Research on Early Screening and Diagnosis of Pulmonary Nodules Based on Novel Non-invasive Technologies.

This is a prospective observational study designed to address the clinical challenge posed by the high false-positive rate associated with CT imaging in early lung cancer screening. The primary objective is to develop a multi-omics technology for early lung cancer screening, leveraging \*\*exhaled breath metabolomics, plasma metabolomics, radiomics, and liquid biopsy. Based on large-sample detection data, the study aims to construct a \*\*multi-dimensional, sequential decision-making system\*\*. This system utilises the high accessibility of metabolomics for primary screening, combined with radiomics and ctDNA technologies for subsequent \*\*differentiation and definitive diagnosis. The research plans to prospectively enrol 300 patients with non-small cell lung cancer, along with corresponding subjects with benign nodules and healthy controls. By optimising the model using machine learning and deep learning algorithms (such as SVM, HRNet, and PAResNet), the ultimate goal is to establish a novel lung cancer early screening system characterised by \*\*high sensitivity, high accuracy, and high accessibility\*\*, enabling the precise differentiation and screening of healthy individuals, benign pulmonary nodules, and early-stage lung cancer.