Towards building a photo-realistic virtual human face for craniomaxillofacial diagnosis and treatment planning

The virtual human face: Superimposing the simultaneously captured 3D photorealistic skin surface of the face on the untextured skin image of the CBCT scan

“…It is recognized that reconstructed 3D-CBCT soft tissue models might show unexplainable defects at the tip of the nose, and streak artefacts extending from the inner aspect of the nose 2,20 . Despite that, the nose region should not be removed as it could be affected by some orthognathic surgical procedures.…”

“…Previous studies have been limited to the analysis of the accuracy of superimposition of the whole face 2,8,20 , which is misleading since some regions of the face are associated with higher errors in superimposition than others. That is why in this study specific regions (patches) in each model were also examined individually.…”

“…The capture is slow; therefore image distortion caused by movement of the subject or change in facial animation is a potential source of errors. In addition, the developed skin surface image lacks the photorealistic appearance and the characteristic surface texture 1,2 .…”

Cephalometric Methods of Prediction in Orthognathic Surgery

“…Different methods are used to display the facial skeleton in combination with the dentition, to determine the exact localisation of the digital dental model in a cone-beam (CB) CT data set, and to fuse the facial soft tissue surface and the facial skeleton [86]. The most applied technique currently is fusing a 3D textured surface derived from a 3D photograph or 3D surface laser scan with reconstruction of multislice CT (MSCT), CBCT data or MRI slices [80][81][82]87].…”

Alternatives to Radiation Investigations in Orthodontics

“…3D studies conducted with Procrustes analysis have determined the translation of soft tissues in monozygotic and dizygotic twins 70 . Another study has used the Procrustes registration method in conjunction with 3D CT scan and Virtual Reality Modelling Language (VRML) to assess the registration error associated with such techniques and found it to be within +/-1.5mm in most parts 82 . A South Korean study 83 found negligible error of 0.37mm and less than 0.66% magnification with laser scanners.…”