Perceptions of Greek radiographers and radiologists on the impact of artificial intelligence in medical imaging: A cross-sectional study.
Authors
Affiliations (4)
Affiliations (4)
- School of Social Sciences, Hellenic Open University, Greece.
- School of Medicine, National and Kapodistrian University of Athens, Greece; General Hospital of Attica KAT, Athens, Greece. Electronic address: [email protected].
- School of Social Sciences, Hellenic Open University, Greece; School of Medicine, National and Kapodistrian University of Athens, Greece.
- School of Social Sciences, Hellenic Open University, Greece; Department of Nursing, University of Thessaly, Greece.
Abstract
This study aimed to examine the perceptions of Greek radiographers and radiologists on integrating artificial intelligence (AI) in medical imaging (MI). A convenience sampling approach was used. The questionnaire was distributed across fifteen public hospitals and two private healthcare groups. A total of 167 valid responses were collected. The data were analysed, conducting both descriptive and inferential statistics. A positive attitude towards the application of AI in MI was identified. Males expressed greater agreement with its positive impact compared to females (p = 0.033), while females reported greater concerns about personal data protection (p = 0.028). Increased age was associated with greater acceptance of AI for diagnostic decision-making (p = 0.014), whereas higher education was associated with lower agreement to statements on the negative impact of AI (p = 0.043). As acceptance of shared responsibility between humans and AI algorithms increased, the agreement with AI's contribution to education decreased (p = 0.006). The acceptance of autonomous AI diagnostic decision-making reinforced the recognition of its impact on the professional role of radiographers and radiologists (p = 0.040). Respondents who were more aware of potential AI-related errors were more likely to believe that diagnosis must remain a human task (p < 0.001). Respondents who recognised the possibility of AI errors were less likely to support autonomous AI decision-making (p = 0.042). The main barriers to implementing AI on MI were a lack of knowledge (64.1%), implementation costs (40.7%), and cybersecurity threats (46.1%). Respondents demonstrated an overall positive attitude towards AI integration, recognising its potential. Compared to previous studies, reservations about fully autonomous diagnostic decision-making and implementation barriers were identified. Despite the willingness of respondents to adopt AI, the findings suggest that the successful integration of AI into radiological clinical practice in Greece will depend on addressing knowledge gaps through structured education and training, which emerged as the most significant barrier to implementation.