Main menu >> Electrical and Computer Engineering >> Digital Signal and Imaging Processing
Electrical and Computer Engineering
Digital Signal and Imaging Processing
- Unwrapping Phase Images
- Brain Image Segmentation by Integrated Multiscale Analysis and Shape Deformation
- Model-based Tomographic Imaging Methods
- Multisensor Information Fusion
- Statistical image reconstruction:
^ Unwrapping Phase Images
R. Koetter,* D. C. Muson,* Z. P. Liang*
National Science Foundation, CCR 01-05719
The primary goal of the project is to develop optimal algorithms for the long-standing problem of unwrapping phase images from various imaging modalities such SAR and MRI. Probabilistic inference algorithms will be developed and tested using SAR and MRI as testbeds. Prof. Liang is responsible for phase unwrapping of MRI data.
^ Brain Image Segmentation by Integrated Multiscale Analysis and Shape Deformation
Z. P. Liang,* S. Wang
NEC Research Lab, University of Illinois Research Board
Brain image segmentation is an important and challenging engineering problem confronting brain mapping. By accurately segmenting gray-scale brain images into various brain structures, we will be able to effectively visualize three-dimensional brain structures and carry out meaningful neuromorphometric studies. The long-term goal of this project is to develop and implement a unified processing software platform to effectively support various information processing tasks in neuroimaging or brain mapping. The specific aim of the project is to capitalize on our recent, novel work on graph-based multiscale image analysis and shape deformation to produce an efficient, accurate and reliable algorithm for identifying brain structures from MR images. We expect to accomplish three specific tasks during the project period: complete the development of a novel graph-theoretic algorithm for multiscale analysis of MR brain images; further develop, perfect, and validate a topology-preserving shape deformation algorithm so that prior shape information of brain structures can be incorporated into the image segmentation process effectively; and integrate multiscale analysis with shape deformation for accurate segmentation of brain images and develop a prototype software system to facilitate the application of the developed algorithms for practical applications in brain mapping.
^ Model-based Tomographic Imaging Methods
Z. P. Liang,* J. Ji, Y. Bresler*
National Institutes of Health, R21 HL62336
The mathematical basis of tomographic imaging is conventionally rooted in the well-established Fourier or radon transform theories, so that image quality is mainly dependent on how the data space is sampled. In practice, physical and temporal constraints often prevent a sufficient coverage of the data space, resulting in various image artifacts, such as Gibbs ringing, resolution degradation, and various motion effects. This project is aimed at overcoming these problems by developing new model-based imaging techniques that can incorporate a priori information into the imaging process effectively. Application of these techniques to cardiac imaging and functional brain mapping is also addressed.
^ Multisensor Information Fusion
Z. P. Liang, H. Pan, and K. Y. Cheng*
DARPA, MDA972-00-1-0020
This project is a component of the DARPA grant on Center for Bio-Optoelectronic Sensor Systems (BOSS). The primary mission of this center is to develop sensor and processing technology for detection of biochemical agents in battlefield situations. Prof. Liang is responsible for developing statistical algorithms for multisensor information fusion.
^ Statistical image reconstruction:
Z. P. Liang,* C. Potter, B. Carragher
National Institutes of Health, RO1 GM61939
The primary goal of the project is to develop practical image reconstruction methods for high-resolution imaging from electron microscopy data, particularly in the presence of uncertainties in data acquisition parameters (say, projection angle). We formulate the problem as a statistical parameter estimation problem by introducing a proper model for the object (say, virus) to be imaged. This research effort promises to provide a brand-new solution to the long-standing problem in electron microscopy.
