Research - Image Processing

Molecular and Cellular Image Analysis

Recent advances in biomedical imaging and nanotechnology have led to new optical imaging modalities based on quantum dot (QD) and molecular beacons (MBs) probes. These techniques enable us to study live biomolecular and cellular behavior for early disease detection and drug response.

However, one of the major roadblocks in translating molecular imaging data into clinical patient care is the lack of accurate quantification techniques to segment single articles, to find co-localization of probes, and to track particle movements in 2D and 3D space.

Previous research has only used centroid or Gaussian mask methods to localize particles. We have developed a powerful set of quantitative image analysis techniques such as active contour (AC) and level set (LS). These techniques allow automated segmentation of single-cell fluorescence images as well as co-localization analysis of single quantum-dot signals.

	Co-localization analysis of single quantum-dot signals (green)
	Automated segmentation of single-cell fluorescence images (red)

Visualisation of cell data in 2D and 3D. The visualisation of this data is important for understanding drug-cell interactions.

The size of anatomy data from visible human project is large. For the male specimen, it is ~16GB (1878 sections); for female, it is ~40GB (5189 sections). We are developing an interactive system to visualize this large volume of data. Eventually, we want to develop a system to explore this data in virtual environment.

Virtual exploration of visible human specimen.