AM Modernization of Bone Age: Comparing the Accuracy and Reliability of Artificial Intelligence Algorithm and Short Hand Bone Age to Greulich and Pyle

Description

In this presentation, Mina Gerges thanks their team and Dr. Cooper from BC Children's Hospital before diving into an overview of bone age assessment methods, primarily focusing on the traditional Greulich and Pyle (GP) method. This conventional approach involves comparing a patient's X-ray to a series of age-standardized images, a process that is described as tedious for modern radiologists. The talk highlights the initiative to explore more efficient alternatives, such as the Short Hand Bone Age (SBA) method developed at Boston Children's Hospital, which simplifies the process by using only five key anatomical landmarks for quicker assessments, albeit with a restricted age range (adolescents only).



Furthermore, the presentation discusses the development of an automated algorithm by a group in Toronto, allowing for immediate bone age determination by simply uploading an X-ray image. Efficacy comparisons reveal that this automated system aligns closely with the gold standard GP method, boasting high correlation coefficients above 0.8, indicating strong agreement. Gerges cites findings involving a study with 213 patients, underscoring the repeated trials and methods used to affirm the accuracy and reliability of these new approaches.



Despite statistical significance, it is noted that the clinically relevant differences between methods are minimal, as they fall within acceptable limits for real-world application. The incapacity of the GP method due to reliance on physical atlases is emphasized, showcasing both the SBA and automated algorithm as superior future options, the latter removing age restrictions entirely, significantly enhancing diagnostic capabilities in pediatric medicine.

Specialties