Back to the list
Congress: ECR25
Poster Number: C-12171
Type: Poster: EPOS Radiologist (scientific)
DOI: 10.26044/ecr2025/C-12171
Authorblock: C. Thouly1, B. Dufour1, N. Heracleous1, D. Moreira1, D. Ribeiro1, B. Rizk1, F. Zanca2; 1Sion/CH, 2Leuven/BE
Disclosures:
Cyril Thouly: Nothing to disclose
Benoît Dufour: Nothing to disclose
Natalie Heracleous: Nothing to disclose
Daniela Moreira: Nothing to disclose
Diana Ribeiro: Nothing to disclose
Benoît Rizk: Nothing to disclose
Federica Zanca: Nothing to disclose
Keywords: Artificial Intelligence, Bones, Paediatric, Conventional radiography, Digital radiography, RIS, Computer Applications-General, Cost-effectiveness, Structured reporting, Education and training, Outcomes
Methods and materials

This study was conducted across eleven diagnostic centers in Switzerland, all utilizing the BoneXpert AI software for bone age estimation. The study was divided into two phases to compare the efficiency of manual versus AI-assisted reporting. In Phase 1 (P1), radiologists manually estimated bone age and generated reports using the Greulich and Pyle atlas. In Phase 2 (P2), BoneXpert provided automatic bone age and height predictions, which were seamlessly integrated into structured reports. Radiologists had the option to review and supplement these reports with additional clinical information using voice recognition technology. To measure the impact of this transition, reporting time was recorded during both phases, with radiologists triggering a chronometer in the Radiology Information System (RIS) at the start and end of reporting. Prior to full implementation, a rigorous quality check ensured that AI-generated reports met clinical standards.
GALLERY