Diana Toneva, Silviya Nikolova, Gennady Agre, Dora Zlatareva, Nevena Fileva, Nikolai Lazarov
{"title":"基于下颌骨测量的性别估计。","authors":"Diana Toneva, Silviya Nikolova, Gennady Agre, Dora Zlatareva, Nevena Fileva, Nikolai Lazarov","doi":"10.1127/anthranz/2023/1733","DOIUrl":null,"url":null,"abstract":"<p><p>Medical imaging and machine learning are beneficial approaches in physical and forensic anthropology. They are particularly useful for the development of models for sex identification based on bone remains. The present study uses machine learning algorithms to create models for sex estimation based on mandibular measurements. The sample included head CT scans of 239 adult Bulgarians (116 males and 123 females). Three-dimensional coordinates of 45 landmarks of the mandible were acquired from segmented polygonal models of the skulls of these individuals. Two datasets of mandibular measurements were assembled. The first dataset included 51 measurements: linear, projective, and angular measurements. The second dataset included 990 interlandmark distances. Seven machine learning algorithms (Support Vector Machines, Neural Network, Naïve Bayes, Random Forest, J48, JRip, and Logistic Regression) were applied to the two datasets, and the classification accuracy was evaluated by 10x5-cross-validation. The selection of the best subsets of attributes specific to each of the abovementioned algorithms was done based on the attribute importance evaluated by an attribute selection scheme. In general, the sub-symbolic algorithms achieved higher results than the symbolic ones, except for the logistic regression. The best classification model was learnt by the Support Vector Machines algorithm, which achieved an accuracy of 95.3% on a dataset described by 19 interlandmark distances. In both datasets, the application of advanced attribute selection has led to an increase in the classification accuracy of all algorithms used in the experiments.</p>","PeriodicalId":46008,"journal":{"name":"Anthropologischer Anzeiger","volume":null,"pages":null},"PeriodicalIF":0.4000,"publicationDate":"2024-01-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Sex estimation based on mandibular measurements.\",\"authors\":\"Diana Toneva, Silviya Nikolova, Gennady Agre, Dora Zlatareva, Nevena Fileva, Nikolai Lazarov\",\"doi\":\"10.1127/anthranz/2023/1733\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>Medical imaging and machine learning are beneficial approaches in physical and forensic anthropology. They are particularly useful for the development of models for sex identification based on bone remains. The present study uses machine learning algorithms to create models for sex estimation based on mandibular measurements. The sample included head CT scans of 239 adult Bulgarians (116 males and 123 females). Three-dimensional coordinates of 45 landmarks of the mandible were acquired from segmented polygonal models of the skulls of these individuals. Two datasets of mandibular measurements were assembled. The first dataset included 51 measurements: linear, projective, and angular measurements. The second dataset included 990 interlandmark distances. Seven machine learning algorithms (Support Vector Machines, Neural Network, Naïve Bayes, Random Forest, J48, JRip, and Logistic Regression) were applied to the two datasets, and the classification accuracy was evaluated by 10x5-cross-validation. The selection of the best subsets of attributes specific to each of the abovementioned algorithms was done based on the attribute importance evaluated by an attribute selection scheme. In general, the sub-symbolic algorithms achieved higher results than the symbolic ones, except for the logistic regression. The best classification model was learnt by the Support Vector Machines algorithm, which achieved an accuracy of 95.3% on a dataset described by 19 interlandmark distances. In both datasets, the application of advanced attribute selection has led to an increase in the classification accuracy of all algorithms used in the experiments.</p>\",\"PeriodicalId\":46008,\"journal\":{\"name\":\"Anthropologischer Anzeiger\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.4000,\"publicationDate\":\"2024-01-25\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Anthropologischer Anzeiger\",\"FirstCategoryId\":\"90\",\"ListUrlMain\":\"https://doi.org/10.1127/anthranz/2023/1733\",\"RegionNum\":4,\"RegionCategory\":\"社会学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"ANTHROPOLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Anthropologischer Anzeiger","FirstCategoryId":"90","ListUrlMain":"https://doi.org/10.1127/anthranz/2023/1733","RegionNum":4,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"ANTHROPOLOGY","Score":null,"Total":0}
Medical imaging and machine learning are beneficial approaches in physical and forensic anthropology. They are particularly useful for the development of models for sex identification based on bone remains. The present study uses machine learning algorithms to create models for sex estimation based on mandibular measurements. The sample included head CT scans of 239 adult Bulgarians (116 males and 123 females). Three-dimensional coordinates of 45 landmarks of the mandible were acquired from segmented polygonal models of the skulls of these individuals. Two datasets of mandibular measurements were assembled. The first dataset included 51 measurements: linear, projective, and angular measurements. The second dataset included 990 interlandmark distances. Seven machine learning algorithms (Support Vector Machines, Neural Network, Naïve Bayes, Random Forest, J48, JRip, and Logistic Regression) were applied to the two datasets, and the classification accuracy was evaluated by 10x5-cross-validation. The selection of the best subsets of attributes specific to each of the abovementioned algorithms was done based on the attribute importance evaluated by an attribute selection scheme. In general, the sub-symbolic algorithms achieved higher results than the symbolic ones, except for the logistic regression. The best classification model was learnt by the Support Vector Machines algorithm, which achieved an accuracy of 95.3% on a dataset described by 19 interlandmark distances. In both datasets, the application of advanced attribute selection has led to an increase in the classification accuracy of all algorithms used in the experiments.
期刊介绍:
AA is an international journal of human biology. It publishes original research papers on all fields of human biological research, that is, on all aspects, theoretical and practical of studies of human variability, including application of molecular methods and their tangents to cultural and social anthropology. Other than research papers, AA invites the submission of case studies, reviews, technical notes and short reports. AA is available online, papers must be submitted online to ensure rapid review and publication.