Background: Quantitative measurements of emphysema extracted from CT scan data are a useful tool in COPD phenotyping, but existing quantitative phenotypes such as low attenuation area (LAA) % at -950HU have not found compelling genetic associations. Texture-based patterns of lung density in CT images may better characterize the amount and type of emphysema. Using a local histogram based emphysema classification method, we quantified the relative amounts of various types of emphysematous and normal lung tissue in CT scans from subjects in the COPDGene Study, and we performed a genome-wide association study on these quantitative phenotypes to identify genetic determinants of emphysema.