A content-based system for human identification based on bitewing dental X-ray images

doi:10.1016/j.patcog.2005.01.011

Pattern Recognition

Volume 38, Issue 11, November 2005, Pages 2132-2142

https://doi.org/10.1016/j.patcog.2005.01.011 Get rights and content

Abstract

This paper presents a system for assisting in human identification using dental radiographs. The goal of the system is to archive antemortem (AM) dental images and enable content-based retrieval of AM images that have similar teeth shapes to a given postmortem (PM) dental image. During archiving, the system classifies the dental images to bitewing, periapical, and panoramic views. It then segments the teeth and the bones in the bitewing images, separates each tooth into the crown and the root, and stores the contours of the teeth in the database. During retrieval, the proposed system retrieves from the AM database the images with the most similar teeth to the PM image based on Hausdorff distance measure between the teeth contours. Experiments on a small database show that our method is effective for dental image classification and teeth segmentation, provides good results for separating each tooth into crown and root, and provides a good tool for human identification.

Introduction

Reliable identification of humans is important for many applications, such as law enforcement, border control, homeland security and airport security. Dental features are regarded as the best candidates for postmortem (PM) biometric identification. Not only they represent a suitable repository for unique and identifying features, but they also survive most PM events that can disrupt or change other body tissues, e.g. bodies of victims of violent crimes, motor vehicle accidents, and work place accidents, whose bodies can be disfigured to such an extent that identification by a family member is neither reliable nor desirable [1], [2], [3]. For these reasons, human identification based on dental features has always played a very important role in forensics.

There are two scenarios for the use of dental identification. In the first scenario, a comparative identification is used to establish the degree of certainty that the dental records obtained from the remains of a decedent and the antemortem (AM) dental records of a missing person are for the same individual. In the second scenario, the AM records are not available, and no clues to the possible identity exist. In this case, a PM dental profile is completed by the forensic odontologist suggesting characteristics of the individual in order to narrow down the search. Traditionally, this kind of identification work is carried manually by forensic odontologists [4].

Basically, comparative dental identification methods, in which PM dental records are analyzed and compared against AM records to confirm identity, relying on dental restoration and dental work features rather than inherent tooth characteristics, e.g., morphology of teeth and roots. Clearly, individuals with numerous and complex dental treatments are often easier to identify than those individuals with little or no restorative treatment. But, in many cases these features are not enough to get correct identifications, moreover, these identification methods are not fully automated as image comparison is carried out manually [5], [6], [7]. In the future, these features may become unreliable and difficult to use due to the advances in dentistry. For example, contemporary generations have less dental decay than their predecessors; also cavities in today's children and their offsprings will be virtually undetectable because of using hi-tech pit and fissure sealants. Consequently, it becomes important to develop automatic dental identification systems using inherent dental features [8], such as shapes of roots and crowns, and space between teeth, for substituting the manual methods [9], [10], [11]. Recent special issues of the IEEE Computer and ACM Communications survey the state of the art in biometric identification technologies and discuss future trends [12], [13].

To build automatic dental identification systems, there are several challenges. For example, dental features may change over time. Therefore, they are not always reliable and the system needs to decide when to rely on them and when to ignore them. Meanwhile, the system needs to handle dental radiographs of poor quality and take view variance of the images into consideration. An important issue is the segmentation of the teeth from the radiograph. This step is crucial for the success of the automated dental identification system, because the accuracy of the extracted features depends on the results of the segmentation.

Current research on automated dental identification systems focuses on shape features [14], [15], [27], [28]. During archiving, they segment the AM images, extract and store either teeth contours or contour descriptors in a database. During retrieval, a PM image is submitted, and these systems segment the image and obtain contours of the teeth or contour descriptors, then match them with the ones in the AM database. The best matches are then presented to the user.

