline TL. Machine learning for ticentre survey. Insights Imaging 2020; 11(1): 14.
medical imaging. Radiographics 2017; 37(2): 505-15. http://dx.doi.org/10.1186/s13244-019-0830-7 PMID: 32025951
http://dx.doi.org/10.1148/rg.2017160130 PMID: 28212054 [21] Li D, Yi PH, Islam N, Verma R, McInnes MDF. Diagnostic radiology
[3] LeCun Y, Bengio Y, Hinton G. Deep learning. Nature 2015; residency application trends: Canadian match results from 2010-2020.
521(7553): 436-44. Can Assoc Radiol J 2021; 72(4): 645-50.
http://dx.doi.org/10.1038/nature14539 PMID: 26017442 http://dx.doi.org/10.1177/0846537120971745 PMID: 33226836
[4] Chen J, Yang N, Zhou M, Zhang Z, Yang X. A configurable deep [22] Yang L, Ene IC, Arabi Belaghi R, Koff D, Stein N, Santaguida P.
learning framework for medical image analysis. Neural Comput Appl Stakeholders’ perspectives on the future of artificial intelligence in
2022; 34(10): 7375-92. radiology: A scoping review. Eur Radiol 2022; 32(3): 1477-95.
http://dx.doi.org/10.1007/s00521-021-06873-z http://dx.doi.org/10.1007/s00330-021-08214-z PMID: 34545445
[5] Topol EJ. High-performance medicine: The convergence of human [23] Pakdemirli E. New subspecialty or close ally emerging on the radiol-
and artificial intelligence. Nat Med 2019; 25(1): 44-56. ogy horizon: Artificial intelligence. Acta Radiol Open 2019; 8(3): 1-
http://dx.doi.org/10.1038/s41591-018-0300-7 PMID: 30617339 2.
[6] Singh R, Kalra MK, Nitiwarangkul C, et al. Deep learning in chest http://dx.doi.org/10.1177/2058460119837498 PMID: 30923633
radiography: Detection of findings and presence of change. PLoS One [24] Hosny A, Parmar C, Quackenbush J, Schwartz LH, Aerts HJWL.
2018; 13(10): e0204155. Artificial intelligence in radiology. Nat Rev Cancer 2018; 18(8): 500-
http://dx.doi.org/10.1371/journal.pone.0204155 PMID: 30286097 10.
[7] Lehman CD, Yala A, Schuster T, et al. Mammographic breast density http://dx.doi.org/10.1038/s41568-018-0016-5 PMID: 29777175
assessment using deep learning: Clinical implementation. Radiology [25] Bluemke DA, Moy L, Bredella MA, et al. Assessing radiology re-
2019; 290(1): 52-8. search on artificial intelligence: A brief guide for authors, reviewers,
http://dx.doi.org/10.1148/radiol.2018180694 PMID: 30325282 and readers—From the radiology editorial board. Radiology 2020;
[8] Arbabshirani MR, Fornwalt BK, Mongelluzzo GJ, et al. Advanced 294(3): 487-9.
machine learning in action: Identification of intracranial hemorrhage http://dx.doi.org/10.1148/radiol.2019192515 PMID: 31891322
on computed tomography scans of the head with clinical workflow in- [26] European Society of Radiology (ESR). What the radiologist should
tegration. NPJ Digit Med 2018; 1(1): 9. know about artificial intelligence – An ESR white paper. Insights Im-
http://dx.doi.org/10.1038/s41746-017-0015-z PMID: 31304294 aging 2019; 10(1): 44.
[9] Chilamkurthy S, Ghosh R, Tanamala S, et al. Deep learning algo- http://dx.doi.org/10.1186/s13244-019-0738-2 PMID: 30949865
rithms for detection of critical findings in head CT scans: A retro- [27] Langlotz CP. Will artificial intelligence replace radiologists? Radiol
spective study. Lancet 2018; 392(10162): 2388-96. Artif Intell 2019; 1(3): e190058.
http://dx.doi.org/10.1016/S0140-6736(18)31645-3 PMID: 30318264 http://dx.doi.org/10.1148/ryai.2019190058 PMID: 33937794
[10] Titano JJ, Badgeley M, Schefflein J, et al. Automated deep-neural- [28] Qurashi AA, Alanazi RK, Alhazmi YM, Almohammadi AS, Alsharif
network surveillance of cranial images for acute neurologic events. WM, Alshamrani KM. Saudi radiology personnel’s perceptions of ar-
Nat Med 2018; 24(9): 1337-41. tificial intelligence implementation: A cross-sectional study. J Multi-
http://dx.doi.org/10.1038/s41591-018-0147-y PMID: 30104767 discip Healthc 2021; 14: 3225-31.
