By combining highly multiplexed imaging microscopy with methods based on computational and systems biology, mathematics and physics, and image processing and machine learning, we aim to  deconvolve the complexity of the tumor microenvironments of solid cancers. Specifically, we are interested in studying, understanding and characterizing how cells of different lineages, and varying signaling and signal processing capabilities come together within the spatial context of the tumor microenvironment to give rise to various malignant phenotypes of neoplastic transformation, cancer progression, recurrence, and response to therapy.
Many patients, such as those with chronic inflammation or a hereditary cancer syndrome, are at a higher-risk of developing cancer. Despite this increased risk, most will not develop cancer. Therefore, an open question is how can we triage surveillence of such at-risk patients that reduces burden on clinics and hospitals, and more importantly, on patients themselves, while being concordant with improved outcomes for them. We have been developing a Fourier-domain Optical Coherence Tomography derived optical imaging approach that is able to capture malignant alterations in the nuclear architecture of epithelial cells. Importantly, it does so with nanoscale sensitivity, before these alterations manifest histologically. Currently, we are studying this ability both from a translationally relevant perspective and at a more basic science level.
The advent of rapid whole slide digital imaging of morphology and biomarker expression in patient tissue sections has created new opportunities to develop multi-scale systems-pathology methods that provide improved and quantitative pathobiological insights into patient diagnostics, prognostics and therapeutics. We are developing such methods. We are also interested in integrating them with omics based bionformatics approaches and analyses. Given that cancer growth lies on a continnuum, with genetic, epigenetic, transcriptomic and preoteomic phenotypes supervening upon each other, such integration will allow capturing of tumor complexity in individual patients in a translationally-relevant manner.
