A Multi-modal Integrating Imaging and Cell-free DNA Methylation in Lung Cancer Early Detection (CMe-Lung): a Multi-center, Observational Study

This study is a multi-center, observational study aiming to develop an early screening multi-model for detecting lung cancer signals through the integration of imaging (non-contrast CT) and blood testing (cfDNA methylation).

This study aims to develop three modality -based models for lung cancer early screening, incuding a CT-based imaging model, a blood-based cfDNA methylation model, and a multimodal framework. First, 10,700 retrospective non-contrast CT scans will be collected to construct the CT imaging-based model. This model will extract multi- scale structural features through a hierarchical modeling framework, enabling the learning of interpretable spatial representations. Second, cfDNA methylation data from lung cancer patients and healthy individuals enrolled in the ProFound study (NCT06217900) wll be used to develop the blood-based model. Third, subjects with paired CT imaging and cfDNA methylation data from the ProFound study will be included to construct a multimodal model by integrating information across modalities. The performance of models based on different modalities will be systematically compared to evaluate their respective and complementary contributions to lung cancer early screening.