Xiaoming Yin1,2, Saurabh Shah3, Andrew C. Larson1,2
1Radiology, Northwestern University, Chicago, IL, United States; 2Electrical Engineering and Computer Science, Northwestern University, Evanston , IL, United States; 3Siemens Medical Solutions, Chicago, IL, United States
R2* is typically estimated via mono-exponential fitting of signal decay within a series of GRE images combined by the root sum-of-square (RSS) approach. However, RSS approaches rectify and bias noise in later TEs, resulting in systematic fitting errors for R2* estimation. Our work investigated the accuracy of low SNR R2* measurements for RSS reconstructed data. Through phantom, ex vivo, and volunteer studies, we compared the accuracy of R2* measurement using SNR-weighted least-square regression and SNR-based truncation methods. We found SNR-weighted least-square regression to be a simple and reliable approach to reduce R2* measurement error.