Predicting and Combining Model Value of 18F-PSMA PET/CT Imaging Omics for Clinically Significant Prostate Cancer
10.11969/j.issn.1673-548X.2025.05.029
- VernacularTitle:18F-PSMA PET/CT影像组学对临床显著前列腺癌的预测及组合模型价值研究
- Author:
Daiyun PENG
1
;
Yimeng ZHU
;
Jingyu FU
Author Information
1. 730030 兰州大学第二医院核医学科
- Publication Type:Journal Article
- Keywords:
Clinically significant prostate cancer;
18F-PSMA PET/CT;
Radiomics;
Logistic regression model
- From:
Journal of Medical Research
2025;54(5):166-173
- CountryChina
- Language:Chinese
-
Abstract:
Objective To investigate the value of a combined model incorporating clinical parameters,conventional metabolic pa-rameters of 18F-PSMA PET/CT,and radiomics features in the early prediction of clinically significant prostate cancer(csPCa).Methods A retrospective analysis was conducted on 124 patients who underwent 18 F-PSMA PET/CT and had complete pathological da-ta at the Second Hospital of Lanzhou University or Gansu Provincial People's Hospital.A total of 96 patients from Gansu Provincial Peo-ple's Hospital were randomly divided into a training set and an internal validation set at a 7∶3 ratio,while 28 patients from the Second Hospital of Lanzhou University served as an external validation set.In the training set,clinical parameters and radiomics features were i-dentified through Pearson correlation analysis and optimized using the least absolute shrinkage and selection operator(LASSO)combined with 10-fold cross-validation.Logistic regression was applied to develop separate clinical PET,radiomics,and combined models.Mod-el performance was assessed through receiver operating characteristic(ROC)curve analysis and calibration curves to evaluate predictive accuracy,decision curve analysis(DCA)to assess clinical utility.Results Three clinical and conventional PET parameters and five ra-diomics features were selected to construct the clinical PET model,radiomics model,and combined model.ROC analysis showed that all three models exhibited good predictive performance,with the combined model achieving the highest performance in the training,internal validation,and external validation sets(AUC were 0.973,0.933,and 0.813,respectively).Calibration curves and DCA indicated that the combined model demonstrated strong generalizability and predictive stability across all datasets.Conclusion The combined model in-corporating clinical parameters,conventional metabolic parameters of 18F-PSMA PET/CT,and radiomics features shows good value in the early prediction of csPCa.
