The Application Of Face Recognition In The Screening Of Down's Syndrome Of Adolescents

ObjectiveThree dimensional imaging system was used to get facial features of adolescents and adolescents with Down's syndrome.A facial model of Down's syndrome was established and compared with the 3D face database of normal adolescents,and the diagnostic parameters of Down's syndrome were obtainedMethodsThe 3D face models of normal Chinese Han adolescents(age from 11-18Y)and the cases(age from 11-18Y)of Down's syndrome which diagnosed in Zhejiang province i were product by 3dMD Face system(3dMD company,USA)from November 2011 to October 2014.After the 3D data preprocessing,face alignment,facial feature points were extracted,and the parameters of face were recorded.Statistical analysis was used to analyze the facial development of adolescents of different ages.Facial feature parameters were measured and compared with the average faces of normal adolescents.The random forest algorithm was used to test and verify the face recognition of the male and female model.ResultsWe completed the automatic localization of 20 facial feature points in 1122 cases of 11-18 year old normal adolescents and 91 cases of the age of 11-18 normal adolescents with Down's syndrome.20 facial feature points and 19 facial feature parameters were obtained.A normal face database was established,and the average faces of adolescents at different ages were obtained.The statistical analysis found adolescents with Down's syndrome with wide outcanthal and small eyes,the eye spacing was relatively wide,and nose length,nasal height,ear length,ear width measurements were mainly distributed in the low percentile area,the face was narrow and small,usually flat.The male and female model successfully completed internal and external validation,with high sensitivity and specificity.It could be applied to the screening and diagnosis of adolescents with Down's syndrome.ConclusionThe special face features of of Down's syndrome of adolescents were distinctly different from the normal faces.This difference ensured the sensitivity and reliability of face recognition for Down's syndrome screening parameters,making it possible to screen for some inherited diseases such as Down's syndrome for large population.