Feature Detection in 3D Images of Dental Models 

            Advances in 3D surface laser-imaging technologies have spawned new applications in dentistry.  Using a laser scanner, dental plaster models can be converted into 3D images that may be viewed in any preferred orientation and subjected to quantitative analysis.  Figure 1 shows a typical scanned image obtained with a high-resolution laser scanner, the Cyberware 3030-HIREZ from Cyberware Laboratories (Monterey, USA).

            The current practice of manually measuring a set of plaster models by the orthodontist to derive clinically important measurements may now be supplanted by computer-based measurements performed on their digitised representations stored in the computer.  Research currently undertaken in a collaborative project between the Department of Electrical & Computer Engineering and the Department of Preventive Dentistry aims to develop algorithms and techniques for the analysis of malocclusions and cleft palate deformities.

            The delineation of the dental arch is important in many orthodontic procedures since features measured from the detected arch are used by orthodontists in treatment planning and monitoring, e.g., in 3D space analysis for the determination of crowding and arch form problems.  We have developed a method for the automated detection of dental arches in 3D images of dental plaster models (Figure 2).  Our approach is based on a novel edge detection algorithm that results in selective and sensitive detection of roof edges, namely, ridge points.  We apply standard curve fitting to these detected roof edges.  Along the fitted third-order polynomial (blue curve), we find the peak position for each perpendicular inspection spoke.  Subsequently, we fit a fourth-order polynomial to these local peaks.  The pink curve in Figure 2 shows the detected dental arch.  It closely follows the important orthodontic features such as the cutting edges of the incisors, the apices of the canines, the bicuspids of the premolars and the cusps of the molars.

            Another aspect of the R&D work focuses on the analysis of cleft palate morphology.  The detection of the palatal arch forms (blue curves) and important landmarks (red dots) in a cleft deformity (Figure 3), allows the quantifiction of the cleft palate deformity through the measurement of linear, angular, areal and volumetric parameters.

            This work was done in collaboration with Dr Kelvin Foong (Dept of Preventive Dentistry) and Assoc Prof AA Kassim (Dept of ECE)

     Figure 1: 3D image of dental model.               Figure 2: Dental arch detection.

 Figure 3: Cleft palate morphology showing detected landmarks and dental arch.