Later stage bone age assessment on hand radiographs

Skeletal maturation estimation or bone age assessment of children is a common procedure performed in pediatrics. The bone age assessed by physicians could vary from time to time and the accuracy depends on physicians´ experiences. In order to provide a stable and more accurate bone age assessment, computer-assisted bone age assessment systems are employed. Once the epiphyseal fusion has started, the degree of fusion and texture features are analyzed. This paper presents an algorithm to extract and analyze the texture features of the regions of interest of phalanges in hand radiographs at the later stage in order to provide a stable and efficient later stage bone age assessment. The presented algorithm first uses wavelet transform to obtain three detail sub-bands from each region of interest. Then singular value decomposition is employed on each sub-band to extract the maximum likelihood feature, λ. Finally, support vector machine is applied for classification. Experimental results show that this algorithm can provide efficient and accurate later stage bone age assessment.