A Preliminary Study Of Quantitative Cervical Vertebral Maturation Based On Computer Aided Cervical Vertebra Analysis

【Objective】To investigate the accuracy and repeatability of semiautomatic identification of landmarks using Computer Aided Cervical Vertebra Analysis(CACVA).【Methods】120 adolescents (60 boys, 60 girls) aged from 9.1 to 17.2 years old were randomly selected. 27 landmarks from the second to fifth Cervical Vertebra were identified three times by CACVA, which identified and calculated the landmarks by fast marching method and parabolic curve fitting. A senior expert(SE) and a junior expert(JE) did the same work with manual method, respectively. The identification accuracy and repeatability among CACVA, SE and JE were compared by ANOVA and group pair-wise comparison.【Results】As for the accuracy, there was no significant difference between CACVA and SE (P>0.05). However, there was significant difference between CACVA and JE, as well as SE and JE (P<0.05). As for the repeatability, CACVA was better than SE and JE.【Conclusion】CACVA has similar accuracy, less workload and better repeatability. It has a bright future in orthodontic clinical and scientific research. 【Objectve】To establish a quantitative cervical vertebral maturation (CVM) system for adolescents with normal occlusion using Computer Aided Landmarking.【Methods】Mixed longitudinal data were used. The subjects included 87 children and adolescents from 8 to 18 years old with normal occlusion (32 boys, 55 girls) selected from 901 candidates. Sequential lateral cephalograms and hand-wrist films were taken once a year for 6 years. The lateral cephalograms of all subjects were divided into 11 maturation groups according to the Fishman skeletal maturity indicators. 31cervical landmarks from the second to fifth Cervical Vertebra were identified. In this study, the System of Computer Aided Cervical Vertebra Analysis(CACVA) was developed, which identified the landmarks by fast marching method and parabolic curve fitting. 153 cervical parameters including length, height, width, area, perimeter, curvature of the concavity of the upper and lower border were calculated. Correlation-based Feature Selection (CFS) algorithm was used to remove irrelevant and redundant parameters, and then establish logistic regression equations to assess cervical vertebral maturation with the best combination of parameters.【Results】5 groups of parameters combination were selected to determine the classification of CVM. The cervical vertebral maturation system included 5 maturational stages(CVM1-CVM5) is shown as follows: CVM 1= 13.59 +( -1.3)×MH3 CVM 2= 1.13 +( -0.22)×AD5+(-0.62)×H4 CVM 3= 2.15+(-0.07)×AD3+(0.24)×LD3+(-5.08)×AH4/PH4 CVM 4=(-4.56) + 0.13×AD4 +0.67×MH3/LW3 +2×AD4/VD4 CVM 5=(-15.41) + 0.93×AH4+0.53×MH4【Conclusions】This method used the computer-aided landmarking with better repeatability. Sensitive parameters were selected from hundreds of cervical parameters. The best combined parameters were used to evaluate the cervical vertebral maturation, which was not dependent on a particular parameter at every stage. This method provided an objective and fair approach to assess the level of skeletal maturation during adolescence.