Predicting face shape from the skull using a combined MR and stereophotographic image database of living individuals

Facial reconstruction is an artistic process with a limited scientific basis. It is a last resort in human identification inwhich an approximate facial appearance is produced fromthe skull in the hope that a resultant candidate identity might be confirmed by other means such as DNA or dental records. The aim of this investigation was to establish whether contemporary medical and stereophotographic imaging methods could be combined to predict face shape from the skull and to assess whether they can be used to reassess current guidelines for the positioning of facial features and other norms in forensic facial reconstruction. A database was collected from 60 individuals (30 male and 30 female) of predominantly White ancestry, consisting of an MRI scan of the volunteer's head and face, and corresponding 3-D stereophotographic and laser-scanned facial surface images. The skull and face surfaceswere landmarked in 3-D at up to 30 traditional anthropometric points. Accuracy in facial landmark positioning in laser, stereophotograph, and MR imageswas compared. A comparative analysis of hard and soft tissue landmarks was used to develop a method of predicting soft tissue landmark position and tissue depth from the skull of an unknown individual. Finally, the combined datasetswere used to produce a revised set of guidelines for positioning and proportions of soft-tissue features.