[11] Allen B, Agarwal S, Coombs L, Wald C, Dreyer K. 2020 ACR data http://dx.doi.org/10.2147/JMDH.S340786 PMID: 34848967
science institute artificial intelligence survey. J Am Coll Radiol 2021; [29] Park CJ, Yi PH, Siegel EL. Medical student perspectives on the im-
18(8): 1153-9. pact of artificial intelligence on the practice of medicine. Curr Probl
http://dx.doi.org/10.1016/j.jacr.2021.04.002 PMID: 33891859 Diagn Radiol 2021; 50(5): 614-9.
[12] Liu X, Yang L, Chen J, Yu S, Li K. Region-to-boundary deep learn- http://dx.doi.org/10.1067/j.cpradiol.2020.06.011 PMID: 32680632
ing model with multi-scale feature fusion for medical image segmen- [30] Bhandari A, Purchuri SN, Sharma C, Ibrahim M, Prior M.
tation. Biomed Signal Process Control 2022; 71: 103165. Knowledge and attitudes towards artificial intelligence in imaging: A
http://dx.doi.org/10.1016/j.bspc.2021.103165 look at the quantitative survey literature. Clin Imaging 2021; 80: 413-
[13] Pu B, Li K, Li S, Zhu N. Automatic fetal ultrasound standard plane 9.
recognition based on deep learning and IIoT. IEEE Trans Industr In- http://dx.doi.org/10.1016/j.clinimag.2021.08.004 PMID: 34537484
form 2021; 17(11): 7771-80. [31] Brandes GIG, D’Ippolito G, Azzolini AG, Meirelles G. Impact of
http://dx.doi.org/10.1109/TII.2021.3069470 artificial intelligence on the choice of radiology as a specialty by
[14] Tang A, Tam R, Cadrin-Chênevert A, et al. Canadian association of medical students from the city of São Paulo. Radiol Bras 2020; 53(3):
radiologists white paper on artificial intelligence in radiology. Can 167-70.
Assoc Radiol J 2018; 69(2): 120-35. http://dx.doi.org/10.1590/0100-3984.2019.0101 PMID: 32587424
http://dx.doi.org/10.1016/j.carj.2018.02.002 PMID: 29655580 [32] Auloge P, Garnon J, Robinson JM, et al. Interventional radiology and
[15] Santomartino SM, Yi PH. Systematic review of radiologist and medi- artificial intelligence in radiology: Is it time to enhance the vision of
cal student attitudes on the role and impact of AI in radiology. Acad our medical students? Insights Imaging 2020; 11(1): 127.
Radiol 2022; S1076-6332(21): 00624-3. http://dx.doi.org/10.1186/s13244-020-00942-y PMID: 33252702
http://dx.doi.org/10.1016/j.acra.2021.12.032 [33] Collado-Mesa F, Alvarez E, Arheart K. The role of artificial intelli-
[16] Park SH, Do KH, Kim S, Park JH, Lim YS. What should medical gence in diagnostic radiology: A survey at a single radiology residen-
students know about artificial intelligence in medicine? J Educ Eval cy training program. J Am Coll Radiol 2018; 15(12): 1753-7.
Health Prof 2019; 16: 18. http://dx.doi.org/10.1016/j.jacr.2017.12.021 PMID: 29477289
http://dx.doi.org/10.3352/jeehp.2019.16.18 PMID: 31319450 [34] Alelyani M, Alamri S, Alqahtani MS, et al. Radiology community
attitude in saudi arabia about the applications of artificial intelligence
in radiology. Healthcare 2021; 9(7): 834.
