High-Throughput Microscopy

vesselsUnderstanding the structure of tissue is crucial for biomedical research. Our laboratory develops new technologies to collect images of whole organs at subcellular resolution using three-dimensional optical imaging techniques, including: confocal microscopy, knife-edge scanning microscopy (KESM), and light-sheet microscopy.

Goals: We are particularly interested in three dimensional structure of tumor biopsies and brain tissue, and hope to develop novel imaging techniques that allow pathologists and researchers to quickly dissect and quantify three dimensional tissue samples in order to further understand the mechanisms of disease progression and tissue development.

Chemical Imaging

Chemical ImagingDetermining the distribution of chemical species in a tissue biopsy is critical for accurate diagnosis of disease. Current techniques use stains and dyes to label biopsies, however these methods are non-quantitative. Our laboratory analyzes vibrational spectroscopic images to identify the chemical composition of tumor biopsies, relying only on optical imaging and computation.

Goals: We are interested in improving clinical cancer diagnosis by developing imaging techniques that will replace existing stains and dyes with more robust, quantitative alternatives.


visualizationVisualizing large images of tissue microstructure poses a unique challenge, since biological samples are often densely packed. We develop computational methods for visualizing and analyzing terabyte-scale data sets containing complex cellular and subcellular components, such as tumor biopsies and brain tissue.

Goals: Our lab is developing open-source visualization methods that are designed to be easy to use and robust for massive data sets. Our goal is to make practical segmentation software available to biologists for analysis and visualization of large biomedical data sets.

Optical Modeling

modelingOne of the major bottlenecks of current optical imaging techniques is the computational complexity of solving inverse problems. In addition, fast forward models are crucial for understanding and improving complex imaging systems. Our lab uses GPU-based algorithms to develop fast forward models based on rigorous optical theory.

Goals: There is a wealth of information hidden within mid-infrared spectroscopic images that can theoretically be retrieved from mid-infrared spectroscopic imaging, if we are able to quickly and accurately model the scattering process. Our goal is to develop the fast forward models that form the basis for nonlinear inversion algorithms.