Sekazu: an integrated solution tool for gender determination based on machine learning models

• Main Authors: Muhammed Kamil Turan, Eftal Sehirli, Zulal Oner, Serkan Oner
• Format: Article
• Language: English
• Published: Society of TURAZ AKADEMI 2021-06-01
• Series: Medicine Science
• Subjects: gender determination; image labeling; bone measurement; machine learning models; gui
• Online Access: http://www.ejmanager.com/fulltextpdf.php?mno=15884

Gender determination is the first stage of identification used in forensic investigation, anthropology, archeology, and bioarchaeology, which helps accelerate the process of narrowing possible matches in a medical-legal context. Without DNA analysis, the dimorphic property of bones comprises a basis for gender determination with measurements taken on only bones. In this work, 9 different bones such as cranium, mandibula, femur, patella, calcaneus, condylus occipitalis, sternum, hand bones, and foot bones were used for gender determination. Machine learning methods and artificial neural networks, especially linear and quadratic discriminant analysis, while determining the gender, machine learning also were technically adopted. 13 different machine learning algorithms were used as a model for gender determination. Many tools were designed to perform processes like designing necessary bookmarks to try models, designing measurements where machine learning algorithms are used as features, determining coordinates of designed bookmarks, and computation of features. A software named Sekazu was developed by presenting an integrated solution proposal. Thanks to the developed software, models used in gender determination were developed and tried in a fast way and researchers can obtain results reported based on performance metrics flexibly. [Med-Science 2021; 10(2.000): 367-73]