| Background and objective:Autism spectrum disorder(ASD)is a neurodevelopmental disorder.Main clinical symptoms of ASD are mainly social,communication disorders,narrow areas of interest,and repetitive patterns of behavior.ASD not only affects the healthy growth of children,but also causes huge economic burden on the children's families,as well as mental and emotional pressure.The etiology of ASD is complex,including genetic factors,the mother's history of disease during pregnancy,and etc.Currently,the diagnosis of ASD is mainly based on Diagnostic and Statistical Manual of Mental Disorders-V and Childhood Autism Rating Scale,meanwhile,magnetic resonance imaging(MRI)technology and the clinical history are used to support diagnosis.Recently,diffusion tensor imaging(DTI),as a non-invasive,emerging MRI technology,is playing an increasingly important role in the diagnosis of ASD.However,the value of combined DTI technology and clinical parameters in diagnosing ASD is still unclear.The aim of this study is to investigate the value of the DTI's parameters,fractional anisortropy(FA),and clinical parameters in the diagnosis of ASD and construct a prediction model in order to easily and conveniently support the clinical diagnosis of ASD.Methodssixty ASD patients and matched normal developing children were selected as the research subjects from January 2019 to December 2019 in our hospital.All samples underwent DTI examinations to calculate FA values in various parts of the brain.At the same time,clinical parameters were collected retrospectively and the differences between the two groups were compared.The count data were tested by chi-square test.The measurement data which have the normal distribution using the t-test,otherwise were analyzed by the Mann-Whitney test.Tukey's multiple comparisons was used to analyze the difference of FA value among the control group,Light/medium group and severe group;R software was used to construct a generalized linear model;Hosmer-Lemeshow test was used to evaluate the calibration of the model,Receiver Operating Characteristic(ROC)was used to assess the discrimination of the model.Decision tree curve analysis was used to assess the clinical efficacy of the model,bilateral p<0.05 represents statistical significance.Results1)Univariate analysis showed that advanced maternal age,premature birth of the fetus,FA values of left and right corona radiate,left and right frontal white matter,genu of corpus callosum,and left inner capsule forelimb were statistically significant in the ASD and control groups.2)Multivariate analysis showed that advanced maternal age,premature birth of the child,FA values of the left corona radiate,left frontal white mattter,and genu of corpus callosum were independent risk factors for predicting the occurrence of ASD.3)The FA values of the left corona radiate,genu of corpus callosum,and right frontal cortex were statistically significant in the control group and the severe ASD group.The FA values of right corona radiate were statistically significant in light/medium and severe group.The FA value of genu of corpus callosum was statistically significant in the control group and the light/medium group.4)A generalized linear model based on the advanced maternal age,the history of child premature birth,FA values of the left corona radiate,the left frontal white matter,and the genu of corpus callosum can accurately predict the occurrence of ASD,AUC was 0.852,with a sensitivity of 80%and a specificity of 86.7%.5)The nomogram based on the model can simply and effectively support to diagnosis ASD in clinic.Conclusion1)Advanced maternal age and fetal premature birth are independent clinical risk factors for ASD?Decreased FA values of the left corona radiate,genu of corpus callosum,and left frontal white matter are independent predictors of ASD.2)A prediction model based on clinical parameters and FA values can accurately predict the occurrence of ASD,and may provide important auxiliary information for clinical diagnosis of ASD. |
PDF Full Text Request