Computer-aided automated landmarking of cephalograms.

The computer-aided method presented is able to interpret and measure radiological structures automatically and reproducibly on a large number of randomly selected lateral head films of different film quality. For example the hard and soft tissue expression of the profile is used to discuss method, application and disadvantages. The performance of the system depends on image quality. The algorithm cannot be influenced by opacifications due to metal structure. In 90% of all cases the contours are identified correctly. Even if the image quality is poor, the recognition rate is about 84%. The selected cephalometric landmarks are placed in the right way in 85% of all cases. The constraint mathematical conditions and the high reproducibility improve the quality process of the cephalometric analysis. Recognition rates of 84% to 90% justify even nowadays the routine use of semi-automatic systems in PC-based analysis. With advancing digital radiography and improved computer performance, image interpreting systems will certainly become established.