Application value of CT radiomics and high-resolution MRI deep neural network in predicting lymph node metastasis of non-small cell lung cancer
10.3969/j.issn.1002-1671.2025.09.009
- VernacularTitle:CT放射组学与高分辨MRI深度神经网络联合预测非小细胞肺癌淋巴结转移的模型构建研究
- Author:
Xuewei FAN
1
;
Feng LI
1
;
Xingzhi SUN
1
;
Weixing LI
1
Author Information
1. 新乡市中心医院东区影像科 新乡医学院第四临床学院,河南 新乡 453000
- Publication Type:Journal Article
- Keywords:
CT radiomics;
high-resolution magnetic resonance imaging;
neural network;
non-small cell lung cancer;
lymph node metastasis
- From:
Journal of Practical Radiology
2025;41(9):1462-1466
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the application value of CT radiomics and high-resolution MRI deep neural network in predicting lymph node metastasis(LNM)of non-small cell lung cancer(NSCLC).Methods A total of 420 NSCLC patients were selected and randomly divided into training group(294 cases)and test group(126 cases)in a ratio of 7∶3.Lymph nodes were annotated using the MRIcroGL software,and radiomics features were extracted from thin-section CT images.Various feature screening methods were applied to optimize the features.A CT radiomics model was established using a support vector machine(SVM),and a high-res-olution MRI deep neural network model incorporating convolutional layers was constructed.Then the model performances were eval-uated,and the diagnostic efficacy of CT,MRI,and the combined models were compared using the receiver operating characteristic(ROC)curves.Results Least absolute shrinkage and selection operator(LASSO)regression identified 8 CT imaging features.The area under the curve(AUC)of the SVM model in the training and test groups were 0.755 and 0.765,and those of the high-resolu-tion MRI deep neural network model in the training and test groups were 0.884 and 0.899,respectively,demonstrating a high pre-dictive value.Conclusion The combined model of CT radiomics and high-resolution MRI deep neural network demonstrates superi-or performance in predicting LNM in NSCLC patients and holds significant clinical application value.
