EE-CS Seminar on Friday, May 11, 10:40-11:30
  • FENS
  • EE-CS Seminar on Friday, May 11, 10:40-11:30

You are here

EE-CS Seminar on Friday, May 11, 10:40-11:30 @FASS 1081

"Processing and segmentation of high angular resolution diffusion imaging (HARDI) data modeled by orientation distribution functions"

by H. Ertan Çetingül, Siemens Corporation, Corporate Research and Technology in Princeton NJ, USA

Title:

Processing and segmentation of high angular resolution diffusion imaging (HARDI) data modeled by orientation distribution functions

Abstract:

Quantitative characterization of the brain circuitry is an important problem in neuroradiology as damage to this circuitry may indicate neurological disease. Diffusion magnetic resonance imaging (DMRI) is presently the only available non-invasive technique to investigate the neural architecture of the brain in vivo. DMRI produces images of biological tissues by measuring the anisotropy of water diffusion. Subsequently, the white matter fiber orientations can be inferred from the directions of maximum diffusion. State-of-the-art DMRI techniques such as high angular resolution diffusion imaging (HARDI) enables the reconstruction of the orientation distribution function (ODF), which offers improved accuracy in resolving intra-voxel complexities over the diffusion tensor model, currently the de facto standard for clinical applications.

Many challenging issues need to be addressed for HARDI to be studied in the correct mathematical setting and to be beneficial in clinical research. In this talk, I will present a mathematical framework to perform essential processing operations such as averaging, interpolation, and filtering of ODF images. By treating the information on the shape and orientation of an ODF as separate entities on Riemannian manifolds, we show how to perform these operations by using a novel group action induced distance for ODFs. I will also present a method to segment ODF images into multiple regions. Our approach integrates tools from sparse representation theory and Riemannian geometry into a graph theoretic segmentation framework. This method can incorporate weak supervision in the form of user interaction to differentiate between anatomically distinct regions with similar ODFs and group different ODFs in the same fiber tract. The proposed methods can be used to generate more accurate white matter atlases for statistical studies of neurological diseases.

Bio:

H. Ertan Cetingul works as a research scientist at Siemens Corporation, Corporate Research and Technology in Princeton NJ, USA. Before joining to Siemens, he received his Ph.D. degree in biomedical engineering from The Johns Hopkins University, Baltimore MD, USA, in 2011, his M.S. degree in electrical and computer engineering from Koc University, Istanbul, Turkey, in 2005, his B.S. degree in electrical and electronics engineering and a minor degree in business administration from Middle East Technical University, Ankara, Turkey, in 2003. His current research focuses on developing new methods by using tools/ideas from computer vision, pattern recognition, Riemannian geometry, harmonic analysis, and compressed sensing to infer the human connectome and promote its use in the clinic.