e17529 Background: A major advance against lung cancer was made with the announcement of results from the National Lung Cancer Screening Trial. A 20% reduction in mortality was shown in a screened population using low dose CT (LDCT) compared to screened chest x-rays. Approximately 95% of the positive findings of LDCT, however, turned out not to be cancer. We propose to use 3D morphometric analysis of cells from enriched sputum to distinguish benign from malignant nodules found using LDCT and to identify cancers that were not identified with LDCT. Methods: Sputum from 30 patients with non-small cell lung cancer were studied. Fluorescence activated cell sorting was used to enrich sputum for epithelial cells. The enriched cells, stained with hematoxylin, were automatically imaged using VisionGate’s 3D cell imaging system (Cell-CT) which unambiguously renders individual cells with isometric, sub-micron resolution. Image libraries of cells were created and 530 morphology features were computed for each cell. Abnormal (moderate and severe dysplasia and cancer) status for each cell was determined by a cytotechnologist and added to the feature set. Cross-validated classifiers were created that combine a sub-set of the features into a single score that best correlated with the binary cell diagnoses. Results: Study of enriched sputum from cancer patients using the Cell-CT shows cancer cells in at least 75% of the cases with cancer cells present in raw sputum. Classifier cell specificity exceeding 98% and a 70% cancer cell detection rate were measured. A 70% cell detection rate translates into a 90% detection rate for the specimen since multiple abnormal cells are present in sputum. Overall detection rate for cancer cell positive sputum is: 68%, computed by finding the product of the rate of cancer cell positive enriched sputum (0.75) with the rate of detection by the classifier (0.90). Conclusions: Automated, 3D morphometric analysis of cells from sputum provides high sensitivity and specificity detection of lung cancer. Adjunctive use with LDCT will aid in determination of the malignancy status of indeterminate pulmonary nodules.