Bioimaging Science and Technology Group
Beckman Institute for Advanced Science and Technology
University of Illinois at Urbana-Champaign
Urbana, IL 61801 USA
My research focuses on merging high-performance computing with new biomedical imaging techniques. The goal of my work is to create databases that provide unprecedented amounts of data at the cellular level for research on tissue anatomy and the mechanisms of disease. My experimental work focuses on developing fast optical imaging methods, while my computational work uses high-performance computing to process and understand large data sets.
Chemical Imaging Advances in hyperspectral imaging allow the collection of spatially-resolved spectra that can be used to characterize the chemical composition of biological samples. I am working with the Chemical Imaging and Structures Laboratory to develop methods for characterizing and classifying hyperspectral images. Our goal is to provide quantitative alternatives to the chemical staining techniques currently used in clinical histology.Read more →
High-Throughput Microscopy Future research in biomedicine relies heavily on acquiring detailed data sets describing the structure, chemical compostion, and function of tissue at the sub-micrometer scale. My research focuses on developing imaging methods and computational tools for collecting massive three-dimensional data sets describing tissue microstructure on the whole-organ scale.Read more →
Visualization The development of new biomedical imaging techniques poses unique and interesting problems in the field visualization, where algorithms must be developed to help researchers interpret data. My visualization research focuses on new methods for data mining and interpreting images acquired using new techniques, such as Serial Electron Microscopy and Spectroscopic Imaging.Read more →
- [2013-5-3] Our paper on coherent super-resolution imaging was accepted to Optics Express.
- [2013-4-25] Our poster on High Performance Computing won 3rd place at the David Kuck Poster Competition.
- [2013-4-10] Our user interface code for the Mackowski and Mishchenko parallel T-Matrix solver is now publicly available
- [2013-4-2] Our paper on chemical visualization was accepted to the BMC Bioinformatics issue on Cancer Bioinformatics
- I am currently on the organizing committee for the 2014 Beckman Visualization Workshop
- [2013-1-18] Our paper on inverse modeling of Mie scattering has been accepted to Applied Spectroscopy
CSE Annual Meeting
- Poster Competition: Using High-Performance Computing to Augment Biomedical Imaging Systems
Chemical Imaging and Structures Laboratory
Presenter: D. Mayerich
Visualization in Serial EM D. Mayerich and J.C. HartRead more →
IEEE Symposium on Biological Data Visualization
Recent advances in electron microscopy allow the collection of three-dimensional data. These imaging techniques have tremendous implications in neuroscience, since they provide sufficient resolution to potentially image and reconstruct neuronal connectivity in the brain. However, these data sets are tremendously feature-rich and difficult to visualize. In this paper, we present methods that make volume visualization of these data sets possible by applying transfer functions based on local variance.
Whole-Brain Vascular Imaging D. Mayerich, J. Kwon, C. Sung, L. Abbott, J. Keyser, Y. ChoeRead more →
Biomedical Optics Express
Microvascular networks are complex structures that span large volumes of tissue and are extremely important for regulating cell behavior. However, most methods capable of whole-organ imaging, such as MRI and CT, do not provide sufficient resolution to reconstruct microvessels. Microscopy provides sufficient resolution, but limits researchers to imaging small subsets of the network. In this paper, we demonstrate that Knife-Edge Scanning Microscopy can be used to image large tissue volumes, such as the whole mouse brain, in less than two days. This results in terabytes of imagery and provides sufficient (sub-micron) resolution to reconstruct the entire microvascular network in three-dimensions.
Quantifying Fiber Segmentation D. Mayerich, J. Taylor, C. Bjornsson, B. RoysamRead more →
Advances in high-throughput microscopy allow researchers to image large tissue volumes at sub-micrometer resolution. This is particularly useful for studying complex networks of fibers, like those formed by neurons and microvessels. However, these structures are difficult to reconstruct, and the development of automated algorithms is an active area of research. One of the reasons for this difficulty is the lack of robust metrics that can be used to validate segmentation results. In this paper, we develop a method for comparing the structure and connectivity of two networks, and propose its use as a quantitative validation metric.