The Value Of CT Radiomics In Predicting The Short-term Efficacy Of TACE In Patients With Colorectal Liver Metastases

Objective: To explore the value of CT radiomics model in predicting the short-term efficacy of transcatheter arterial chemoembolization(TACE)in patients with colorectal cancer liver metastasis(CRCLM).Methods: Retrospective analysis from February 2014 to May 2022,Clinical and imaging data of 126 patients with colorectal cancer confirmed by endoscopy or histopathology and liver metastasis confirmed by pathology or imaging examination in Yijishan Hospital of Wannan Medical College.All patients underwent plain and enhanced abdominal CT scans within 2 weeks before TACE treatment and 1 to 3 months after treatment.Then,according to modified response Evaluation Criteria In Solid Tumors(m RECIST),the patients were divided into good response group(n = 74)and poor response group(n = 52).The patients were divided into a training set and a validation set(89 cases and 37 cases,respectively)with a ratio of 7∶3 by using computer randomized analysis method.Using ITK-SNAP software,the region of interest(ROI)was manually drawn on all slices of liver metastases on plain scan,arterial phase,venous phase and delayed phase images,respectively.The volume of the lesions was obtained by 3D fusion,and then imported into FAE software to extract the internal texture information of each lesion.The minimum redundancy maximum correlation(m RMR)and minimum absolute convergence and selection operator(LASSO)were used to screen the radiomics features of plain scan,arterial stage,venous stage,delayed stage and multiple sequences(plain scan + arterial stage + venous stage +delayed stage),respectively,for the prediction of short-term efficacy after TACE surgery,establish the radiomics model and calculate the rad-score for each patient.The reliability of the radiomics model was determined by 100-fold leave-P cross-validation.Logistic regression analysis was used to establish a conventional model based on the variables with statistically significant differences between groups in clinical and imaging data,and then the radiomics model with the highest prediction efficiency was combined to establish a comprehensive diagnostic model,a nomogram was drawn to analyze the predicted probabilities.The receiver operating characteristic(ROC)curve was used to evaluate the predictive efficacy of the model,and the decision curve analysis(DCA)was used to evaluate the net clinical benefit of the model.Results: The area under the curve(AUC)of plain scan,arterial phase,venous phase,delayed phase and multi-sequence imaging model in the training group and validation group were 0.79 and0.68,0.78 and 0.76,0.78 and 0.80,0.84 and 0.79,0.75 and 0.69,respectively.The delayed period image radiomics model has better predictive efficiency in both the training set and the verification set.The AUC of conventional model and comprehensive diagnostic model in training group and validation group were 0.73 and 0.71,0.87 and 0.86,respectively.The comprehensive diagnostic model was significantly in predicting the short-term efficacy of CRCLM patients after TACE.The delayed period rad-score and CA19-9 were independent predict factors(OR = 1.79,P<0.001 and OR =1.00,P = 0.011).The DCA showed that both the comprehensive diagnostic model and the delayed period image radiomics model had better net benefits.Conclusions :1.In terms of clinical data and CT image features,tumor size and CA19-9 level are important factors for the short-term efficacy after TACE in CRCLM patients.2.In the established radiomics models of plain scan,arterial phase,venous phase,delayed phase and multi-sequence,the delayed phase radiomics model had high prediction efficiency in both the training set and the validation set.3.The comprehensive diagnostic model based on clinical data,CT imaging features and rad-score of delayed phase imaging model has a high value in predicting the short-term efficacy of TACE in CRCLM patients,which is significantly superior to the clinical model.