T1 and T2 quantification from standard turbo spin echo images.

Abstract

To extract longitudinal and transverse (T1 and T2 ) relaxation maps from standard MRI methods. Bloch simulations were used to model relative signal amplitudes from standard turbo spin-echo sequences: proton density weighted, T2 -weighted, and either T2 -weighted fluid attenuated inversion recovery or T1 -weighted images. Simulations over a range of expected parameter values yielded a look-up table of relative signal intensities of these sequences. Weighted images and flip angle maps were acquired in 8 subjects at 3 T using both single and multislice acquisitions. The T1 and T2 maps were fit by comparing the weighted images to the look-up table, given the measured flip angles. Results were compared with inversion recovery and multi-echo spin-echo experiments. A region analysis showed that relaxation maps computed from single-slice proton density, T2 and T1 weighting provided a mean T1 error of 4% in gray matter and 11% in white matter, and a mean T2 error of 3% and 4%, respectively, in comparison to reference measurements. In multislice acquisitions that are optimized to reduce cross-talk and incidental magnetization transfer, the mean T1 error was 7% in gray matter and 1% in white matter, and the mean T2 errors were 3% and 4%, respectively. The best T1 results were achieved using proton density, T2 and T1 weighting rather than the fluid attenuated inversion recovery, although T2 maps were largely unaffected by this choice. Incidental magnetization transfer reduced T1 accuracy in standard interleaved multislice acquisitions. Through exact sequence modeling and separate flip angle measurement, T2 and T1 may be quantified from a turbo spin-echo brain protocol with proton density, T2 , and T1 weighting.