Chemical microscopy and image analysis for biomedical applications

Abstract

Chemical microscopy enables the identification and spatial analysis of chemical composition within a sample, offering powerful insights across a range of applications. This dissertation demonstrates the capability of chemical microscopy tools in diverse settings, from solid pharmaceutical materials to living organisms to human subjects in a clinical environment. The overarching principle guiding these studies is that, for simpler static samples, highly complex imaging techniques can be employed to extract rich chemical detail, whereas, for complex, dynamic biological systems, faster, more mobile tools, complemented by computational analysis, become essential. The progression of this work begins with the application of high-resolution coherent Raman microscopy to analyze well-defined, abiotic samples, demonstrating its ability to rapidly visualize and quantify chemical components in solid pharmaceutical tablets. Extending this approach to biological systems, the technique is then applied to measure the uptake of cryoprotective agents in Daphnia in correlation with toxicity, a critical step towards improving the development of cryopreservation protocols. These studies illustrate how chemically specific imaging can enhance material characterization and biological assessment, while also highlighting the increasing challenges of applying sophisticated tools to living systems. As the complexity of the sample increases, so do the demands on imaging methodology. In human subjects, where motion, heterogeneity, and physiological variability complicate data acquisition, a shift toward simpler, more mobile imaging modalities combined with computational approaches becomes essential. We use a cart-based multiphoton microscopy tool to characterize atopic dermatitis in humans in vivo, integrating imaging with spatial transcriptomics to improve our understanding of how RNA expression correlates with cellular level features. Further, the use of machine learning with these microscopy images enables disease classification and severity prediction, with the ultimate goal of replacing biopsies during longitudinal clinical studies of skin disease. Through this progression from controlled, static samples to complex, dynamic systems, this dissertation illustrates both the power and the limitations of chemical microscopy across diverse applications. Ultimately, this work contributes to the advancement of chemical microscopy as a versatile, noninvasive tool for biomedical and translational research.