Tuesday, November 28 | 8:30 a.m.-8:40 a.m. | T1-SSPD03-3 | Room E353B
In this session on pediatric imaging, a deep-learning model will be presented that is designed to assess bone age by visualizing features of the olecranon on lateral elbow x-rays.
“Accurate bone age evaluation is crucial for assessing skeletal maturation, especially during puberty. Hand bone age has limitations during this period, and elbow bone age evaluation can be a valuable alternative,” noted Gayoung Choi, MD, PHD, of Korea University in Seoul, who will present the research.
Significantly, the group first reviewed and classified 3,508 lateral elbow radiographs based on morphological changes of the olecranon bone ossification process and compared this novel approach with previously established elbow bone and hand and wrist bone age methods. Next, they used this dataset to train, validate, and test the deep learning-based model.
The deep-learning model for estimating olecranon bone age showed an accuracy of 96% and a specificity of 98% on the internal test set. In an external validation, the accuracy was 86%, according to the results.
“The novel olecranon method can be a simple and practical choice for bone age evaluation in puberty,” Choi and colleagues suggest.
Tune in on Tuesday morning in this session for all the details.
