CT Slice: The Art of Facial Reconstruction
By Josh Hildebrand
Vol. 24 No. 1 P. 8
With rapid technological developments, the radiology field is constantly innovating and integrating more effective types of imaging. Recently, a form of 3D imaging prominently used by forensic artists and archeologists is piquing the interest of radiologists.
Whithorn, Scotland, is a famous historical site said to be the “cradle of Scottish Christianity.” In September 2022, with funding from the Whithorn Trust and in collaboration with the Cold Case Whithorn Project, the skulls of three key figures from the 12th to 14th century were recovered during an archeological dig. The people are believed to be a female resident of the local area, a priest, and a bishop. Following their retrieval, Christopher Rynn, PhD, a self employed forensic artist, was recruited to reconstruct the faces of these figures using solely their skulls.
“It was all physical sculpture back in the early 2000s,” Rynn says. “We used to have to carry human skulls across borders to plaster cast them. Thanks to modern technology, the process is much easier; now, it’s international.”
Within the last decade, facial reconstruction has received an upgrade. Thanks to newer, more advanced technology, physical skulls are no longer needed to reconstruct faces. With this new technology, a CT scan provides more than enough data for a forensic artist to reconstruct a face. Organizations from around the globe are now able to send CT data to artists like Rynn, who can then translate and transform those data into a 3D model.
3D Systems, a US-based additive manufacturing solutions provider, manufactures the device Rynn uses for facial reconstructions. The device is called Touch X, a haptic, brushlike, 3D sculpting tool that provides feedback when users interact with 3D models. Oqton’s Geomagic Freeform is Touch X’s native software—a pseudo-art room suite that grants artists the means to create models in 3D.
“This device has been under the radar since I started using it,” Rynn says. “I don’t know how people can sculpt in 3D using a 2D interface such as a mouse. I can feel the skull I’m working on because of the device’s haptic feedback, which avoids the limitations of 2D interfaces.”
The reconstruction process, according to Rynn, is both scientific and artistic. Every skull has the same muscles, but not everyone looks the same. Variations in face shapes and orifices come about from the differing shapes of human skulls.
“It’s a manual step-by-step process,” Rynn says. “I start with tissue depth pegs, which are averages. I then join them together based on the facial muscles, which are sculpted to fit the skull. There is a certain amount of bone, muscle, and soft tissue, and everybody has the same amount. There’s a sort of [natural] balance in the face.”
Two facial features that are particularly difficult to reconstruct are mouths and ears. Accurate reconstruction of a mouth requires some knowledge of dentistry, according to Rynn. Whether someone has an overbite or underbite can greatly affect the shape of their mouth.
Ears are another challenging area because they are strictly cartilage. CT scans of a skull do not reveal much about ear shape, especially when considering differences between right and left ears, which are not identical. However, it is possible to tell the length of a person’s ears based on their mastoid process, a projection of bone located toward the base of the skull.
“It’s a shame we can’t tell more about ears from the skull,” Rynn says. “It’s obvious that you wouldn’t be able to because they aren’t functionally attached to the skull. Something like a nose is functionally attached to the skull, which makes it easier to recreate.”
Noses and eyes do not pose nearly as much trouble. Eye and brow shapes are comparatively easy to reconstruct based on a skull’s supraorbital ridge, the part of the skull directly above the eye sockets. The supraorbital ridge provides the shape, angle, and protrusion of the eyes, as well as the shape of the eyebrows.
Reconstructing a nose is a matter of measuring the nasal aperture, the pear-shaped inlet on a skull, and running the measurements through regression equations. The shape of the nasal aperture also provides artists with the shape of the nose, whether it is upturned or downturned, and the shape of it in profile.
Photo textures are added once the skull has been sculpted in 3D. These textures include photographs of similar looking faces, facial features that align with the artist’s interpretation of the skull, and the artist’s best estimation of what an individual looks like. When the artist has achieved the desired level of realism, an AI program can then be used to animate it, allowing the reconstructed face to blink and move slightly.
“The only part of the reconstruction process that is automated is the animation of the face by an AI,” Rynn says. “I keep that part of the process on a tight leash because the animation is someone else’s movement mapped to the face I’ve reconstructed. If an animation distorts the face even slightly, I don’t use it.”
Although this technology and its software are most notably used for forensic and archeological purposes, there are clinical uses as well. A less known use of this technology is medical legal visualization, which involves private patient data in a legal setting. The other main use for this kind of imaging is the creation of 3D-printed prosthetic implants.
“Because of the Data Protection Act, none of the medical professionals who use the software [clinically] can talk about it,” Rynn says. “Though, we do know it is also used for 3D printing and prosthetics. Say a patient has a hole in the side of their skull. By using only a CT scan, an artist or doctor can mirror that patient’s skull and work from there to create a prosthetic.”
The ability to create interactive 3D models suggests that the technology could be useful for facial reconstruction surgeries, as well. While it may be some time before it is widely integrated into medical departments, its use for forensic investigations and archeological reconstructions highlights its potential.
— Josh Hildebrand is an editorial assistant for Radiology Today.