Development of a New Software for Automatic Anatomical landmark Detection in Cephalometric X-ray Images

Background & Objectives: As a routine clinical procedure in orthodontics, there are many geometrical calcula-tions usually taken place on the lateral x-ray images of the head in order to develop an appropriate treatment plan. This is currently done manually by the expert radiologists and/or orthodontists. It can take an experi-enced orthodontist up to thirty minutes to analyze one x-ray image and the results are still influenced by both subjective and objective errors. Here a new software for automatic anatomical landmark detection in cepha-lometric x-ray images is presented. The software extracts ten commonly used main landmarks. Methods and Materials: In this research, a combination of various image processing techniques including Template Matching, Edge Walking, and Shape Registration is employed to develop an automatic cephalomet-ric landmark detection algorithm within the x-ray images of the head. Ten essential and commonly used landmarks were selected as the target points. Results: The accuracy of the automatic landmark detection algorithm was evaluated with comparing the re-sults obtained by the automatic algorithm to those manually identified by an expert orthodontist on a data base of 10 x-ray images. The statistical analysis showed that 92% of the landmarks were determined accurately within an acceptable region of 4 millimeters from the points an expert had defined.. Conclusions: A MATLAB-based software was developed accordingly to help an orthodontist automatically identify the most commonly used 10 main cephalometric landmarks. The method can be extended for more landmarks in the future.