Impact of radiomics features from chest imaging on lung cancer diagnosis and treatment response prediction

HOU Xin; XU Jiang; DIAO Zhushuai; LI Huijian; ZHOU Yunpeng

doi:10.20258/j.cnki.1006-9011.2026.01.014

Abstract：Objective To investigate the value of radiomics features extracted from chest CT imaging in the diagnosis of lung cancer and the prediction of treatment response.Methods A total of 120 patients with pulmonary nodules admitted to our hos-pital were enrolled.All patients underwent contrast-enhanced chest CT scans and were pathologically confirmed.Using the Pyradiomics platform,285 radiomics features were extracted,and the least absolute shrinkage and selection operator(LASSO)algorithm was applied for feature selection to construct a radiomics model.With pathological results as the gold stan-dard,the diagnostic efficacy of the model for lung cancer was evaluated.Additionally,clinical follow-up data were analyzed to assess the correlation between radiomics features and progression-free survival(PFS)and overall survival(OS)in these pa-tients.Results The radiomics model demonstrated robust performance in both the training and validation sets.In the validation set,the AUC reached 0.912,with a sensitivity of 88.33%,specificity of 85.90%,and accuracy of 87.00%.Survival analysis re-vealed that patients in the radiomics high-risk group had significantly shorter progression-free survival(PFS)(13.46 months vs.18.79 months,P<0.01)and overall survival(OS)(21.57 months vs.33.12 months,P<0.01)compared to the low-risk group(Figure 4).Multivariate analysis identified original_glcm_ClusterProminence(HR=1.321,95%CI:1.082-1.614,P=0.009)and SurfaceVolumeRatio(HR=0.703,95%CI:0.542-0.901,P=0.007)as independent predictors of PFS and OS,respectively.Conclusion Radiomics features can effectively assist in lung cancer diagnosis and provide quantitative evidence for treatment response and prognosis evaluation.