A Hierarchical Multi-modal AI Framework for Pathological and Genetic Subtyping of Lung Cancer Based on PET/CT Imaging

What this Trial Is About

Researchers are investigating a multi-modal artificial intelligence (AI) framework designed to improve the classification and genetic subtyping of lung cancer using PET/CT imaging and clinical information such as age, gender, smoking history, family cancer history, and tumor biomarkers. This observational study aims to accurately distinguish between small cell lung cancer and non-small cell lung cancer (NSCLC), and then further divide NSCLC into subtypes including adenocarcinoma and squamous cell carcinoma. Additionally, the study evaluates the prediction of the EGFR gene mutation status, especially common in lung adenocarcinoma patients. All participants undergo a pre-treatment 18F-FDG PET/CT scan. The study cohort is divided into training, validation, test, and prospective groups for analysis. The AI framework is developed and tested using retrospective data, followed by prospective validation to assess the model's accuracy in pathological and genetic subtyping. This structured approach aims to facilitate a precise and hierarchical stratification process for lung cancer patients. Participants will provide clinical information and have their PET/CT images analyzed as part of the study. Researchers will measure the accuracy of lung cancer subtype differentiation and EGFR mutation prediction over a one-year period. The study involves no drug interventions, focusing on imaging data, clinical records, and AI model development. The total participation time varies, with assessments primarily based on imaging and clinical data collected before treatment begins.

A Hierarchical Multi-modal AI Framework for Pathological and Genetic Subtyping of Lung Cancer Based on PET/CT Imaging