Implementation and statistical evaluation of a web-based software for bone age assessment

Bone age assessment is a tedious procedure carried out for assessing growth disorders of children using the left hand radiograph. The purpose of this work was to implement and evaluate a web-based software based on the Tanner and Whitehouse method in a pediatric endocrine department of a social security hospital processing 600-1,000 radiographs per year. The system was evaluated by using a statistical technique for comparing measurement methods in order to test the performance of the procedure and a time study to assess its feasibility under local conditions. It was found that the intra-observer variation for the web-based Tanner and Whitehouse method was smaller (95% confidence limits, -0.77 to 0.97 vs. -0.45 to 0.37) then the conventional Greulich and Pyle manual method and the average net time required for an age assessment was 2.4 min. We therefore concluded that the web-based system should be adopted for its higher precision and relatively low turnaround time for cases requiring serial readings on the same patient. The statistical method demonstrated in this study can also serve as an example for evaluating similar biomedical parameter assessing software.