PET/CT radiomics for predicting Ki-67 expression level of non-small cell lung carcinoma
10.13929/j.issn.1672-8475.2025.09.005
- VernacularTitle:基于PET/CT影像组学预测非小细胞肺癌Ki-67表达水平
- Author:
Xiaogang ZHANG
1
;
Yanjia ZHU
1
;
Xiaofeng LI
1
;
Wengui XU
1
Author Information
1. 天津医科大学肿瘤医院分子影像及核医学诊疗科,恶性肿瘤国家临床医学研究中心,天津市肿瘤防治重点实验室,天津市恶性肿瘤临床医学研究中心,天津 300060
- Publication Type:Journal Article
- Keywords:
lung neoplasms;
positron-emission tomography;
tomography,X-ray computed;
radiomics;
Ki-67
- From:
Chinese Journal of Interventional Imaging and Therapy
2025;22(9):579-582
- CountryChina
- Language:Chinese
-
Abstract:
Objective To observe the value of PET/CT radiomics for predicting Ki-67 expression level of non-small cell lung carcinoma(NSCLC).Methods 18F-FDG PET/CT data of 139 NSCLC patients were retrospectively analyzed.The patients were divided into high-expression group(≥40%,n=75)and low-expression group(＜40%,n=64)according to Ki-67 expression level of NSCLC.CT,PET and PET/CT data were divided into training set and test set at a ratio of 7∶3 and make the distribution of Ki-67 expression levels balanced between sets,respectively.CT,PET and PET/CT radiomics features of NSCLC were extracted,and the optimal radiomics features were screened,then random forest(RF),categorical boosting(CatBoost)and extreme gradient boosting(XGBoost)algorithms were used to construct models,respectively.Receiver operating characteristic curve was plotted,and the area under the curve(AUC)was calculated to screen the radiomics model with the highest efficacy for predicting Ki-67 expression level of NSCLC.Results RF models had the highest performance among radiomics models constructed based on CT,PET and PET/CT.The efficacy of RFCT,RFPET and RFPET/CT models for predicting Ki-67 expression level of NSCLC in test set increased sequentially,with AUC of 0.830,0.870 and 0.940,respectively(all P＜0.05),and RFPET/CT was the best radiomics model.Conclusion PET/CT radiomics could be used to effectively predict Ki-67 expression level of NSCLC,and RFPET/CT model had the best performance.
