Reconstruction of facial morphology from laser scanned data. Part I: reliability of the technique

The aim of this study was to evaluate the effects of scanning parameters on the precision of the data acquired using a facial laser scanner and to assess the reliability of automatic model recording in humans.

Data were acquired using a laser scanner (Cyberware 3030RGB); analysis and measurements were performed with Rapid Form 2004 software. A mannequin and six volunteers were scanned to investigate the effects of environmental conditions, positioning, head orientation, and software procedures. Precision and accuracy of the data were evaluated comparing six linear measures calculated on scanned data with those obtained directly. Two sessions with different head inclination were performed. The reliability of repeated scans was also assessed measuring the distance between the surfaces reconstructed from two separate scans of the same subject, at 12 anatomical points, in 5 subjects, during two sessions using a different head inclination. Differences were analysed using paired t-tests or analysis of variance (ANOVA).

The accuracy of scanning was +/-0.65 mm. The development of a specific protocol resulted in a mean scanning error of 1-1.2 mm and a recording error of 0.3-0.4 mm on repeated scans of human subjects.

This study indicates that scanning of the human face may be hampered by errors and artefacts, mainly due to movements. While the effect of trembling and involuntary movements during the exam may be minimized using faster scanning devices, comparative observation over time may be affected by unreal differences due to the uncertainty of facial expression. The overall error is, however, in the range useful for most clinical studies.