Defining normal


April 17, 2014

A Perth-based research group is defining the boundaries of a normal face to better understand and treat disorders of the head and face.

University of Western Australia PhD candidate Stefanie Kung says is part of a research team at Princess Margaret Hospital for Children that is defining what a normal face is.

“You need to understand what is normal to understand the abnormal,” Ms Kung says.

The research team is applying its findings to treatment, disease monitoring and eventually diagnosis of patients with craniofacial disorders, such as Treacher Collins Syndrome where effected people typically have downward slanting eyes and are missing cheekbones.

When preparing for corrective facial surgery, surgeons would ordinarily use a generalised reference face to base their incisions on. This reference face is built on an average of healthy individuals. This average face, however, does not really represent anybody.

Using 3D facial imaging, the PMH group in collaboration with researchers from Melbourne and Belgium have developed a better method of determining what a normal face for a patient should look like.

Their method involved compiling a ‘face-space’ and constructing a normal version of the face for the patient – a virtual identical twin.

The face-space project was constructed using the facial images of about 1000 normal, healthy people between the ages of five to 25 from Perth. More than 200,000 markers were mapped onto their faces and used to define the entire spectrum of normal for a given population. Constructing a space, or range, rather than a static image of an average face allowed the research group to construct different, yet still harmonious, faces for patients.

The purpose of the face-space database was to capture the average facial form of a given population as well as the variation between facial forms, and use this range as the template for facial reconstruction.

Image of a 19-year-old girl with an asymmetric face, after a mapping mask has been applied.

The 3D facial images were captured using equipment which takes two photos of the patient’s face at the same time, one of each side . The images overlap and the computer stitches them together to created a 3D image. This piece of equipment costs more than $500,000 and looks deceptively simple. A long, narrow bar about 1.5 metres across holds out two cereal box sized cases which contain the cameras.

The patient sits in between the two ‘cereal boxes’ and the camera snaps away.

“The hard part is getting little kids to sit still,” Ms Kung says.

Researchers clean up the image by removing the hair, ears and any visible clothing. The image is then sent to the University of Leuven, Belgium where a team of experts, led by Peter Claes, uses super computers to create a heat map of the dysmorphic face.

The heat map shows the areas of the face which are furthest from the norm using a colour scale. Red represents the furthest from normal and blue is the closet to normal.

The photograph takes seconds to capture but processing the data can take days.

Along with the Perth face-space, the group applied dysmorphometrics – a term coined by Dr Claes for the measure of abnormal physical forms – to further tailor the virtual face to the patient. The system used the normal parts of the face to fill out the rest of the dysmorphic face.

“When we say dysmorphologies, we mean the regions of the face that are characteristic of that condition,” Ms Kung explains.

This means the team is not creating just another averaged face to compare patients to, but a tailored range inside which the patient can fit.

On top of developing software to model these virtual twins for patients, the team is trying to train the computer to automatically pick out abnormalities for diagnosis.

They are trying to develop a toolkit for an academic, researcher or clinician to investigate conditions in which they are interested.

“It takes an expert to be able to pick out those features just by looking at someone’s face and getting a hunch that there is some underlying condition,” Ms Kung says.

“Not everybody can do that.

“What we hope the end product will be is to be able to give a non-expert the tools where they can acquire expert knowledge.”

Categories: Health, Technology

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.