In this paper we present a system for archiving and retrieval of dental images to be used in identification based on dental images. The system includes steps for dental image classification, automatic segmentation of bitewing dental X-ray images, and teeth shape matching. The main reason for working with bitewing images is that the shapes of molar teeth in bitewing images are considered more distinctive than other teeth. Also, the bones are usually visible and distinguishable from the teeth and can be used to separate the roots from the crowns of the teeth. As we will see in the experiments, since the quality of bitewing images is usually not poor, most of the teeth in these images could be successfully separated into crowns and roots, which is potentially important for extracting features for identification.

The paper is organized as follows: Section 2 introduces our method for classifying the three types of dental images, details the algorithm for segmenting bitewing images, and describes the teeth matching algorithm. Section 3 discusses the experimental results of the classification, segmentation, and retrieval using the proposed methods in Section 2. The conclusions and the future work are discussed in Section 4.

Section snippets

System components

Fig. 1 shows a high-level diagram of our system. The system contains two stages: archiving and retrieval. During archiving, the system processes AM images, classifies and segments them, extracts the teeth contours and archives them in a database. In the retrieval stage, a PM image is submitted; the system classifies and segments the image and uses extracted contours of the teeth to calculate shape distance between the PM and the AM images, and presents AM images with the smallest distances to

Experimental results

In the dental image classification experiment, we used 60 images as the training set, with 15 images obtained from each of the four types of dental images, e.g. panoramic, bitewing, lower periapical and upper periapical images. The proposed image classification algorithm was tested on a set of 123 images, in which 9 are panoramic images, 82 are bitewing images, 15 are lower periapical images and 17 are upper periapical images. Fig. 15 shows few of the images that we used in the experiments. In

Conclusion and future work

In this paper we presented a content-based archiving and retrieval system of dental images for use in human identification. It contains three major stages: dental image classification, bitewing image segmentation and retrieval based on teeth shapes using bidirectional Hausdorff distance. In the classification stage, two features are proposed for dental X-ray image classification. The classified bitewing images are segmented to extract the contours of molars and premolars, which are then used to

About the Author.—JINDAN ZHOU received the B.S. and M.S. degrees in Biomedical Engineering from Southeast University, Nanjing, China, in 1999 and 2002. She is now a Ph.D. student at the Department of Electric and Computer Engineering, University of Miami, Coral Gables, FL, USA. Her current research interests include image processing, pattern recognition and biometrics.

References (28)

F.S. Malkowski
Forensic dentistry, a study of personal identification
Dent. Stud.
(1972)
V.W. Weedn
Postmortem identifications of remains
Clin. Lab. Med.
(1998)
R.B. Dorion
Disasters big and small
J. Can. Dent. Assoc.
(1990)
I.A. Pretty et al.
A look at forensic dentistry—Part 1: The role of teeth in the determination of human identity
Br. Dent. J.
(2001)
American Society of Forensic Odontology, Forensic Odontology News, vol. 16, no. 2, Summer...
United States Army Institute of Dental Research Walter Reed Army Medical Center, Computer assisted post mortem...
Dr. Jim McGivney et al., WinID2 software...
Jonasson, Bankvall, Kiliaridis, Estimation of skeletal bone mineral density by mean of the trabecular pattern of the...
The Canadian Dental Association, Communique, May/June...
P. Stimson et al.
Forensic Dentistry
(1997)

Gustafson, Ghosta, Forensic Odontology, American Elsevier Pub. Co.,...

A. Jain et al.

Biometric identification

Commun. ACM

(2000)

S. Pankanti, R. Bolle, A. Jain, Biometrics: the future of identification, IEEE Comput. (2000)...

A.K. Jain, H. Chen, S. Minut, Dental biometrics: human identification using dental radiographs, Proceedings of the...

