The Value Of CT-based Radiomics Nomogram In Predicting Fungal Sinusitis

Objective:A nomogram model based on CT radiomics and clinical independent predictors was developed and validated to predict whether sinus inflammation is fungal infection.Methods:A retrospective analysis of 100 cases of fungal sinusitis and 100 cases of non-fungal chronic sinusitis confirmed by surgery and pathology in our hospital from January 2017 to November 2020.All patients received sinus CT examination before operation.Randomly divided into training set(n=140)and validation set(n=60).Selecting the thin CT images of each patient to extract the imaging features by 3D slicer image processing software,An radiomics signature was established after dimension reduction using univariate analysis,Least Absolute Shrinkage and Selection Operator(LASSO)for the training set.Evaluation of clinical data and CT image features,And using single factor and multiple factor Logistic regression to screen independent clinical predictors,And then using multivariable Logistic regression combined with radiomics signature to build prediction model and make nomogram,Evaluation of diagnostic efficacy of nomogram and radiomics signature in training set using ROC curves and validation by validation set.Based on all cases,the ROC curve was used to evaluate the diagnostic efficacy of nomogram model,radiomics signature and calcification factors.Results:A total of 1130 imaging features were extracted from CT images.After dimensionality reduction,6 features(including 1 first order feature and 5 texture features)were selected and then the image group labels were formed.Clinical independent predictors of age,lateralization and calcification were screened from clinical data and CT image features.A prediction model containing clinical independent factors and radiomics signature was constructed using multi-factor Logistic regression,and a line diagram was made,The area(AUC)under the ROC curve of fungal sinusitis was 0.961 and 0.957 in the training set and the validation set,assessment of AUC values of nomogram.AUC values of nomogram,radiomics signature,and calcification factors in all patients were 0.961,0.846,and 0.895,AUC value of radiomics nomogram is greater than calcification factor,difference was statistically significant(p<0.001).Conclusion:The Nomogram based on CT radiomics can be used as a quantitative tool to predict fungal sinusitis.