Recent interest in the development of multispectral (MS-) qMRI pulse sequences (1-10) has been motivated by the increasing awareness of the clinical benefits afforded by: a) continuous contrast navigation via on-demand MR image synthesis, and b) the improved diagnostic accuracy possible with tissue integrity and hydration measurements; e.g. by the relaxation times (T1, T2) and the water-normalized proton density (PD). Therefore, there is great need for platform-independent MS-qMRI pulse sequences that are scantime efficient and produce superior image quality. Here we report the development of the quadra fast spin-echo (quadra-FSE), which is the concatenation of two dual-echo FSE acquisitions differing solely in the repetition time (Figure 1) and describe the theoretical underpinnings of the accompanying qMRI algorithms for mapping T1, T2, and PD. In addition, we validate the quantitative accuracy of this dual-saturation and dual-echo FSE pulse sequence and the corresponding qMRI algorithms, with a multi-compartment qMRI phantom (Figure 2) independently calibrated with the reference “gold standard” conventional pulse sequences for T1 (multi-TI inversion recovery single-shot-FSE) and multi-TE conventional spin echo for T2.

All qMRI algorithms were programmed in Mathcad (PTC, Needham, MA). The T1 mapping algorithm is based on the T1-implicit equation:

see Eq. 1 in Figure 4

which leads to a root-finding problem solvable the Ridder and Brent numerical methods (standard Mathcad functions); is the pixel value ratio between the first echoes of each DE-FSE acquisition and the function F is given by (11, 12):

see Eq. 2 in Figure 4

The T2 mapping algorithm combines the exact dual-echo formula applicable for short and intermediate T2 values and a maximum-likelihood algorithm valid for long T2s. Finally, the PD mapping algorithm is based on the principle of pulse sequence relaxation time unweighting (T1 and T2), followed by low spatial frequency demodulation to correct for receiver coil profile inhomogeneities, and pixel value calibration relative to water. Phantom experimentation was performed at room temperature (20±0.5ºC) using a 3T clinical scanner (Discovery MR750w, GE Healthcare, Waukesha, WI).

The logical architecture of the MS-qMRI algorithms is shown in the flowchart of Figure 3. The top row shows the four directly acquired images per slice generated by the quadra-FSE pulse sequence; these are labelled as Sat1_E1&2 and Sat2_E1&2. In the current implementation, Sat1_E1 and Sat2_E1 are used for T1 mapping via Eq. 1. Furthermore, Sat2_E1 and Sat2_E2 are used for T2 mapping. In the final steps, the residual T1 and T2 weightings of the Sat1_E1 image are reversed, the low spatial frequencies are demodulated, and the pixel values are normalized to that of water. Typical resulting maps are displayed in the bottom row. Additionally, the quadra-FSE T1 and T2 values are in very good agreement (within 10%) with the reference values afforded the gold standard conventional multi-TI and multi-TE conventional pulse sequences.A new MS-qMRI pulse sequence and accompanying mapping T1, T2, and PD algorithms have been developed. This pulse sequence termed quadra-FSE is simple, does not use inversion pulses which are prone to flip angle errors, is scan-time efficient, and nearly platform independent. This work could have implications for synthetic MRI, clinical and research applications of qMRI, and for correlation time diffusion MRI.