Cited by (146)

A comprehensive survey of deep learning algorithms and applications in dental radiograph analysis
2023, Healthcare Analytics
The Integration of machine learning and traditional image processing in dentistry has resulted in many applications like automatic teeth identification and numbering, caries, anomaly, disease detection, and dental treatment prediction. They have a broad scope in different applications observed in the dentistry literature review. This study reviews the literature on deep learning and dental radiograph analysis. We present an overview of machine learning algorithms in different areas of dentistry: tooth identification and numbering, Dental disease detection, and dental predictive treatment models. The methods under each area are briefly discussed. The dental radiograph data set required for performing experiments is summarized from the available literature. The study concludes by discussing new research opportunities and initiatives in this field. This paper offers a comprehensive overview of this innovative, challenging, and growing area in dentistry.
A novel method of human identification based on dental impression image
2023, Pattern Recognition
In large-scale natural disasters and special criminal cases, surface features of bodies, such as faces and fingerprints, are easily destroyed. Teeth possess strong high-temperature resistance, corrosion resistance, and high hardness, which can compensate for the shortcomings of the aforementioned situations. This paper proposes an identification method based on the aggregated features of multi-scale dental impression images. Firstly, a method exploiting the adaptive object detection method based on YOLOv8 is proposed to segment toothprints. Next, a novel geometric feature named calibrated offset distance is extracted, combined with the SIFT feature method, to extract multi-scale and multi-dimensional features from the global toothprint, local toothprints, and single-tooth prints. Finally, all features are aggregated to enhance the descriptive ability and robustness. Experimental results indicate that the method proposed in this paper demonstrates good identification performance.
A fine-grained network for human identification using panoramic dental images
2022, Patterns
When accidents occur, panoramic dental images play a significant role in identifying unknown bodies. In recent years, deep neural networks have been applied to address this task. However, while tooth contours are significant in classical methods, few studies using deep learning methods devise an architecture specifically to introduce tooth contours into their models. Since fine-grained image identification aims to distinguish subordinate categories by specific parts, we devise a fine-grained human identification model that leverages the distribution of tooth masks to distinguish different individuals with local and subtle differences in their teeth. First, a bilateral branched architecture is designed, of which one branch was designed as the image feature extractor, while the other was the mask feature extractor. In this step, the mask feature interacts with the extracted image feature to perform elementwise reweighting. Additionally, an improved attention mechanism was used to make our model concentrate more on informative positions. Furthermore, we improved the ArcFace loss by adding a learnable parameter to increase the loss of those hard samples, thereby exploiting the potential of our loss function. Our model was tested on a large dataset consisting of 23,715 panoramic X-ray dental images with tooth masks from 10,113 patients, achieving an average rank-1 accuracy of 88.62% and rank-10 accuracy of 96.16%.
A relation-based framework for effective teeth recognition on dental periapical X-rays
2022, Computerized Medical Imaging and Graphics
Citation Excerpt :
For oral disease diagnosis, dentists first need to recognize the teeth in dental X-rays, including finding their location and classify the teeth positions, which means that the teeth recognition is a key step in the whole process. Meanwhile, unlike the face and fingerprint features which are difficult to obtain when one died from disaster or traffic accident, the teeth can remain unbroken because they are the hardest part of humans, so that the court doctors can use teeth for person recognition (Tohnak et al., 2007; Nomir and Abdel-Mottaleb, 2005; Zhou and Abdel-Mottaleb, 2005) and postmortem identification (Daniels and Troy, 1984; Nomir and Abdel-Mottaleb, 2007; Corrêa et al., 2020). Therefore, the teeth detection and classification is very important for both dentists and court doctors.
Dental periapical X-rays are used as a popular tool by dentists for diagnosis. To provide dentists with diagnostic support, in this paper, we achieve automated teeth recognition of dental periapical X-rays by using deep learning techniques, including teeth location and classification. Convolutional neural network(CNN) is a popular method and has made large improvements in medical image applications. However, in our specific task, the performance of CNN is limited by lack of data and too many teeth positions in X-rays. Addressing this problem, we consider to utilize the prior dental knowledge, and therefore we propose a relation-based framework to handle the teeth location and classification task. According to the relation in teeth labels, we apply a special label reconstruction technique to decompose the teeth classification task, and use a multi-task CNN to classify the teeth positions. Meanwhile, for teeth location task, we design a proposal correlation module to use the information in teeth positions, and insert it into the multi-task CNN. A teeth sequence refinement module is used for the post processing. Our experiment results show that our relation-based framework achieves high teeth classification and location performance, which is a big improvement compared to the direct use of famous detection structures. With reliable teeth information, our method can provide automated diagnostic support for the dentists.
Gender classification on digital dental x-ray images using deep convolutional neural network
2021, Biomedical Signal Processing and Control
Human teeth are more durable in nature and tend to survive even after the death of a person. For this reason, teeth play a momentous role in the forensic department to identify a missing or dead person. Gender discrimination is also a key process in forensic analysis. However, designing an automated dental gender estimation based on deep learning techniques still needs more research efforts. This paper proposed an algorithm to estimate the gender of a human from dental x-ray image (DXI). It consists of three steps namely pre-processing, segmentation, and gender classification. In the initial step, the DXI is denoised to remove the unwanted noises such as Gaussian, speckle, and impulse using a prime magic square filter. In the second step, the gradient-based recursive thresholding segmentation is used to segment the denoised DXI. At the final stage, the image is classified using the Resnet50 deep convolutional neural network classification algorithm. The database of 1000 dental images has taken and divided into 600 training samples and 400 testing samples during experimental evaluation. Various performance measures of the proposed segmentation and classification method are evaluated. The proposed system has achieved 98.27% of classification accuracy. Moreover, the performance of the proposed work is compared with similar existing methods. It shows that the proposed system produced better results. Therefore, the proposed gender classification system supports the dentist in forensic applications to record the dental post mortem details in case of a big disaster.
Cascaded convolutional networks for automatic cephalometric landmark detection
2021, Medical Image Analysis
Cephalometric analysis is a fundamental examination which is widely used in orthodontic diagnosis and treatment planning. Its key step is to detect the anatomical landmarks in lateral cephalograms, which is time-consuming in traditional manual way. To solve this problem, we propose a novel approach with a cascaded three-stage convolutional neural networks to predict cephalometric landmarks automatically. In the first stage, high-level features of the craniofacial structures are extracted to locate the lateral face area which helps to overcome the appearance variations. Next, we process the aligned face area to estimate the locations of all landmarks simultaneously. At the last stage, each landmark is refined through a dedicated network using high-resolution image data around the initial position to achieve more accurate result. We evaluate the proposed method on several anatomical landmark datasets and the experimental results show that our method achieved competitive performance compared with the other methods.

