DUNE (Data Mining of Uncertainty Data 2007)
Keynote Speaker: Prof. Ambuj Singh (University of California at Santa Barbara)
Title: Uncertainty in bioimage databases: querying and mining challenges
Time: 13:30-14:05, 28 October 2007
Venue: Room "Big Blue C" (Room H)
Abstract:
Rapid advances in imaging technologies have opened the door for a new era of biological research, leading to quantitative understanding of complex systems. The realization of this potential, however, critically depends on the availability of tools and technologies to extract information from imaging data so as to achieve new biological insights and understanding. The inherent nature of bioimages leads to uncertainty in data, e.g., to which retinal image layer does a given pixel belong, what is the thickness of a neurite, how many photoreceptors are there, or what is the length of a microtubule. Probability distributions are the natural way to model such phenomena. The aggregation of data (such as the thickness of a layer measured at different spatial locations or the length of all microtubules in an image) again leads to probability distributions. I will discuss new probabilistic image analysis algorithms for measuring biological processes. These include image segmentation, computing sizes of neurons and their spatial distributions. I will discuss new techniques for querying probabilistic data through index structures that can store distributions at multiple resolutions and scale to large datasets. Finally, I will discuss techniques for exploratory mining of uncertain bioimage information.
[Slide]