Lung CT Radiomics Dataset of Clinically Healthy Adults

This dataset contains structured lung CT radiomics features from 100 clinically healthy adults acquired on a single scanner using a homogeneous routine chest CT protocol. Lungs were segmented with TotalSegmentator to generate anatomically consistent ROIs: individual lobes, left lung, right lung, and both lungs combined. Radiomics features were extracted with PyRadiomics using a fixed, fully documented configuration under a 2×2 design: two attenuation ranges (raw: −1000 to 200 HU; parenchyma-focused: −950 to 0 HU) and two anatomical mask modes (including intrapulmonary vessels and airways vs parenchyma-only). The dataset provides feature tables with more than 100 first-order, shape, and texture descriptors per ROI, accompanied by anonymized demographics, a detailed data dictionary, a README file, and the Python extraction script with the full PyRadiomics settings. This resource is intended for methodological benchmarking, exploring feature stability across windows and mask definitions, and constructing normative reference ranges for studies of diffuse and region-predominant pulmonary diseases. Raw DICOM/NIfTI images are not distributed due to legal and ethical constraints.