Facing our ancestors. A new method of facial reconstruction using geometric morphometric technique

The human skull usually retains very little information about the specific individual's identity. Virtually restoring the soft facial tissue has been a challenge for many years. Today, soft tissue depth is being assessed from x-ray, ultrasonography, MRI or CT, followed by average soft tissue reconstruction. While this process was used in many forensic and historical cases, it still poses methodological problems, such as the dataset used for reconstruction and even subjective reconstructors' interpretation. Most importantly, as the human face is highly variable, relaying on average tissue depth at few selected points, holds inherent problems. Here we suggest a new, simple and more accurate method to evaluate the shape of a face from a skull. 30 CT scans of healthy individuals from routine medical imaging were acquired. All skulls and corresponding faces were 3D reconstructed, and 125 equivalent landmarks and semi-landmarks were placed on their surfaces. To explore to covariation between skulls and faces, partial least squares (PLS) method was applied. Next, principal component analysis (PCA) was used, in order to explore the space-shape relation between skulls and faces. PLS shows strong covariation between skulls and faces (R=0.85), supporting a connection between the shape of corresponding skull and face. From the PCA, a clear connection is visible between individual skull and its equivalent face, suggesting the possibility to predict the shape of the face from a skull. This method was applied to an early human and a Neanderthal skull in order to appreciate what they look like.