Abstract #2737

The Use of Iteratively Reweighted Least Square (IRLS) in the Calculation of Tissue Susceptibility

Tian Liu^1,2, Cynthia Wisnieff^1,2, Craig Horenstein³, Krishna Surapaneni³, Yi Wang^1,2

¹Biomedical Engineering, Cornell University, Ithaca, NY, United States; ²Radiology, Weill Cornell Medical College, New York, NY, United States; ³Radiology, Columbia University, New York, NY, United States

Calculation of susceptibility from measured field map is an ill-posed inverse problem. In addition, the noise on the field map may not be Gaussian in signal void regions or due to improper phase unwrapping, complicating the inversion. In this abstract, we propose to use an Iteratively Reweighted Least Square (IRLS) algorithm to automatically identify the outlier pixels with non-Gaussian field noise, and attenuate the weighting for these pixels. Preliminary results showed that IRLS is able to suppress streaking artifacts in the presence of clusters of outlier pixels.