A minimally-invasive, optical strategy to detect and discriminate between inflammation and neoplasia could improve early cancer detection by reducing the number of false positive exams due to benign inflammation. This thesis describes research to optimize optical molecular contrast agents to observe architectural, metabolic, and biomolecular changes from inflammation and cancer in the gastrointestinal tract. My goal was to: 1) understand the limitations of autofluorescence imaging for cancer detection, 2) image exogenous fluorescent contrast agents specific to inflammation and neoplasia in rodent models, and 3) topically deliver a contrast agent cocktail in vivo in a mouse model. Wide field autofluorescence imaging of oral tissue utilizes endogenous tissue contrast to discriminate neoplastic from normal tissue; clinical studies of this technique show good sensitivity but poor specificity. I conducted a confocal microscopy study of 47 biopsies from 20 patients; results showed a similar decrease in autofluorescence in the stroma of inflamed and neoplastic tissue. This finding helps explain the low specificity of wide field autofluorescence imaging. Topically applied exogenous contrast agents could be used to improve discrimination between neoplasia and inflammation. I tested individual fluorescent contrast agents and contrast agent cocktails in chemically induced rodent models of inflammation and neoplasia. The first model used autofluorescence imaging with fluorescence imaging of proflavine to highlight cell nuclei and 2-NBDG to assess metabolic activity for oral cancer detection. A classification algorithm based on proflavine and 2-NBDG staining separated neoplastic from non-neoplastic areas on the tongue with 91% sensitivity and specificity. In the second model, a contrast agent cocktail composed of proflavine, a fluorescently labelled CD45-targeted antibody to identify inflammatory cells, and permeation enhancers was evaluated for topical in vivo delivery to image ulcerative colitis. The antibody identified the presence of inflammation and established topical delivery of antibody sized agents in vivo. These results provide evidence that topically applied contrast agent cocktails could improve discrimination between inflammation and neoplasia when endogenous contrast is insufficient. An optical-based strategy utilizing contrast agent cocktails to observe architectural, metabolic, and biomolecular changes associated with inflammation and cancer could improve early cancer detection by reducing the number of false positives from inflammation.