Dental Identification Abstract

Abstract— This thesis will analyze several similarity measures in a dental identification context. This identification will be based on a 3D scan of a patient’s skull and 2D radiographic images taken respectively post- and ante-mortem. This approach allows to circumvent the variable projection direction and/or incorrect patient positioning when capturing the classic 2D bitewing and periapical images. Three classes of experiments can be distinguished in this paper. First, there are the registrations with the classic ‘real-life’ 2D bitewing and periapical images. Secondly, experiments with dynamically reconstructed radiographs out of the 3D volumes were performed. These allowed to exclude the segmentation errors introduced in the first category
Dunn et al., [7], describe the use of invariant geometric structures to determine the projection geometry used to acquire an image. He demonstrated the relationship of these two-dimensional points independent of the projection geometry of which they were acquired.
D. Flint et al., [19], mention, for digital dental images, the challenges to the visual identification of the forensic odontologist. They include lack of dental morphology, changes to the radiographic appearance with replacement of restorations, differences in projection geometry between AM and PM images and the experience of the forensic odontologist. The significant difference in similarity measure between images of the same individual and images from dissimilar individuals is stressed as well. Factors affecting the registration of images and similarity measures were wide differences in the projection geometry used to capture the 2D images (focus of this thesis), artifacts and variations in location and extent of restorations on the teeth. Important to note is the use of semi-automated methods (the use of manually placed markers) in the research describe above.
C. Feature- vs. Intensity-based
[24] T.Chomdej, W.Pankaow and S.Choychumroon, “Intelligent dental identification system (IDIS) in forensic medicine,” Forensic Science International, vol. 158, pp. 27-38, 2006.
[25] A.Marana, E.Barboza, J.Papa, M.Hofer and D.Oiveira, “Dental biometrics for Human Identification,” Biometrics - Unique and Diverse Applications in Nature, Science, and Technolog, 2011-04-04.
[26] D.E.Nassar, A.Abaza, X.Li and H.Ammar, “Automatic Construction of Dental Charts for Postmortem Identification,” IEEE Transactions on Information Forensics and Security, vol. 3, no. 2, pp. 234-246, 2008.
[27] T.Lehmann, A.Sovakar, W.Schmitt and R.Repges, “A comparison of similariy measures for digital subtraction radiography,” Comput. Biol. Med. , vol. 27, pp. 151-167, 1997.
[28] D.Flint, S.Dove, P.C.Brumit, M.White and D.R.Senn, “Computer-aided Dental Identification: An Objective Method for Assessment of Radiographic Image Similarity,” J. Forensic Sci., vol. 54, no. 1, pp. 177-184, 2009.
[29] R.McLaughlin, J.Hipwell, D.Hawkes, A.Noble, J.Byrne and T.Cox, “A comparison of intensity-based registration and feature-based registration for neurointerventions,” LNCS, vol. 2489, pp. 517-524,