Evaluation Of Hand Function Of Children With Cerebral Palsy Based On Structural MRI Of Hand Knob

Objective:Based on structure magnetic resonance imaging（MRI）of hand knob,the relationship between hand knob morphology and its structural indicators and hand function in children with cerebral palsy（CP）was evaluated.Method:1.For 70 children with clinically diagnosed CP collected from May 2019 to September 2021,three-dimensional T1-weighted image（3D-T1WI）,manual ability classification system（MACS）assessment.According to MACS,children with CP were divided into mild impairment group（grade Ⅰ-Ⅲ）and severe impairment group（grade Ⅳ-Ⅴ）.2.Distribution of hand knob types:Based on 3D-T₁WI,Kappa statistic was used to evaluate the consistency of hand knob types within and among assessors;Fisher’s exact probability method was used to compare tistributional differences of hand knob types in CP clinical type,MRI injury type and the classification of hand function.3.To construct a hand function evaluation model for children with CP:Based on3D-T₁WI,measure the hand knob structure indicators（hand segment width and height;the distance between the midpoint of the hand knob and the midline of the brain;the thickness of the hand knob cortex and the height of the white matter of the hand knob）.Using intra-class correlation coefficients to test the intra-rater and inter-rater consistency of hand knob structure indicators.（1）Based on structure indicators of hand knob,Logistic Ridge regression was used to construct a hand konb scoring model;（2）Multivariate Logistic regression was used to construct a clinical feature model based on the clinical characteristics of CP children;（3）Based on CP children’s hand knob score and clinical characteristics,using multivariate Logistic regression to build a hand knob score-clinical features combined model.Stepwise regression was performed on the model according to the maximum likelihood method,and the best model was selected using the Akaike information criterion;the model discrimination and fit were evaluated by receiver operating characteristic curve and calibration curve.Finally,create a nomogram to visualize the model.Results:1.A total of 70 children with CP were included,including 53 cases in the hand function mild impairment group and 17 cases in the severe hand function impairment group.There was no significant difference in gender,age and gestational age between the two groups（P>0.05）.1.There were 53 cases in the hand function mild impairment group and 17 cases in the severe hand function impairment group.There was no significant difference in gender,age and gestational age between the two groups（P>0.05）.2.Distribution of hand knob types:（1）The invertedΩ-shaped,inner asymmetricω-shaped and multi-peak hand knob of children with CP are mainly distributed in the type of spastic cerebral palsy.The brain MRI manifestations of children with CP with these three types of hand knob were mainly white matter injury,and the degree of hand functional injury was mild;the proportion of hand knob morphology distribution was not statistically significant in the clinical types of CP,MRI manifestations and manual ability classification（P>0.05）;（2）CP children in the left and right cerebral hemisphere hand knob consistent group had less hand function impairment than the inconsistent group（P<0.05）.3.To construct a hand function evaluation model for children with CP:（1）In the hand knob score model,the odds ratio（OR）value of the hand knob score was 85.344;（2）In the clinical feature model,the OR of white matter injury among MRI injury types was obtained the value was 0.048;（3）In the hand knob score-clinical features combined model,the OR values of hand knob score and white matter injury were 8.633 and 0.052,respectively.The areas under the curve of hand knob score model,clinical feature model and hand knob score-clinical feature combined model for predicting the degree of hand function impairment were 0.752,0.819,and 0.880,respectively.The calibration curve showed that the predicted probability of hand function impairment and the actual occurrence probability of the hand knob score-clinical features combined model had good consistency.Conclusion:Based on the structural index of hand knob in children with CP and the classification of MRI injuries,the hand knob score-clinical features combined model can be used to preliminarily assess the degree of hand impairment in children with CP.