Towards automated early cancer detection: Non-invasive, fluorescence-based approaches for quantitative assessment of cells and tissue to identify pre-cancers

Cancer is the second leading cause of death globally, second only to heart disease. As in many diseases, patient survival is directly related to how early lesions are detected. Using conventional screening methods, the early changes associated with cancer, which occur on the microscopic scale, can easily go overlooked. Due to the inherent drawbacks of conventional techniques we present non-invasive, optically based methods to acquire high resolution images from live samples and assess cellular function associated with the onset of disease. Specifically, we acquired fluorescence images from NADH and FAD to quantify morphology and metabolic activity. We first conducted studies to monitor monolayers of keratinocytes in response to apoptosis which has been shown to be disrupted during cancer progression. We found that as keratinocytes undergo apoptosis there are populations of mitochondria that exhibit a higher metabolic activity that become progressively confined to a gradually smaller perinuclear region. To further assess the changes associated with early cancer growth we developed automated methods to rapidly quantify fluorescence images and extract morphological and metabolic information from life tissue. In this study, we simultaneously quantified mitochondrial organization, metabolic activity, nuclear size distribution, and the localization of the structural protein keratin, to differentiate between normal and pre-cancerous engineered tissues. We found the degree mitochondrial organization, as determined from the fractal derived Hurst parameter, was well correlated to level of cellular differentiation. We also found that the metabolic activity in the pre-cancerous cells was greater and more consistent throughout tissue depths in comparison to normal tissue. Keratin localization, also quantified from the fluorescence images, we found it to be confined to the uppermost layers of normal tissue while it was more evenly distributed in the precancerous tissues. To allow for evaluation of the early cancerous changes in vivo, we developed video-rate confocal reflectance/multi-photon fluorescence microscope as a clinical prototype. This device was specifically designed to rapidly acquire and assess non-invasively acquire fluorescence images using the automated methods we have developed. We have demonstrated the ability of this microscope to simultaneously acquire fluorescence, confocal reflectance, and second-harmonic generation images as well as assess blood flow in vivo.