Math Colloquium: Statistical Quantitative Imaging

Monday, February 8, 2016

3:00 PM-4:00 PM

Russell Shinohara, Department of Biostatistics and Epidemiology, University of Pennsylvania

While conventional magnetic resonance imaging (MRI) is measured in arbitrary intensity units, much work has centered on the development of normalization procedures. In this talk, we propose a simple, fast, and robust statistical normalization method which does not require prior segmentation of a reference tissue. This technique allows us to measure changes across the brain using conventional MRI in interpretable statistical units, providing a promising new tool for large multi-center studies and clinical practice in which conventional MRI are used routinely.

We then propose methods for the identification of multiple sclerosis (MS) white matter lesions both cross-sectionally and longitudinally at the population level using these statistically normalized images which are now quantitative. Finally, we examine the prospect of quantifying disease activity and tissue damage using our statistical quantitative MRI, derived solely from widely available conventional MRI, compared to contrast-enhanced and advanced quantitative MRI modalities. Our preliminary results indicate that statistical quantitative MRI is significantly less costly, less invasive, and yet may provide more precise and sensitive biomarkers for disease-related changes in MS on MRI.

Contact Information

Pawel Hitczenko
phitczenko@math.drexel.edu