The study’s purpose was to evaluate radiologists’ perceptions and attitudes toward the adoption of artificial intelligence (AI) in their clinical practice. Understanding these attitudes is crucial for successfully integrating AI technologies in healthcare settings.
Method
A comprehensive survey was conducted among members of the SIRM Lombardy to gather data on radiologists’ attitudes toward AI. The survey was designed to cover various topics, including satisfaction with AI-based tools, openness to innovation, and optimism about AI’s future impact.
The questionnaire consisted of two sections:
Demographic and Professional Information: This section collected categorical data on the participants’ backgrounds.
Attitudes Toward AI: This section used Likert-type responses ranging from 1 to 5 (1 being extremely negative, 3 neutral, and 5 extremely positive) to assess radiologists’ views on AI.
The questionnaire underwent an iterative refinement process with expert panels and a pilot phase to ensure consistency and eliminate redundancy. Exploratory data analysis was performed using descriptive statistics and visual assessment of Likert plots. Non-parametric tests were used for subgroup comparisons to analyse specific emerging patterns thoroughly.
Results
The survey received 232 valid responses. Key findings include:
General Outlook: There is a generally optimistic outlook on AI adoption among radiologists, particularly among young radiologists (<30) and seasoned professionals (>60, p<0.01).
Daily Use of AI: Although 36.2% (84 out of 232) of the respondents reported using AI-based tools daily, only a third considered their contribution decisive (30%, 25 out of 84).
AI Literacy: AI literacy varied, with a significant proportion of radiologists feeling inadequately informed (36%, 84 out of 232), especially younger radiologists (46%, p<0.01).
Perceived Benefits: There were overwhelmingly positive attitudes towards AI’s potential to improve detection, characterise anomalies, and reduce workload (positive responses >80%), consistent across subgroups.
Scepticism in Decision-Making: Radiologists were more sceptical about AI’s role in enhancing decision-making processes, such as the choice of further investigations and personalised medicine.
AI’s Impact on the Profession: Most respondents viewed AI as an opportunity (61%, 141 out of 232) rather than a threat (18%, 42 out of 232). Additionally, many believed in AI’s relevance to future radiologists’ career choices (60%, 139 out of 232).
Concerns Among Breast Radiologists: There were specific concerns among breast radiologists (20 out of 232 responders) regarding AI’s potential impact on their profession.
Importance of Radiologists’ Final Assessment: A significant majority (84%) of respondents believe that the final assessment by a radiologist is still essential, even with AI integration.
Conclusion
The study indicates an overall positive attitude towards AI adoption in radiology, tempered by concerns about training and practical efficacy. Addressing the gaps in AI literacy, particularly among younger radiologists, is essential for the effective integration of AI. Radiologists must proactively adapt to technological advancements to fully leverage AI’s potential benefits.
While the outlook on AI is generally positive, significant work remains to enhance its integration and widespread use in clinical practice. Ensuring that radiologists are well-informed and trained in AI applications will be key to maximising the benefits of AI in radiology.
Source : https://www.sciencedirect.com/science/article/abs/pii/S0720048X24003061