Accurate identification of malignant lung lesions is a prerequisite for rational clinical management to reduce morbidity and mortality of lung cancer. However, classification of lung nodules into malignant and benign cases is difficult as they show similar features in computer tomography and sometimes positron emission tomography imaging, making invasive tissue biopsies necessary. To address the challenges in evaluating indeterminate nodules, the authors investigate the molecular profiles of small extracellular vesicles (sEVs) in differentiating malignant and benign lung nodules via a liquid biopsy-based approach. Aiming to characterize phenotypes between malignant and benign groups, they develop a single-molecule-resolution-digital-sEV-counting-detection (DECODE) chip that interrogates three lung-cancer-associated sEV biomarkers and a generic sEV biomarker to create sEV molecular profiles. DECODE capturessEVs on a nanostructured pillar chip, confines individual sEVs, and profiles sEV biomarker expression through surface-enhanced Raman scattering barcodes. The author utilize DECODE to generate a digitally acquired sEV molecular profiles in a cohort of 33 people, including patients with malignant and benign lung nodules, and healthy individuals. Significantly, DECODE reveals sEV-specific molecular profiles that allow the separation of malignant from benign (area under the curve, AUC = 0.85), which is promising for non-invasive characterisation of lung nodules found in lung cancer screening and warrants further clinincal validaiton with larger cohorts.