Automatic gender determination from 3D digital maxillary tooth plaster models based on the random forest algorithm and discrete cosine transform. (May 2017)

• Record Type: Journal Article
• Authors: Akkoç, Betül; Arslan, Ahmet; Kök, Hatice
• Is Part Of: Computer methods and programs in biomedicine. Volume 143(2017)
• Page Start: 59
• Page End: 65
• Publication Date: 2017-05
• Subjects: Gender determination -- Feature extraction -- Image processing -- Random forest algorithm; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2017.03.001
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: In this study, DCT and RF based gender determination system has been proposed. 3D digital models of the maxillary tooth plaster models have been used to determine gender. Segmentation and feature extraction process has been carried out automatically. A multi-disciplinary study has been realized. Abstract: Background and Objective: One of the first stages in the identification of an individual is gender determination. Through gender determination, the search spectrum can be reduced. In disasters such as accidents or fires, which can render identification somewhat difficult, durable teeth are an important source for identification. This study proposes a smart system that can automatically determine gender using 3D digital maxillary tooth plaster models. Methods: The study group was composed of 40 Turkish individuals (20 female, 20 male) between the ages of 21 and 24. Using the iterative closest point (ICP) algorithm, tooth models were aligned, and after the segmentation process, models were transformed into depth images. The local discrete cosine transform (DCT) was used in the process of feature extraction, and the random forest (RF) algorithm was used for the process of classification. Results: Classification was performed using 30 different seeds for random generator values and 10-fold cross-validation. A value of 85.166% was obtained for average classification accuracy (CA) and a value of 91.75% for the area under the ROC curve (AUC). Conclusions: A multi-disciplinary study is performed here that includes computer sciences, medicine and dentistry. A smart system is proposed for the determination of gender from 3D digital models of maxillary tooth plaster models. This study has the capacity to extend the field of gender determination from teeth.