View all citing articles on Scopus

About the Author—MOHAMED ABDEL-MOTTALEB received his Ph.D. in computer science from University of Maryland, College Park, in 1993. He is an associate professor in the department of Electrical and Computer Engineering, University of Miami, where his research focuses on 3D face recognition, dental biometrics, visual tracking, and human activity recognition. Prior to joining the University of Miami, from 1993 to 2000, he worked at Philips Research, Briarcliff Manor, NY. At Philips Research, he was a Principal Member of Research Staff and a Project Leader, where he led several projects in image processing, and content-based multimedia retrieval. He holds 20 US patents and published over 60 papers in the areas of image processing, computer vision, and content based retrieval. He is an associate editor for the Pattern Recognition journal.

^☆: This research is supported in part by the US National Science Foundation under Award number EIA-0131079, the research is also supported under Award number 2001-RC-CX-K013 from the Office of Justice Programs, National Institute of Justice, US Department of Justice.

View full text

A content-based system for human identification based on bitewing dental X-ray images☆

Abstract

Introduction

Section snippets

System components

Experimental results

Conclusion and future work

Forensic dentistry, a study of personal identification

Dent. Stud.

Postmortem identifications of remains

Clin. Lab. Med.

Disasters big and small

J. Can. Dent. Assoc.

A look at forensic dentistry—Part 1: The role of teeth in the determination of human identity

Br. Dent. J.

Forensic Dentistry

Biometric identification

Commun. ACM