Improved Look-Locker Acquisition Scheme and Curve Fitting Procedure
C Hui1*, P Narayana2, (1) UTHSC-Houston, Houston, TX, (2) UT Medical School, Houston, TXTU-G-217A-5 Tuesday 4:30:00 PM - 6:00:00 PM Room: 217A
Purpose: To improve the quality of 3D Look-Locker (LL) T1 mapping by optimizing the k-space segmentation in MRI. In addition, a multi-step curve fitting procedure is developed to reduce the effect of curve fitting error on the T1 estimation.
Methods: Traditional 3D LL acquisition generally utilizes centric encoding scheme that is limited to single phase encoding direction. To minimize reconstruction artifacts, we optimized the k-space segmentation by implementing the elliptical segmentation scheme along both phase encoding directions. We also developed a multi-step curve fitting procedure to compute the T1 values. First, four-parameter curve fit (flip angle, inversion angle, equilibrium magnetization and T1) is performed to obtain the angle maps. Then, the acquired angle maps are smoothed based on the chi-square weighting of the fit. These smoothed angle maps are treated as the actual angle maps and a two-parameter (equilibrium magnetization and T1) fit is used to create the final T1 map. Simulation and phantom study were performed to assess the performance of the proposed LL modification and the multi-step fitting algorithm.
Results: Reconstruction artifacts are reduced using the elliptical segmentation scheme relative to the traditional encoding method. In addition, the subsequent T1 computation using the elliptical encoding data along with the chi-square weighted angle maps is shown to yield more precise T1 values and was more immune to fitting errors.
Conclusions: An elliptical centric k-space segmentation scheme is implemented on a 3D LL sequence to reduce reconstruction artifacts. A multi-step curve fitting procedure that uses chi-square weighting is used to reduce the effect of false angles obtained from spurious signal. These methods were shown to improve the precision and accuracy of 3D LL T1 estimation.