Artificial Intelligence-based Model for the Prediction of Occult Lymph Node Metastasis and Improvement of Clinical Decision-making in Non-small Cell Lung Cancer

Artificial Intelligence-based Model for the Prediction of Occult Lymph Node Metastasis and Improvement of Clinical Decision-making in Non-small Cell Lung Cancer: A Multicenter, Prospective, Observational Study

This nationwide, multicenter observational study aims to develop and validate a multimodal artificial intelligence (AI) model for detecting occult lymph node metastasis in early-stage non-small cell lung cancer (NSCLC) patients. Despite advances in lymph node staging, 12.9%-39.3% of occult nodal metastasis cases remain undetected preoperatively, affecting treatment decisions. This study will use deep learning to extract imaging features of occult metastasis and combine them with clinical data to build an AI model for risk prediction. This study will provide insights into the feasibility of AI-driven detection of occult metastasis, supporting clinical decision-making and potentially revealing underlying biological mechanisms of lymph node metastasis in NSCLC.

Inclusion Criteria:

• Pathologically confirmed non-small cell lung cancer;
• Clinical stage I (AJCC, 8th edition, 2017);
• Age≥18 years old;
• KPS score≥70;
• Patients who have undergone primary NSCLC radical surgery or SBRT treatment;
• Complete systemic lesion imaging assessment before primary NSCLC radical surgery or SBRT treatment (Note: Tumor size ≥ 3 cm or centrally located tumor requires PET/CT and/or invasive mediastinal staging);
• Patients willing to cooperate with the follow-up after primary NSCLC radical surgery;
• informed consent of the patient.

Exclusion Criteria:

• Poor quality of computed tomography imaging;
• Baseline imaging shows pure ground-glass nodules (GGO);
• Uncontrolled epilepsy, central nervous system disease, or history of mental disorders, judged by the researcher to potentially interfere with the signing of the informed consent form or affect patient compliance.;
• Loss to follow-up.