Radiomics Based on Enhanced CT in Predicting the Risk Classification of Gastric Stromal Tumors
10.3969/j.issn.1005-5185.2024.09.008
- VernacularTitle:基于增强CT影像组学预测胃肠道间质瘤危险度分级
- Author:
Juan PENG
1
,
2
;
Xianli LUO
;
Ruxue FAN
;
Hong YU
;
Bangguo LI
Author Information
1. 遵义医科大学附属医院放射科,贵州遵义 563000
2. 贵州茅台医院放射科,贵州仁怀 564500
- Keywords:
Gastrointestinal stromal tumors;
Tomography,X-ray computed;
Radiomics;
Risk assessment;
Pathology,surgical
- From:
Chinese Journal of Medical Imaging
2024;32(9):908-913
- CountryChina
- Language:Chinese
-
Abstract:
Purpose To explore the value of predicting risk classification of gastrointestinal stromal tumors(GIST)based on the model established by dural-phase enhanced CT based radiomics.Materials and Methods Totally 200 patients with pathologically confirmed GIST from October 2017 to July 2023 in the Affiliated Hospital of Zunyi Medical University were enrolled,including 69 cases with low-risk group(very low-risk,low-risk)and 131 with high-risk group(medium-risk,high-risk).All patients were randomly divided into training set(n=139)and validation set(n=61)at the ratio of 7∶3.Univariate and multivariate Logistic regression analysis were used on clinical data and CT sings in the training set to obtain clinical-CT features for predicting the risk grade of GIST,and clinical-CT models were constructed.The radiomics features were extracted and screened from the three data sets of enhanced CT arterial phase,venous phase and arterial+venous phase,and the radiomics model was constructed to obtain the optimal radiomics features,respectively.The optimal radiomics features were obtained and combined with the clinical-CT features,a combination model was constructed and the normogram was drawn.The predictive efficiency of these models was evaluated by area under the curve(AUC).Results Tumor diameter was an independent predictor of GIST risk classification(OR=1.070,P＜0.001).The AUC of the combination model,model arterial+venous phase radiomics and model clinical-CT in the training set were 0.948,0.896 and 0.873,respectively;those in the validation set were 0.886,0.825 and 0.870,respectively.The AUC of the above three models showed statistical difference(Z=-3.167,-2.316,P＜0.05).Conclusion The radiomics features based on enhanced CT have good value in predicting risk classification of GIST.Compared with model clinical-CT and model radiomics,the combination model is the most effective in predicting the risk classification of GIST.
