The BoneXpert bone age method has been validated in a study of healthy American children of four ethnicities in a paper published by the American journal Academic Radiology, thereby bringing the total number of peer-reviewed papers on BoneXpert up to ten. 

The study was based on a database of hand X-rays created by Prof. Huang’s bone age research group at University of Southern California in Los Angeles. The images were recorded from 1993 to 2006.

Each image had been rated twice manually and the root mean square deviation between the two manual ratings was 0.62 years.

BoneXpert was able to analyse all images with bone age above 2.5 years for boys and 2.0 years for girls, and the root mean square deviation between the automated rating and the average of the two manual ratings was 0.61 years.

Re-examining the 15 images with the largest deviation between automated and manual rating showed that 12 of these had been rated wrongly in the original manual rating, indicating that it is safer to use BoneXpert than manual rating, even if the manual rating is the average of two independent ratings. Link to the publisher