We have previously demonstrated a fast quantitative, T₁ mapping method based on inversion recovery echo planar imaging (IR-EPI) [1]. The method relies on the acquisition of multiple EPI slices following a broadly slab-selective inversion. The slice acquisition order is then permuted by a fixed ‘skip-factor’ for each of the subsequent inversion recovery periods. By rapidly acquiring multiple inversion times (TI) for each slice, accurate T₁ maps can be obtained. Nevertheless, the reordering scheme used is empirically chosen and may not be optimal for the desired range of tissue parameters. Rather than a fixed skip-factor, our goal was to find an optimal reordering scheme, i.e. one that maximizes the discrimination between different tissues. In this work we demonstrate that optimizing the reordering scheme permits a 5-fold reduction in the number of inversions required resulting in significant scan time savings without adversely affecting the quantitative maps obtained. To find the optimal reordering we simulated the magnetization for a given slice ordering of all possible tissue T₁ values. The dot product matrix was then computed and a similarity metric was defined as the sum of all off-diagonal elements, similarly to [2]. A genetic optimization algorithm [3] was then used to find an ordering that minimized the chosen similarity metric. The original skip-factor ordering with 21 inversions and a skip-factor of 7 was used as a baseline. To minimize computation time, the optimization was done assuming 4 inversions and was only allowed to run for 30 generations (iterations). All experiments were conducted on a 3T Siemens Trio whole body scanner (Siemens Healthcare, Erlangen, Germany) equipped with a 32-channel phased-array head coil. A phantom composed of vials with different concentrations of doped water was used to mimic the variety of T₁ values in the brain. The phantom was scanned at a resolution of 1.2 mm³ using both the unoptimized (21 inversions) and the optimized (4 inversions) ordering with the following acquisition parameters: TR/TE/flip angle/BW/matrix size/slices/R = 6690 ms/26.9 ms/ 90°/ 1240 Hz/pixel / 192×192/100/4 with FLEET autocalibration [3] (with α=10° and 5 preparatory pulses) and online GRAPPA reconstruction. Total scan time for the IR-EPI data was 140/27 seconds for the unoptimized/optimized sequence, plus a fixed 30 s for GRAPPA autocalibration. The acquired data was processed offline in Matlab. The signal in each voxel was compared to a pre-computed dictionary of signal magnetizations and the best match selected [3]. The mean of the T₁ in each vial obtained using both orderings was used to calculate a best-fit curve whose R² value was used to measure the quantitative similarity between the two orderings. A healthy 24 y.o. female volunteer was recruited and provided written informed consent. The subject was scanned with both slice orderings.A visual comparison between the skip-factor 7 unoptimized and optimized slice orderings is shown in Figure 1. A comparison between the T₁ values obtained with each ordering is shown in Figure 2 along with a best-fit curve whose R² value was 0.9994 indicating the closeness of the quantitative values obtained with the two orderings. The standard deviation for the optimized ordering, however, is substantially smaller. Representative images from the phantom for both acquisitions are shown in Figure 3, showing that the T₁ values match known values for the solutions. Maps from the human subject for both slice ordering are shown in Figure 4, where the T₁ values approximate reported values for brain tissue at 3T [5] and are similar to those of the unoptimized ordering.Despite its clinical importance, quantitative tissue mapping is hindered by the long scan time required. Our optimized IR-EPI method requires less than 30 seconds for fully quantitative, 1.2 mm³ isotropic T₁ maps representing a 5-fold saving over previous techniques. Similar protocols have been successfully used at higher (7T) fields making this method easily portable to higher fields. Current limitations include the long processing time required for the optimization but may be resolved by use of improved optimization algorithms and parallel processing.An optimization method to reduce scan time in IR-EPI was developed and tested on a clinical 3-T scanner. Potential applications for the increased acceleration include clinical scanning and imaging of dynamic processes such as dynamic contrast enhancement [6].