Multi-Dimensional Image Analysis Methods for Complex Systems

Badri Roysam, Professor, Electrical, Computer and Systems Engineering, Rensselaer Polytechnic Institute.

Optical microscopy is one of the "work horses" of biology. There has been an explosive growth in the sophistication of research microscopes. They can record cell and tissue structure and functional behavior along multiple dimensions (Fig. 1). For a start, they can capture structures in full three-dimensional detail without fragmenting the tissue as in biochemical assays, micro-arrays or flow cytometry. Time-lapse multi-photon and second-harmonic imaging can capture morphological dynamics with minimal photodamage. Multi-spectral excitation and sensing enables simultaneous observation of multiple structural and functional labels. Burgeoning fluorophore libraries now permit ever larger numbers of highly-specific structures and phenomena to be labeled simultaneously. Finally, there is a trend towards instruments that support multiple imaging modalities such as phase-contrast and DIC on the same platform.

Modern quantitative cell- and tissue-level studies (e.g., hypothesis testing, combinatorial methods, high-content assays, tissue engineering, systems modeling) generate vast quantities of data. There is a widespread need for automated software tools that are capable of translating this wealth of imaging data into quantitative insight.

This talk will describe highly automated, programmable, intuitive, and integrative multi-dimensional image analysis algorithms for modern microscopy. The dimensions of interest include 3-D space (x, y, z), time (t), spectra (molecular markers), and multiple imaging modalities. These tools are intended to support systems biology studies, with the added benefit of rich spatial information derived from modern microscopes. The generality of these tools is evidenced by the diversity of applications that can be handled in a common framework.