clinical study

Assessment of a deep learning algorithm for the automatic detection of rib fractures on trauma CTs

Materials & Methods 

We retrospectively identified all trauma CTs referred from our emergency department between Jan. 2018 and Dec. 2018 (n=511). Examinations were categorized as positive (n=102) or negative (n=409) for rib fractures according to the clinically approved written CT reports. After anonymization, the bone kernel series (1.5 mm slice thickness) served as input for a rib fracture detection prototype algorithm based on a deep convolutional neural network (DCNN) that was previously trained on an independent sample (n=11,000).


75 fractures (50 acute; 25 chronic) detected by the algorithm were not mentioned in the written CT reports.


The AI solution was able to detect rib fractures that were previously not mentioned in CT reports. The algorithm has potential clinical application to be used in reading assistance.