Plaque measurement from intra-oral frames

The development and optimisation of tooth cleaning products requires the exact measurment of dental plaque before and after cleaning. The conventional procedure is to test a representative number of people by colouring the plaque and roughly estimating its quantity by taking a look at the teeth. The result of this procedure is subjectiveto a high degree and varies due the inter and itra observer variety. This paper presents a system for the computer assisted evaluation of dental images, aiming at the reduction of costs for the evaluation as well as the objective measurement of plaque. The presented system is robust to process images, taken under everyday conditions. It implements an image processing pipeline, which consists of three processes: 1. specular-reflection supression 2. tooth contour detection to segment the tooth of interest by B-Snakes 3. plaque measurements by classification applying discriminate analysis. The procedures`s result is a qotient of the area of the measured pl aque and the tooth. This paper discusses the applied techniques and examines the reproducibility of their results. In order to enhance the tooth contours, to suppress specular-reflections, and to detect plaque pixel, three colour image transformations (HSV, CIE-Luv, Discrete Karhunen Loeve Transformation) are applied. In order to calculate the B-Snake's external energy, a new morphologic method for edge enhancement is introduced, which guarantees a robust snake behaviour. Finally, the system's interaction techniques are introduced. They are designed to enable the user to fully control the quality of each individual of the above processes.