Meeting Banner
Abstract #3577

A Robust and Automated Method for Estimating the Expected Signal Standard Deviation in DWI Datasets

Lin-Ching Chang1, Carlo Pierpaoli2

1Department of Electrical Engineering and Computer Science, The Catholic University of America, Washington, DC, United States; 2National Institutes of Health, Bethesda, MD, United States


Correctly estimating the expected signal standard deviation (SD) due to thermal noise in diffusion weighted images (DWIs) is important for controlling image quality, correctly computing the chi-squares value, image registration and outlier detection. For single channel acquisitions, signal SD could be estimated from a ghost-free region of the image background. However, the signal in the background of DWIs acquired on modern clinical scanners cannot be used for this purpose. This paper proposes an object-based method taking advantage of robust regression and residual analysis to estimate signal SD. Results from simulation indicate that the proposed method performs very well even in the presence of DWI volumes which are completely corrupted.