Development of a 3D morphable face model for planning, guidance and assessment of craniofacial surgery

Overview

This new collaboration between the Department of Computer Science, HYMS (York) and the Craniofacial Unit at Great Ormond Street Hospital (London) aimed to provide surgical planning aids and surgical outcome assessment for patients suffering from chronic facial abnormalities, in particular congenital conditions, such as Apert syndrome. The project was the first step in building an accurate, state-of-the-art deformable 3D face model that is able to show how the face should look were abnormalities removed.

A software tool has now been built that allows the fast manual mark-up of 3D face scans - features of this software development include multiple simultaneous views of the surface to allow accurate placement of manual landmarks and auto-piloting between different regions of the face that require landmarking. A training set of 3D faces has been marked up using the "me-in-3D" dataset at GOSH and the auto-landmarking software developed in a recent PhD at York. The results were successful on adults, but less successful on children – most likely because the training set consisted exclusively of adults.

The findings were encouraging and, as a follow on to this, a team of collaborators have been gathered who are working on an EPSRC proposal to continue the work, including developing the spare model fitting process into a dense model fitting process. As well as with Great Ormond Street Hospital, there have been several new collaborations generated by the project, including with 3DMD manufacturers of medical industry standard 3D scanning equipment, Cincinnati Childrens Hospital Medical Center, and University College London's Department of Medical Physics.

Ouputs

Grants

• Nick Pears, Leverhulme Trust, Royal Academy of Engineering Senior Research Fellowship, £48,713