Gray level co-occurrence and random forest algorithm-based gender determination with maxillary tooth plaster images
Introduction
Teeth are important for breaking down food and chewing, as well as producing correct sounds while talking, providing development of supportive tissues by simultaneously protecting them, and determining identity information about a person. Teeth have durable, resistant features, and can protect their own structure despite catastrophes thus being helpful in identification in forensic and archeology science. Gender is one of the properties of identity, even while improvements can be made by reduction in the cluster when a search is being performed.
In the literature, it has been seen that teeth have differences depending on gender. Parekh et al. [1] stated that the width of the canine arch in males was significantly larger than that of females. In the study of Hasanreisoğlu et al. [2], it was indicated that the size of the central incisors and canines in males was wider than those of females. It was concluded by Forster et al. [3] that the width of the dental arch was related to gender and vertical morphology of the face. In the study of Louly et al. [4], when the dental arch was investigated in a population of Brazilian children, it was found that males had more maxillary depth. In another study by Olmez and Doğan [5], it was discovered that the dental arch of males had more depth and width when compared to females. Horvath et al. [6] found the correlation between maxillary anterior form and gender in 3D data indicated that the maxillary anterior had differences according to gender. In the study of Shin [7], eighteen properties of Principle Component Analysis of maxillary tooth plaster models were classified according to k Nearest Neighbors (kNN) and its algorithm, with a 76% success rate.
When the literature was further investigated, gender determination from teeth can be seen by manual measurements in forensic science and dentistry. However, in the computer sciences, there are inadequate studies on this subject. The purpose of this study is to identify gender determination with features extracted automatically from maxillary tooth plaster model images, without requiring manual measurement. The main contributions of our study to the literature are as follows:
- •
Automatic segmentation was carried out with image processing methods, with features being extracted automatically. Rapid gender determination was performed with this fully automatic designed system.
- •
This study is multi-disciplinary, one in which there are benefits in dentistry, forensic science, and computer sciences.
- •
Size of teeth vary from one population to another. Our proposed method is an automatic system in which every population can adapt to it.
The manuscript is organized as follows: materials and methods used for the proposed system are discussed in the second part. The experimental results are presented in the third part. Finally, the manuscript is finalized with focused remarks.
Section snippets
Materials and methods
To perform fully automated gender determination from maxillary tooth plaster model images of individuals, a system is designed where steps can be seen in Fig. 1.
In the first stage of the system, the segmentation process is handled by standard image. Gray Level Co-occurrence Matrix (GLCM) of segmented image is formed, with specific features extracted from this matrix. The extracted features are classified by the RF algorithm, at which point gender determination is performed.
Data set
To create a data set, a working group consisted of 40 people (20 females, 20 males) belonging to the Turkish population between 21–24 years old by taking 2015/002 numbered Ethics Committee approval from Necmettin Erbakan University Faculty of Dentistry. Maxillary tooth plaster models of each person were used.
The tooth's shape, size, and position can be altered by prosthetic treatment, such that the prosthesis is harmonized with dentofacial and dentoalveolar structures. Prosthodontists can make
Conclusion
Gender determination is one of the most popular fields in computer science. Gender determination studies were carried out with face, walking, and signature patterns when the literature was investigated [14], [15], [16].
Since teeth have a hard and resistant structure, they are important sources for gender determination in situations such as accidents and disasters. Gender determination studies through teeth are generally performed in dentistry and forensic science. In this study, a
Acknowledgments
The authors acknowledge TUBITAK (The Scientific and Technological Research Council of Turkey) for 2211 scholarship.
Betül Akkoç received the B.Sc. degree from Selçuk University in 2010 and the M.Sc. degree from Selçuk University, Konya, Turkey, in 2012, both in Computer Engineering. She is currently working towards her Ph.D. at Selçuk University. Her research interests are in data mining, machine learning and image processing.
References (20)
- et al.
An analysis of maxillary anterior teeth: facial and dental proportions
J. Prosthet. Dent.
(2005) - et al.
Calculation of grey level co-occurrence matrix-based seismic attributes in three dimensions
Comput. Geosci.
(2013) - et al.
Predicting habitat suitability with machine learning models: The potential area of Pinus sylvestris L. in the Iberian Peninsula
Ecol. Model.
(2006) - et al.
Learning to classify e-mail
Inf. Sci.
(2007) - et al.
Odontometric study of Maxillary Canine teeth to establish sexual dimorphism in Gujarat population
Int. J. Biol. Med. Res.
(2012) - et al.
Relationship between dental arch width and vertical facial morphology in untreated adults
Eur. J. Orthod.
(2008) - et al.
Dental arch dimensions in the mixed dentition: a study of Brazilian children from 9 to 12 years of age
J. Appl. Oral Sci.
(2011) - et al.
Comparison of the arch forms and dimensions in various malocclusions of the Turkish population
Open J. Stomatol.
(2011) - et al.
The correlation between anterior teeth form and gender – a 3D analysis in humans
Eur. J. Esthet. Dent
(2012) Gender identification on the teeth based on principal component analysis representation
Lect. Notes Comput. Sci.
(2006)
Cited by (0)
Betül Akkoç received the B.Sc. degree from Selçuk University in 2010 and the M.Sc. degree from Selçuk University, Konya, Turkey, in 2012, both in Computer Engineering. She is currently working towards her Ph.D. at Selçuk University. Her research interests are in data mining, machine learning and image processing.
Ahmet Arslan received the B.Sc. and the M.Sc. degree in electric–electronic engineering both from Fırat University, Elazığ, Turkey and the Ph.D. degree in electric–electronic engineering from Bilkent University, Ankara, Turkey, in 1984, 1987 and 1992, respectively. He is the Professor of Department of Computer Engineering since 2015 in the University of Konya Food & Agriculture, Konya, Turkey. His research interests are in data mining, machine learning, computer graphics, and computer aided design.
Hatice Kök received the M.Sc. degree from Selçuk University Faculty of Dentistry in 2003 and the Ph.D. degree in Selcuk University Faculty of Dentistry Department of Orthodonti in 2009. She is the Assistant Professor Dr. of Department of Orthodonti since 2015 in the University of Necmettin Erbakan, Konya, Turkey. Her research interests are in canine retraction, segmental springs.