Value of Pre-treatment CT Imaging in Predicting Lauren's Classification in Patients with Gastric Cancer
Objective To establish a radiomics model based on pre-treatment CT image features to predict the classification of Lauren gastric cancer.Methods Pre-treatment CT enhancement and clinical feature data of 167 gastric cancer patients with Lauren's classification confirmed by pathological biopsy were retrospectively analyzed,including 71 cases with intestinal type and 96 cases with diffuse/mixed type.It is randomly divided into training set and validation set according to 7:3 ratio.The region of interests were segmented on venous phase CT images by two senior abdominal radiologists,and the imaging features were extracted by 3D Slicer software.This study used least absolute shrinkage and selection operator(LASSO)regression algorithm to filter out the optimal feature combination,established radiomics labels,and applied logistic regression algorithm to build the model.The receiver operating characteristics(ROC)area under curve(AUC)was used to diagnosis efficiency and applied accuracy index to assessment model.Calibration curves were used to verify the matching of model prediction probabilities with actual results,and decision curves were used to evaluate the validity of clinical information.Results A total of 16 features were selected to predict intestinal type and diffuse/mixed type gastric cancer,and the radiomics model was established.In the training set,the accuracy,sensitivity and specificity of CT radiomics model were 79.7％,73.5％and 88.0％,respectively,and the AUC was 0.852(0.785-0.920).In the validation set,the accuracy,sensitivity,specificity,and AUC of the radiomics model were 73.3％,65.8％,80.5％,and 0.714(0.565-0.860).Conclusion The radiomics model based on pre-treatment CT imaging can predict enteric type and diffuse/mixed type gastric cancer,and provide objective basis for rational clinical treatment strategy.
万方数据
维普
