[Automated system of bone maturity evaluation applied to the Tanner-Whitehouse 2 method]

Employing automated methods of image analysis to determine bone maturity helps to avoid subjectivity, reduce the execution time and to increase continuity in maturity scales. We have studied bone maturity by the Tanner-Whitehouse method (TW2) and by image analysis based on Fourier transformation.

Bone maturity (BM) was evaluated in 200 X-rays of the left hand and wrist from 45 children between 2 and 18 years of age employing the TW2 method, manual or traditional, to determine the state of maturity of each bone. At the same time, BM was evaluated by using an automated form of a recognizance program (IMAGO 2). As a standard we used the schemes of the phases described in the TW2 method. The image of the bone in question captured by a video camera after digitalization was classified by statistical comparison of its Fourier coefficient matrix with that previously determined by each of the standard schemes. Manual and computerized bone age concordance was also studied.

The computerized procedure showed a tendency to give lower values than the manual procedure. There were no significant differences between mean phases when calculated by both methods (manual and computerized) for the radius and the fifth metacarpal. The least degree of concordance was observed in the third metacarpal and phalanges. We found better concordance in the carpal bones than in the phalanges. This concordance was lower than that found in the radius and fifth metacarpal. The worst results were observed in the carpus and large bones and the best results were observed in the semilunar and scaphoid bones.

Divergence in the reproducibility between this computerized system and others (CASES method) are probably related to the differences between the digitalization process and the standard images used.