Prediction of p53 Mutation in Endometrial Carcinoma Based on Radiomics Nomogram of Intratumoral and Peritumoral MRI
10.3969/j.issn.1005-5185.2025.05.018
- VernacularTitle:基于瘤内和瘤周MRI影像组学列线图模型预测子宫内膜癌p53突变状态
- Author:
Hua ZHANG
1
;
Yunze YANG
;
Junhong HE
;
Mengtong LIU
;
Mingjie WANG
Author Information
1. 承德医学院研究生学院,河北 承德 067000;保定市第一中心医院医学影像科,河北 保定 071000
- Publication Type:Journal Article
- Keywords:
Endometrial neoplasms;
Genes,p53;
Mutation;
Magnetic resonance imaging;
Radiomics;
Machine learning;
Intratumoral;
Peritumoral;
Forecasting
- From:
Chinese Journal of Medical Imaging
2025;33(5):553-559
- CountryChina
- Language:Chinese
-
Abstract:
Purpose To investigate the value of radiomics nomogram for the prediction of p53 abnormal in patient with endometrial carcinoma based on intratumoral and peritumoral MRI.Materials and Methods A total of 145 female patients were pathologically confirmed endometrial carcinoma who underwent pelvic MRI before treatment in Baoding First Central Hospital from January 2020 to April 2024,including 96 patients with p53 wild type and 49 with p53 abnormal.Radiomics features were extracted from both intratumoral and peritumoral regions(2 mm)in diffusion weighted imaging and equilibrium phase of dynamic contrast enhanced MRI,which were selected using least absolute shrinkage and selection operator.Three machine learning algorithms of random forest,K-nearest neighbors and extra trees were conducted to develop the intratumoral,peritumoral and intratumoral combined peritumoral radiomics models.Multivariate Logistic regression was used to constitute the clinical model and nomogram.The performance of these models was evaluated using receiver operating characteristic curve,decision curve analysis and calibration curve.Results The K-nearest neighbors model of the intratumoral combined peritumoral regions performed the best in all radiomics models,the area under the curve were 0.921 and 0.773 in the training cohorts and test cohorts.The radiomics nomogram,which was composed of age,apparent diffusion coefficient and radiomics signatures,achieved the best performance with area under the curve of 0.970 and 0.777 in the training cohorts and test cohorts,respectively.Calibration curve analysis and decision curve analysis demonstrated favorable calibration and clinical utility of the nomogram model.Conclusion The nomogram based on intratumoral and peritumoral MRI radiomics yields a favorable diagnostic value for predicting p53 abnormal in patient with endometrial carcinoma.
