Fei Fan1, Meng Tu1, Rui Li2, Xinhua Dai1, Kui Zhang1, Hu Chen2, Feijun Huang1, Zhenhua Deng1,3. 1. West China School of Basic Medical Sciences & Forensic Medicine, Sichuan University, Chengdu, China. 2. College of Computer Science, Sichuan University, Chengdu, China. 3. Key Laboratory of Evidence Science (China University of Political Science and Law), Ministry of Education.
Abstract
OBJECTIVES: This study aimed to explore whether computed tomography (CT) images of cranial sutures can contribute to adult age estimation in Chinese Han individuals. MATERIALS AND METHODS: This study was based on cranial CT scans of 230 Chinese Han males aged 23.33-76.93 years. A total of 160 images from 16 suture segments were scored after volume reformation and multiplanar reconstruction in each individual. Decision tree regression, linear support vector regression, Bayesian ridge regression, and gradient boosting regression were developed for adult age estimation by a training set using leave-one-out cross-validation and further evaluated by the test set. The inaccuracy and bias were calculated to evaluate the four models and the previously used models from the literature. RESULTS: The degree of suture closure was associated with adult age. The minimum inaccuracy of the test set was 7.73 years obtained by linear support vector regression, while the inaccuracy of previous simple linear regression models was 13.09 and 10.97 years. The accuracy was higher in the age group from 40.00 to 59.99 years compared to the other age groups. DISCUSSION: The accuracy of our models for adult age estimation was superior to those in previous studies based on cranial sutures. Hence, the application of novel statistical data mining tools helps to improve aging issues. Nevertheless, age estimation of adults should be combined with other methods, since the accuracy level is still not satisfactory.
OBJECTIVES: This study aimed to explore whether computed tomography (CT) images of cranial sutures can contribute to adult age estimation in Chinese Han individuals. MATERIALS AND METHODS: This study was based on cranial CT scans of 230 Chinese Han males aged 23.33-76.93 years. A total of 160 images from 16 suture segments were scored after volume reformation and multiplanar reconstruction in each individual. Decision tree regression, linear support vector regression, Bayesian ridge regression, and gradient boosting regression were developed for adult age estimation by a training set using leave-one-out cross-validation and further evaluated by the test set. The inaccuracy and bias were calculated to evaluate the four models and the previously used models from the literature. RESULTS: The degree of suture closure was associated with adult age. The minimum inaccuracy of the test set was 7.73 years obtained by linear support vector regression, while the inaccuracy of previous simple linear regression models was 13.09 and 10.97 years. The accuracy was higher in the age group from 40.00 to 59.99 years compared to the other age groups. DISCUSSION: The accuracy of our models for adult age estimation was superior to those in previous studies based on cranial sutures. Hence, the application of novel statistical data mining tools helps to improve aging issues. Nevertheless, age estimation of adults should be combined with other methods, since the accuracy level is still not satisfactory.