The study, published in the September 2024 issue of Elsevier, had radiologists and an AI solution analyze 94 adult CT scans of pelvis and/or hips and extremities with suspected bone fractures between January 2022 and August 2023. The results of each were compared to the findings of a senior radiologist.
Of the patients, 47 had at least one fracture and a total of 71 fractures were deemed present using the standard reference. The AI solution had 58 true positives, 13 false negatives, 33 true negatives and 15 false positives, resulting in a 76% accuracy rate.
Radiologists scored 90% accuracy with 65 true positive, 6 false negative, 42 true negative and 6 false positive findings. The radiologists also outperform AI in sensitivity and specificity.
“This raises the question of whether a strong standard of reference for evaluating AI solutions should be used in future studies comparing AI and human reading in fracture detection using radiographs,” the study authors wrote.
