Published in:
01-03-2013 | Original Article
An optimized video system for augmented reality in endodontics: a feasibility study
Authors:
D. D. Bruellmann, H. Tjaden, U. Schwanecke, P. Barth
Published in:
Clinical Oral Investigations
|
Issue 2/2013
Login to get access
Abstract
Objectives
We propose an augmented reality system for the reliable detection of root canals in video sequences based on a k-nearest neighbor color classification and introduce a simple geometric criterion for teeth.
Material and methods
The new software was implemented using C++, Qt, and the image processing library OpenCV. Teeth are detected in video images to restrict the segmentation of the root canal orifices by using a k-nearest neighbor algorithm. The location of the root canal orifices were determined using Euclidean distance-based image segmentation. A set of 126 human teeth with known and verified locations of the root canal orifices was used for evaluation.
Results
The software detects root canals orifices for automatic classification of the teeth in video images and stores location and size of the found structures. Overall 287 of 305 root canals were correctly detected. The overall sensitivity was about 94 %. Classification accuracy for molars ranged from 65.0 to 81.2 % and from 85.7 to 96.7 % for premolars.
Conclusion
The realized software shows that observations made in anatomical studies can be exploited to automate real-time detection of root canal orifices and tooth classification with a software system.
Clinical relevance
Automatic storage of location, size, and orientation of the found structures with this software can be used for future anatomical studies. Thus, statistical tables with canal locations will be derived, which can improve anatomical knowledge of the teeth to alleviate root canal detection in the future. For this purpose the software is freely available at:
http://www.dental-imaging.zahnmedizin.uni-mainz.de/.