Prediction Of Survival Factors After Resection Of Childhood Neuroblastoma And Application Of CT Radiomics Features In Predicting Liver Regeneration Abilit

OBJECTIVE:To investigate the predictive value of preoperative neutrophil-tolymphocyte ratio（NLR）and serum C-reactive protein to albumin ratio（CAR）on the prognosis after resection of neu-roblastoma in children.METHODS: Subject operating characteristic（ROC）curves were used to determine the optimal cut-off values for the continuous variable CAR versus NLR.And patients were divided into three groups: low NLR with low CAR was defined as NLR-CAR 0,high NLR with high CAR was defined as NLR-CAR 2,and low NLR with high CAR or high NLR with low CAR was defined as NLR-CAR 1.Kaplan-Meier method and Logrank method were used for survival analysis.Univariate versus multivariate Cox proportional risk regression was used to determine independent factors of prognosis in NB patients.RESULTS: According to the ROC curve,the optimal cut-off value for NLR was 2.49 and for CAR was 0.035.Kaplan-Meier method showed that high NLR（>2.49）,high CAR（>0.035）and NLR-CAR 2 had poorer overall survival（P < 0.01）.Univariate and multifactorial Cox regression analysis showed that age（>18 months）,INSS stage（III-IV）,and NLR-CAR 2（high NLR and high CAR）were independent risk factors for the prognosis of NB patients（P < 0.05）.CONCLUSION: NLR-CAR 2（high NLR and high CAR）is a valuable biomarker for the prognosis of children with NB.OBJECTIVE: With the continuous development of medical technology,the application of CT imaging omics features and machine learning algorithms has attracted more and more attention.Liver regeneration is a special reaction after liver trauma or disease,usually requiring surgical intervention.However,effective methods for evaluating patients’ liver function and regeneration capacity are still lacking before and after liver resection.Therefore,this study aimed to explore the application of CT imaging omics features based on machine learning in predicting liver regeneration capacity.METHODS: In this study,CT images of 117 patients who underwent partial hepatectomy were collected,and the volume changes of liver were judged by a computeraided surgery system.The liver regeneration capacity was divided into groups,and the CT imaging omics features were analyzed to screen out the prediction-related features.The Lasson regression method was used to select features,and a logistic regression model was established.In addition,internal validation was performed using Bootstrap sampling,and the accuracy of the prediction model was verified using ROC curves and DCA.RESULTS: A total of 117 patients who underwent partial liver resection were included in our study,including 100 males and 17 females with an average age of 52 years.The preoperative average liver volume was（1433.829 ± 479.719）ml,and the mean tumor volume was（141.545 ± 396.335）ml.The postoperative average liver volume was（1032.624 ± 224.965）ml,and the average liver volume at half a year after surgery was（1218.197 ± 255.711）ml.The median value of liver regeneration rate was 13.34%.According to the size of the median value,we classified the patients into high liver regeneration group（59 cases）and low liver regeneration group（58 cases）.Through analyzing CT imaging omics and Lasson regression analysis,9 significant feature variables were included in the Logistic analysis and model building.Finally,three feature variables,lbp₂D_glrlm_Run Length Non Uniformity,lbp₂D_glrlm_Run Length Non Uniformityy,and wavelet_HLL_firstorder_Maximum,were used to construct a column graph.In addition,the area under the ROC curve was0.693,and the DCA curve showed that the predictive efficiency of the model was high.CONCLUSION:In conclusion,the prediction method of liver regeneration capacity based on machine learning algorithms and CT imaging omics features has high accuracy and practicality.The application of machine learning algorithms can improve the automated processing ability of medical images and thus improve the prediction accuracy of liver regeneration capacity.This prediction model can also play a role in treatment planning,surgical planning and prognostic evaluation,providing better medical services and treatment outcomes for patients.