March 28, 2005

This Is Your Brain on Statistical Complexity (This Week at the Complex Systems Colloquium)

Attention conservation notice: This is an attempt to increase the attendance at the complex systems colloquia by blogging about them in advance. Of minimal relevance if you're not in Ann Arbor or don't care about complex systems, information theory, quantitative measures of complexity, brain imaging, or summarizing ridiculously largge amounts of data.

For this week's colloquium, we are very happy to have as our speaker Karl Young, from UCSF. In the early 1990s, Karl wrote what are (in my humble and quite unbiased opinion) some of the most important papers on quantitative, informational measures of complexity (e.g., 1, 2, 3). Then, mysteriously, he left the lush and easy life of complexity for the arid desolation of neuroscience, sustained in his wanderings in the wilderness by the ravens of NMR physics and brain imaging. Now, happily, he has returned to us with good news, prophesying the utility of complexity measures in making sense of brain images:

"MRI Diagnostic Tools Via Statistical Complexity Measures"
Magnetic resonance imaging (MRI) data, which are available in a large variety of modalities, has led to challenges regarding how to best utilize and interpret combined information for diagnostic purposes. For example, a MRI study of the brain may involve structural, spectroscopy, perfusion, and functional MRI in the same session, providing anatomical, metabolic, physiological, and functional information. Great progress has been made in registering different MRI modalities via the use of brain atlases, so that regional information is also maintained. However, a major problem with this approach is identification of relevant information for diagnosis from the huge amount of regional and multi-modal information available. In an attempt to generate sensitive methods for diagnostic classification this research explores a complimentary, global approach that utilizes entropy and statistical complexity measures applied to multi-modal data to obtain global measures of brain function. As a demonstration the BrainWeb Simulated Brain Database is used to simulate data based on a common model of neurodegeneration. Entropy and statistical complexity are shown to sensitively track simulated disease state.

A preliminary paper has already appeared in Physical Review Letters, but I know Karl has a lot more done on this topic than showed up in the PRL, and I hope we'll get to hear about it.

4 pm, Thursday, March 31, in room 335 West Hall, Central Campus

Complexity; Minds, Brains, and Neurons; Engimas of Chance

Posted at March 28, 2005 08:26 | permanent link

Three-Toed Sloth