Differentiation of vascular and non-vascular skin spectral signatures using in vivo hyperspectral radiometric imaging: Implications for monitoring angiogenesis

Abstract

Molecular imaging techniques can detect and monitor characteristics of the tumor microenvironment, such as angiogenesis, hypoxia, metabolism, and apoptosis that may better correlate with response to cancer therapy and may provide information in real-time. We investigated the use of a novel, spatially discrete, hyperspectral, multi-fiber optical system to characterize selected regions of skin in living mice. We determined the reproducibility and robustness of the spectral signatures derived from comparable regions of interest. Additionally, we characterized spectral differences in vascular and non-vascular fields to determine their potential use in monitoring angiogenesis. The macroscopic Prism and Reflectance Imaging Spectroscopy System (MACRO-PARISS) was calibrated against a National Institute for Standards and Technology (NIST)-certified lamp, allowing for reproducible spectra with any instrument similarly calibrated. Spectra were classified using a linearity-independent algorithm over a wavelength range of 450-920 nm. Classified spectra were integrated into a spectral library and subsequent acquisitions were correlated with the library set to a minimum correlation coefficient (MCC) of 99%. The results indicated that similar regions of interest with respect to vascularity consistently generated a unique spectral signature. As the field of view (FOV) moved from vascular to non-vascular areas, the acquired spectra changed in a step-wise and predictable fashion. Additionally, vascular fields that were deprived of their blood supply subsequently generated a non-vascular spectral signature. This work has implications for monitoring of various physiologic or pathological processes including tumor angiogenesis and the therapeutic effects of anti-vascular agents.