The use of a computerized method of bone age assessment in clinical practice

We assessed the reliability and repeatability of a new computerized bone age system, both versions 3.4 and 3.5 (licensed by Discerning System Inc. and Ares Service SA, Serono), able to automatically assess bone age on a left hand and wrist radiograph. This computer system is based upon Tanner and Whitehouse's method (TW2), but there are important differences. Our sample included an initial group of 40 patients who had growth delay/constitutional delay of growth and puberty (n = 10), growth hormone insufficiency/deficiency (n = 15), low birth weight/Silver-Russell syndrome (n = 9), precocious puberty (n = 6), as well as 20 patients with various skeletal dysplasias (multiple epiphyseal dysplasia n = 7, pseudoachondroplasia n = 7, acrodysostosis n = 5, achondroplasia n = 1), 7 girls with Turner syndrome, and 10 boys with nephrotic syndrome on chronic corticosteroid treatment. Multiple anthropometric readings of the same radiographs demonstrated excellent repeatability of the assessment. In addition, the number of times that a manual insertion of a grade was required was similar in four different assessments. The computerized method did not entirely avoid errors in interpretation as the position of the x-ray on the screen was critical. There was a high manual insertion rate in radiographs of children with skeletal dysplasia. However, the computer assessment system, version 3.5, performed adequately for radiographs of children with normal bone morphology and Turner syndrome and had the advantage of a continuous scale of assessment.