Discover more from Wajdi Archives
AI for Chest Radiograph
Generative AI Matches Radiologists' Performance in Chest Radiograph Interpretation.
In a study published in JAMA Network Open on October 5, 2023,1 researchers investigated the use of generative artificial intelligence (AI) for interpreting chest radiographs in the emergency department (ED).
Their objective was to evaluate the accuracy and quality of AI-generated chest radiograph reports in comparison to reports produced by on-site radiologists and teleradiology services.
Thanks for reading Wajdi Archives! Subscribe for free to receive new posts and support my work.
The researchers conducted a retrospective diagnostic study involving 500 chest radiographs randomly selected from an ED in a tertiary care institution.
Each radiograph was independently assessed by a teleradiology service, an attending on-site radiologist, and an AI model.
Emergency department physicians rated the quality and accuracy of these reports using a Likert scale.
The primary outcome of the study was to determine any differences in Likert scores among the three types of reports.
The results indicated that the AI model produced reports with comparable clinical accuracy and textual quality to those of radiologists, outperforming teleradiology reports.
Secondary analyses revealed no significant differences in the probability of reports containing clinically significant discrepancies for all three report types.
The study showed that AI-generated reports can provide valuable clinical support in the ED, with performance on par with on-site radiologists and superior to teleradiology services.
The research suggests a promising role for generative AI in enhancing clinical decision-making, particularly in situations requiring timely identification of critical conditions.
Further research is necessary to assess the broader applicability of AI-generated reports in various clinical settings.
Huang J, Neill L, Wittbrodt M, et al. Generative Artificial Intelligence for Chest Radiograph Interpretation in the Emergency Department. JAMA Netw Open. 2023;6(10):e2336100. doi:10.1001/jamanetworkopen.2023.36100