Splenic CT radiomics nomogram predicting the risk of upper gastrointestinal hemorrhage in cirrhosis

Abstract

ObjectiveTo explore the application value of splenic CT radiomics nomogram in predicting the risk of upper gastrointestinal hemorrhage. MethodsFrom October 2019 to September 2021, a total of 110 patients with cirrhotic portal hypertension were retrospectively analyzed. All patients underwent abdominal CT plain scan and gastroscopy. According to the results of the gastroscopy, the patients were divided into a low and medium-risk bleeding group (n = 45) and a high-risk bleeding group (n = 65). The samples were randomly divided into a training set and a validation set according to the ratio of 7:3. A total of 1647 unscanned splenic radiomics features were extracted, and 12 radiomics features closely related to gastrointestinal hemorrhage were analyzed and screened by the national variance threshold method and LASSO algorithm. The clinical bleeding risk factors PLT obtained from multiple regression analysis were modeled, and the radiomics model was combined to construct a nomogram predicting the risk of upper gastrointestinal hemorrhage. ResultsThe area under the curve values of the combined radiomics model and clinical high-risk bleeding factor model in the training and test sets were 0.947 and 0.924, which were higher than those of the radiomics model alone and the clinical high-risk bleeding factor PLT model. ConclusionThe splenic CT radiomics nomogram combined with clinical high-risk factors has high consistency in predicting upper gastrointestinal hemorrhage with digestive endoscopy.