Original scientific paper
Semi-automatic Assessment of Cervical Vertebral Maturation Stages using Cephalograph Images and Centroid-based Clustering
Emir Sokic
orcid.org/0000-0003-3520-2094
; Faculty of Electrical Engineering Sarajevo, Bosnia and Herzegovina
Alisa Tiro
; School of Dentistry University of Sarajevo, Bosnia and Herzegovina
Elma Sokic-Begovic
; Federal Ministry of Health, Bosnia and Herzegovina
Enita Nakas
; School of Dentistry University of Sarajevo, Bosnia and Herzegovina
Abstract
Introduction: Effectiveness of different numerical techniques for use in semi-automatic assessment of cervical vertebral maturation stages (CVM) using radiograph images was investigated. Methods: Lateral cephalographs of 211 patients were recorded and stored in a digital form. Using the specially developed software application, orthodontic experts marked and measured several characteristic cephalometric parameters for every patient. The results of these measurements were used to automatically determine the cervical vertebral maturation stage using numerical techniques, including the K-means clustering and the Fuzzy C-means clustering. These results were compared with the assessment made manually by the trained orthodontists.Results: The best results were achieved using the modified Fuzzy-C means clustering. Identification of the correct CVM stage was around 70%, while the assessment including the adjacent classes [+/- 1 developmental stage] was over 99%. Conclusions: Experimental results show that it may be possible to develop a fully automated system to assess CVM stages, although there are still minor issues that need to be addressed before the method’s implementation in the clinical practice.
Keywords
Cephalometry; Cervical Vertebrae; Automatic Data Processing; Fuzzy Logic
Hrčak ID:
94355
URI
Publication date:
28.12.2012.
Visits: 2.103 *