Comparison of Greulich-Pyle and Gilsanz-Ratib Atlases Through An Intelligent Bone Age Assessment System

In this study, Greulich-Pyle Atlas which is the most commonly using method of bone age assessment for children and Gilsanz-Ratib Atlas that also has a recently rising popularity are compared. The comparison is examined by an intelligent system utilizing a novel improved method and artificial neural networks. The novel method is basic as utilizing distal radius, ulna and their epiphysis only, and it is accurate as defining qualified parameters for assessing bone age. Additionally, artificial neural networks provide the method to be applied as fast and accurately. As a result of comparison, it is demonstrated that Greulich-Pyle Atlas and Gilsanz-Ratib Atlas are highly compatible expect only for several ages. The mean square error between the atlases is obtained 0.16 year for males and 0.42 year for females, respectively. Thus, reliability of both Greulich-Pyle Atlas based system and Gilsanz-Ratib Atlas is demonstrated.