The optimization of T 1 relaxation measurements is a longstanding problem and numerous articles have addressed the subject (1-11). For a given application, the optimization of T 1 measurement depends on a number of factors, including the range of T 1 times of interest, the measurement technique used to estimate the relaxation time, and the acquisition time available for the measurements. Most work on T 1 optimization to date has focused on the efficiency of T 1 estimation (precision per unit time), trading off the number and distribution of inversion times against the number of acquisitions (signal averaging) to maximize the efficiency without consideration of explicit limits on the total length of the experiment (1-4,12). In this article we investigate optimization of the precision of brain T 1 measurements when time constraints preclude optimal efficiency of T 1 estimation. In particular, we are interested in the choice of the number of TI times and their values to optimize the precision of T 1 measurements in the human brain in a fixed acquisition time that limits two important parameters of T 1 measurement. First, the number of acquisitions is fixed and the number of inversion times must be traded off with the range of inversion times that can be sampled. Second, the total acquisition time is limited such that all TI Յ 3 ϫ T 1 max. For this application, the range of T 1 times is moderate (500 -1500 ms in healthy brain, 300 -2000 in pathology) (13). We consider modified fast inversion-recovery (MFIR) (14) with segmented k-space acquisition in the MFIR pulse sequence (15,16) and nonlinear least-squares curve-fitting to estimate T 1 (17).
MATERIALS AND METHODS
NumericalCurve-fitting to estimate T 1 from the inversion-recovery images was based on the Levenburg-Marquardt method (17). With the MFIR method, the repetition time (TR) is constant for all TI times and the total measurement time is proportional to the product of TR and the number of TI times, N. The 3-parameter signal model for the MFIR experiment is given by Eq. [1] (14,16,18):