3D atlas based models for orthognathic surgery planning

The primary objective of orthognathic surgery is to improve facial aesthetics, dental aesthetics and function to the patient’s satisfaction (Kiyak et al., 1981). Patients are aware if they look different from others, it affects their behaviour and self-esteem negatively (Stirling et al., 2007).

Patients need to be aware of the improvements expected in their extra oral facial appearance as a result of the surgical intervention. Predicting the changes in facial soft tissue (prediction planning) is dependent upon the complex interaction of the teeth, the attached bone and the overlying facial soft tissue.

At present facial prediction planning techniques are two-dimensional, based on a profile image, or more recently three-dimensional. The majority of surgical centres routinely use of 2D planning due to its ease of use; accepting the inherent deficiency of being able to predict only the facial profile. Patients are unaccustomed to viewing themselves in profile and report difficulties in visualising profile changes in their facial appearance following surgery (Turpin, 1995).

Three-dimensional imaging has increased the diagnostic ability in the oral and maxillofacial area. The use of cone beam CT (CBCT) to capture 3D hard tissue and surface scanners i.e. laser scanners or stereophotogrammetry, to capture 3D soft tissue has revolutionised orthognathic surgery prediction. However the equipment is expensive, limited to only a few expert centres, complex to use and the CBCT exposes the patient to substantially more radiation than a profile radiograph. As a result the uptake of 3D facial planning in the NHS has been disappointingly slow. Clinicians are trained to plan using 2D images whilst consumers (patients) are more accustomed to viewing 3D content.
Aims: We are seeking a student with a strong background in physical and computational techniques for this project. This interdisciplinary proposal aims to combine 3D models of faces with the two-dimensional images to produce a novel quantitative method of prediction planning, bridging the gap between Clinicians and patients. The use of 3D models and atlases are common in many clinical areas, such as fMIR. A universal atlas is commonly used to map data from multiple subjects onto a common platform, hence providing a unified means of analysis. In this work, we will develop and evaluate the use of atlas based 3D model registration and profile generation, from 2D and 3D images, to provide an anatomically correct segmented 3D dataset which will then be used to generate patient specific 3D models. This work will be ground breaking in both the fields of Vision and orthognathic and will bridge the gap between physical, computational and clinical sciences.

The project will require the development of computer vision based solutions to
1. Provide, assess and quantify the accuracy of seamless registration of 2D face profiles with 3D models of the human head.
2. Study the interaction of conventional 2D radiographic hard tissue surgical changes on 3D facial soft tissue outcomes.
3. Investigate the use of topographical features extracted from conventional radiographs to deform generic 3D models.
4. Examine the ability of deformed generic hard tissue models and 3D facial soft tissue to predict 3D facial soft tissue outcomes.

This study challenges the concept of profile planning orthognathic surgery and assess whether it is possible to improve 3D facial soft tissue predictions. The results of the study will have important clinical implications, especially in showing the patient their predicted post-surgical result; this has medico-legal implications and in managing patient expectations.