Lizhe Xie1,2, Wen Tang1,3, Iman Izadikhah1,3, Zhenqi Zhao4, Yang Zhao5, Hu Li2,3, Bin Yan6,7,8. 1. Jiangsu Province Key Laboratory of Oral Diseases, Nanjing Medical University, Nanjing, China. 2. Jiangsu Province Engineering Research Center of Stomatological Translational Medicine, Nanjing Medical University, 136 Hanzhong Street, Gulou District, Nanjing, 210029, China. 3. Department of Orthodontics, Affiliated Hospital of Stomatology, Nanjing Medical University, Gulou District, 136 Hanzhong Street, Nanjing, 210029, China. 4. Department of Stomatology, The First People's Hospital of Nantong, Nantong, China. 5. School of Public Health, Department of Biostatistics, Nanjing Medical University, Jiangning District, Nanjing, China. 6. Jiangsu Province Key Laboratory of Oral Diseases, Nanjing Medical University, Nanjing, China. byan@njmu.edu.cn. 7. Jiangsu Province Engineering Research Center of Stomatological Translational Medicine, Nanjing Medical University, 136 Hanzhong Street, Gulou District, Nanjing, 210029, China. byan@njmu.edu.cn. 8. Department of Orthodontics, Affiliated Hospital of Stomatology, Nanjing Medical University, Gulou District, 136 Hanzhong Street, Nanjing, 210029, China. byan@njmu.edu.cn.
Abstract
PURPOSE: Nowadays, the integration of Artificial intelligence algorithms and quantified radiographic imaging-based diagnostic procedures is hailing amplified deliberation particularly in assessment of skeletal maturity. So we intend to formulate a logistic regression model for intelligent and quantitative estimation of Fishman skeletal maturation index (SMI) based on the parameters attained from the cervical vertebrae CBCT images of Chinese girls. METHODS: From 709 hand wrist radiographs and CBCT images, 447 samples were randomly selected (called as G1) to build a logistic regression model. The reliability and reproducibility were assessed by the intraclass correlation coefficient (ICC) and weighted Cohen's kappa, followed by Spearman's rank correlation coefficient to identify the parameters significantly associated with the SMI. Two hundred and sixty-two other subjects (named G2) were recruited for external examination of the models by direct visual comparison and the receiver operating characteristic (ROC) curve. In cases of confusion and mispredictions, the model was modified to improve the consistency. RESULTS: Five significant parameters (Chronological age, C3 height (H3)[Formula: see text], C4 upper width (UW4), C4 lower width (LW4), and the ratio of posterior height to lower width of C4 ([Formula: see text]) were administered into logistic regression model. Despite total agreement percentage which was 84% (total AUC = 0.92), unsatisfactory performance was noticed for the 6th and 8th stages which were confused with their neighboring stages. After adjustments of the models, the total agreement percentage and AUC were upgraded to 88% and 0.96, respectively. CONCLUSION: Consistency and fitness evaluation of our models demonstrated adequate prediction percentage and reliability for automated classification of skeletal maturation. The presented constructed logistic regression model has the potential to serve as a maturity evaluation index in clinical craniofacial orthopedics in Chinese girls. The proposed model in this study showed promising strength for being expended in the event of other clinical multi-stage conditions.
PURPOSE: Nowadays, the integration of Artificial intelligence algorithms and quantified radiographic imaging-based diagnostic procedures is hailing amplified deliberation particularly in assessment of skeletal maturity. So we intend to formulate a logistic regression model for intelligent and quantitative estimation of Fishman skeletal maturation index (SMI) based on the parameters attained from the cervical vertebrae CBCT images of Chinese girls. METHODS: From 709 hand wrist radiographs and CBCT images, 447 samples were randomly selected (called as G1) to build a logistic regression model. The reliability and reproducibility were assessed by the intraclass correlation coefficient (ICC) and weighted Cohen's kappa, followed by Spearman's rank correlation coefficient to identify the parameters significantly associated with the SMI. Two hundred and sixty-two other subjects (named G2) were recruited for external examination of the models by direct visual comparison and the receiver operating characteristic (ROC) curve. In cases of confusion and mispredictions, the model was modified to improve the consistency. RESULTS: Five significant parameters (Chronological age, C3 height (H3)[Formula: see text], C4 upper width (UW4), C4 lower width (LW4), and the ratio of posterior height to lower width of C4 ([Formula: see text]) were administered into logistic regression model. Despite total agreement percentage which was 84% (total AUC = 0.92), unsatisfactory performance was noticed for the 6th and 8th stages which were confused with their neighboring stages. After adjustments of the models, the total agreement percentage and AUC were upgraded to 88% and 0.96, respectively. CONCLUSION: Consistency and fitness evaluation of our models demonstrated adequate prediction percentage and reliability for automated classification of skeletal maturation. The presented constructed logistic regression model has the potential to serve as a maturity evaluation index in clinical craniofacial orthopedics in Chinese girls. The proposed model in this study showed promising strength for being expended in the event of other clinical multi-stage conditions.
Authors: Gustavo Echevarría-Sánchez; Luis Ernesto Arriola-Guillén; Violeta Malpartida-Carrillo; Pedro Luis Tinedo-López; Ricardo Palti-Menendez; Maria Eugenia Guerrero Journal: Int Orthod Date: 2020-01-31