930 Current Medical Imaging, 2023, Vol. 19, No. 8 Ali Alamer

http://dx.doi.org/10.3390/healthcare9070834 PMID: 34356212 [43] Wood EA, Ange BL, Miller DD. Are we ready to integrate artificial
[35] Smith-Bindman R, Miglioretti DL, Larson EB. Rising use of diagnos- intelligence literacy into medical school curriculum: Students and
tic medical imaging in a large integrated health system. Health Aff faculty survey. J Med Educ Curric Dev 2021; 8
2008; 27(6): 1491-502. http://dx.doi.org/10.1177/23821205211024078 PMID: 34250242
http://dx.doi.org/10.1377/hlthaff.27.6.1491 PMID: 18997204 [44] Dumić-Čule I, Orešković T, Brkljačić B, Kujundžić Tiljak M,
[36] Ganeshan D, Rosenkrantz AB, Bassett RL Jr, Williams L, Lenchik L, Orešković S. The importance of introducing artificial intelligence to
Yang W. Burnout in academic radiologists in the United States. Acad the medical curriculum – assessing practitioners’ perspectives. Croat
Radiol 2020; 27(9): 1274-81. Med J 2020; 61(5): 457-64.
http://dx.doi.org/10.1016/j.acra.2019.12.029 PMID: 32037261 http://dx.doi.org/10.3325/cmj.2020.61.457 PMID: 33150764
[37] Krupinski EA, Berbaum KS, Caldwell RT, Schartz KM, Kim J. Long [45] Hedderich DM, Keicher M, Wiestler B, et al. AI for doctors—A
radiology workdays reduce detection and accommodation accuracy. J course to educate medical professionals in artificial intelligence for
Am Coll Radiol 2010; 7(9): 698-704. medical imaging. Healthcare 2021; 9(10): 1278.
http://dx.doi.org/10.1016/j.jacr.2010.03.004 PMID: 20816631 http://dx.doi.org/10.3390/healthcare9101278 PMID: 34682958
[38] Alexander A, Jiang A, Ferreira C, Zurkiya D. An intelligent future for [46] Cheng CT, Chen CC, Fu CY, et al. Artificial intelligence-based edu-
medical imaging: A market outlook on artificial intelligence for med- cation assists medical students’ interpretation of hip fracture. Insights
ical imaging. J Am Coll Radiol 2020; 17(1): 165-70. Imaging 2020; 11(1): 119.
http://dx.doi.org/10.1016/j.jacr.2019.07.019 PMID: 31918875 http://dx.doi.org/10.1186/s13244-020-00932-0 PMID: 33226480
[39] Chen J, Li K, Zhang Z, Li K, Yu PS. A survey on applications of [47] Reeder K, Lee H. Impact of artificial intelligence on US medical
artificial intelligence in fighting against COVID-19. ACM Comput students’ choice of radiology. Clin Imaging 2022; 81: 67-71.
Surv 2022; 54(8): 1-32. http://dx.doi.org/10.1016/j.clinimag.2021.09.018 PMID: 34619566
http://dx.doi.org/10.1145/3465398 [48] Grimm LJ, Fish LJ, Carrico CW, et al. Radiology stereotypes, appli-
[40] Kwee TC, Kwee RM. Workload of diagnostic radiologists in the cation barriers, and hospital integration: A mixed-methods study of
foreseeable future based on recent scientific advances: Growth expec- medical student perceptions of radiology. Acad Radiol 2021; S1076-
tations and role of artificial intelligence. Insights Imaging 2021; 6332(21): 00378.
12(1): 88. http://dx.doi.org/10.1016/j.acra.2021.08.020
http://dx.doi.org/10.1186/s13244-021-01031-4 PMID: 34185175 [49] Lo L, Awan OA. To engage or not to engage: A new era for medical
[41] Bin Dahmash A, Alabdulkareem M, Alfutais A, et al. Artificial intel- student education in radiology. Radiographics 2020; 40(7): 1830-1.
ligence in radiology: Does it impact medical students preference for http://dx.doi.org/10.1148/rg.2020200002 PMID: 33136477
radiology as their future career? BJR|Open 2020; 2(1): 20200037. [50] Heiligers PJM. Gender differences in medical students’ motives and
http://dx.doi.org/10.1259/bjro.20200037 PMID: 33367198 career choice. BMC Med Educ 2012; 12(1): 82.
[42] Tajaldeen A, Alghamdi S. Evaluation of radiologist’s knowledge http://dx.doi.org/10.1186/1472-6920-12-82 PMID: 22913471
about the Artificial Intelligence in diagnostic radiology: A survey- [51] Sarikhani Y, Ghahramani S, Bayati M, Lotfi F, Bastani P. A thematic
based study. Acta Radiol Open 2020; 9(7) network for factors affecting the choice of specialty education by
http://dx.doi.org/10.1177/2058460120945320 PMID: 32821436 medical students: A scoping study in low-and middle-income coun-
tries. BMC Med Educ 2021; 21(1): 99.
http://dx.doi.org/10.1186/s12909-021-02539-5 PMID: 33568113