摘要：This work addresses the problem of automatic determination of skeletal maturity in children and young adults. Skeletal age assessment is important for diagnosing and monitoring growth and endocrine disorders. We have constructed a system which uses Statistical Models of Shape and Appearance to locate bones in a radiograph and to predict skeletal maturity. By analysing the performance on a dataset of about 600 digitised radiographs of normal children we show that different variants of Part+Geometry (P+G) models are sufficient to initialise an automatic registration algorithm. We built global models of whole hand and local models of individual bones. We used the same P+G models to locate salient bones of the hand to initialise an Active Appearance Model (AAM) to match all the bones of the hand in a radiograph. We improved our age estimation results by using multiple local age group models and multiple local age estimators. We obtained an accuracy of 0.75mm and 0.70mm on sparse points placement for initialization of automatic registrations and Active Appearance models fitting respectively. We achieved a sub-millimeter accuracy for automatic model annotation and for locating the bones in a new radiograph. Our skeletal maturity methodology achieved an accuracy in estimating Skeletal Age of mean absolute error of 0.41 ± 0.02 years and 0.47 ± 0.03 years for female and male respectively
