AUTOMATIC CLASSIFICATION OF TEETH IN BITEWING DENTAL IMAGES 1 Mohammad Hossein Mahoor and Mohamed Abdel-Mottaleb Department of Electrical and Computer Engineering, University of Miami 1251 Memorial Dr., Coral gables, FL 33146 mmahoor@umsis.miami.edu, mottaleb@miami.edu 1 This research is supported in part by the U.S. National Science Foundation under Award number EIA-0131079. ABSTRACT We present an automated algorithm to classify teeth in bitewing dental images, using Bayesian classification, and assign an absolute number to each tooth based on common numbering system used in dentistry. Fourier descriptors of the contours of the molar and the premolar teeth in bitewing images are used in the Bayesian classification of these two types of the teeth. Then, the spatial relation between the two types of the teeth is considered to number each tooth and correct the misclassification of some teeth in order to obtain high precision results. Experiments with 50 bitewing images containing more than 400 teeth show that our method is capable of classifying and assigning absolute index number to the teeth with high accuracy. 1. INTRODUCTION Forensic radiography is part of forensic medicine, which is concerned with identifying people using the postmortem radiological images of different parts of the body including skeleton, skull and teeth. Radiological images of the decedent's body are compared with antemortem records of a missing person to evaluate similarities between them. Traditionally, dental-based identification relies on information such as missing teeth and dental works [1]. Nowadays with the advancements in dentistry and care of teeth by people, these methods might not be reliable; hence, developing new methods that use inherent dental features for identification is important [2]. In order to build an automated dental identification system (ADIS), we need to extract features from the teeth in the dental images of missing people and archive them in a database. During retrieval, the features for each tooth in the query image need to be extracted and compared to those stored in the database. If we limit the comparison of the teeth to the ones that have the same index number, this will help limit the search space and increase the robustness of the system. In this paper, we present an algorithm for the classification and numbering of teeth to be used during archiving and retrieval in/from the database. The algorithm starts by classifying each tooth in a bitewing image based on its inherent shape and then it considers the relationship between the neighboring teeth in the bitewing image to correct any initial misclassification. Finally, using the results of the classification, it assigns a number to each individual tooth based on the common numbering system of dentistry [1]. The Adult dentition contains 32 teeth, 16 teeth in each jaw. We divide the jaws into four equal quadrants which each quadrant contains eight teeth, two incisors, one cuspid (canine), two premolars (bicuspid), and three molars. As shown in figure 1, The numbering system numbers permanent teeth from 1 to 32, beginning at the maxillary right third molar (#1), extending across the maxilla to the left third molar (#16), then continuing to the left mandibular third molar (#17), and going around the mandibular arch to the right third molar (#32). Figure 1. The universal numbering system of adult teeth. In this paper we deal with bitewing images. These images usually contain two types of teeth, i.e., molar and premolar teeth. Section 2 presents our method for classification. Section 3 gives properties of Fourier 9 10 11 12 13 14 15 16 32 31 30 29 28 27 26 25 1 2 3 4 5 6 7 8 24 23 22 21 20 19 18 17 Right Maxilla Right Mandible Left Mandible Left Maxilla 0-7803-8554-3/04/$20.00 ©2004 IEEE. 3475