Application research of radiomics based on enhanced CT venous phase for preoperatively predicting poorly differentiated esophageal squamous cell carcinoma
10.3969/j.issn.1002-1671.2025.09.012
- VernacularTitle:基于增强CT静脉期的放射组学在术前预测低分化食管鳞状细胞癌的应用研究
- Author:
Meng LIU
1
;
Zeqiang GAO
;
Chunyue YAN
;
Weili LONG
;
Ming YANG
;
Fei WANG
Author Information
1. 泸州市人民医院医学影像科,四川 泸州 646000
- Publication Type:Journal Article
- Keywords:
esophageal squamous cell carcinoma;
pathologi-cal grade;
computed tomography;
radiomics
- From:
Journal of Practical Radiology
2025;41(9):1477-1481
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore a nomogram of intratumor and peritumor radiomics based on enhanced CT venous phase to pre-operatively predict the pathological grade of poorly differentiated esophageal squamous cell carcinoma(ESCC).Methods A retro-spective selection was made of 266 ESCC patients confirmed by pathology(76 cases of poorly differentiated;190 cases of non-poorly differentiated),and all patients were randomly divided into training set(n=186),validation set(n=80),and full data set(n=266).Tumors were segmented on the enhanced CT venous phase to create three-dimensional region of interest(ROI)of intratumor,peritu-mor 0.3 cm,and intratumor+peritumor 0.3 cm.A total of 2 553 radiomics features were extracted.After feature dimensionality reduc-tion,XGboost machine learning algorithm was utilized to rank the top fifteen features.Stepwise forward multiple logistic regression was employed to identify the most significant features.The radiomics scores of the intratumor,peritumor 0.3 cm,and intratumor+peritu-mor 0.3 cm were calculated.The diagnostic efficacy of the model was evaluated using the area under the curve(AUC)of the receiver operating characteristic(ROC)curve,calibration curve and decision curve analysis(DCA).Results The nomogram constructed based on radiomics scores of intratumor,peritumor 0.3 cm,intratumor+peritumor 0.3 cm in the training set for preoperative prediction of poorly differentiated ESCC had an AUC of 0.899[95％confidence interval(CI)0.846-0.938],and it was well validated in the vali-dation set(AUC 0.869,95％CI 0.775-0.934)and the full data set(AUC 0.889,95％CI 0.845-0.924).Additionally,calibration curves and DCA indicated that the nomogram achieved good calibration ability in the three cohorts and offered greater clinical net benefit.Conclusion The nomogram based on enhanced CT venous phase intratumor and peritumor radiomics achieves a high and stable diagnostic efficacy for preoperatively predicting poorly differentiated ESCC,which may help with individualized surgical selec-tion and management before surgery.
