Machine learning models based on CT radiomics can effectively predict invasiveness of pulmonary pure ground-glass nodules
ZHANG Na
DU Jing
GUO Ziquan
WU Zhifeng
Abstract：Objective To evaluate the value of a radiomics-based machine learning model in predicting the invasiveness of pulmonary pure ground-glass nodules(pGGNs).Methods A retrospective cohort study was conducted on 208 pGGNs identified in our department from August 2022 to August 2024.Based on pathological results,the nodules were classified into non-invasive and invasive groups.CT characteristics of the nodules were recorded,and radiomics features were extracted from CT images.Optimal radiomics features were selected to construct a predictive model.ROC curves were plotted and the area under the curve(AUC)was calculated.The diagnostic performance of the radiomics model was compared with that of radiologists alone and radiologists assisted by the radiomics model.Results After dimensionality reduction,six optimal features were selected to build a logistic regression model.In the training set,the model achieved an AUC,sensitivity,and specificity of 0.786,0.771 and 0.875,respectively,while in the validation set,these values were 0.776,0.735 and 0.859.The radiomics model outperformed radiologists in diagnostic accuracy and enhanced radiologists'diagnostic performance when used in combination.Conclusion The CT radiomics-based machine learning model demonstrates high predictive efficacy for determining the invasiveness of pulmonary pGGNs and provides valuable guidance for clinical decision-making.
Keywords：pure ground-glass nodulesradiomicsmachine learningtomographyX-ray computed tomography
Publication Date：2025-05-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 614-619 )