Figure 1: Various dental images: (a) dental casts. (b) lateral cephalogram. (c) facial model.

Different medical imaging modalities produce complementary diagnostic information. Image data that is of interest to the dentofacial clinicians, e.g., orthodontists and oral surgeon, include lateral cephalograms (which are X-rays of the head), dental casts (plaster models), and facial shape models. Examples of these image modalities are shown in Figure 1. The availability of a 3D environment of the craniofacial region that integrates this data will provide information that is useful to the clinician for diagnosis and treatment planning.

The 3D images of the dental casts and face contain only surface data and there are no common landmarks that can be employed to register them. H owever, common features exist between the lateral cephalogram and facial model, and also between the lateral cephalogram and the dental casts. We can thus use the lateral cephalogram to link the two 3D images so that they may be positioned accurately together in virtual 3D space. This project involves the registration of the facial model and dental casts, respectively, with the lateral cephalogram.

To register the lateral cephalogram and the 3D facial model, we first extract the facial profiles from the side view. The lateral cephalogram is preprocessed to enhance the soft tissue outline us ing image processing techniques. An edge detector extracts the soft tissue facial profile by detecting the ridge of the nose from horizontal cross-sections of the face. From the extracted profiles, the appropriate transformation parameters for translation, scaling and rotating the images can be determined.

The registration procedure for the images of the dental casts and cephalogram require, accurate positioning of the maxillary (upper jaw) and mandibular (lower jaw) casts so that they are in their natural biting posture. Two prominent hard-tissue features, the superior incisor and the last molar, are selected as the common landmarks on both images. The geometric transformation for aligning the dental casts with the cephalogram is then calculated and applied.

With the lateral cephalogram as the reference object, the spatial relationship between the facial model and dental casts can be determined. The lateral cephalogram is superimposed onto the 3D facial model followed by registering the 3D dental casts with the above cephalogram. Finally, the data of the dental casts and facial model are integrated and visualised (Figure 2).

Figure 2: Registration result: dental casts and facial model.

The registration system permits the creation of a virtual 3D dental and facial environment for orthodontic diagnosis and treatment planning.Routine clinical records of the orthodontic patients can be converted into a digital platform for ease of storage as well as analysis. Furthermore, through the accurate merging of the three different image modalities in this system, subsequent enhancements will take the form of accurate merging of root shapes to the tooth crowns on the dental casts. Simulation of tooth movements within the created 3D virtual environment will permit the clinician to visualise how changes in the spatial position of the teeth affect the spatial position of the lips and facial contours.