Purpose: To develop a T1rho-prepared, balanced gradient echo (b-GRE) pulse sequence for rapid three-dimensional (3D) T1rho relaxation mapping within the time constraints of a clinical exam (<10 minutes), examine the effect of acquisition on the measured T1rho relaxation time and optimize 3D T1rho pulse sequences for the knee joint and spine.
Materials and methods: A pulse sequence consisting of inversion recovery-prepared, fat saturation, T1rho-preparation, and b-GRE image acquisition was used to obtain 3D volume coverage of the patellofemoral and tibiofemoral cartilage and lower lumbar spine. Multiple T1rho-weighted images at various contrast times (spin-lock pulse duration [TSL]) were used to construct a T1rho relaxation map in both phantoms and in the knee joint and spine in vivo. The transient signal decay during b-GRE image acquisition was corrected using a k-space filter. The T1rho-prepared b-GRE sequence was compared to a standard T1rho-prepared spin echo (SE) sequence and pulse sequence parameters were optimized numerically using the Bloch equations.
Results: The b-GRE transient signal decay was found to depend on the initial T1rho-preparation and the corresponding T1rho map was altered by variations in the point spread function with TSL. In a two compartment phantom, the steady state response was found to elevate T1rho from 91.4+/-6.5 to 293.8+/-31 and 66.9+/-3.5 to 661+/-207 with no change in the goodness-of-fit parameter R2. Phase encoding along the longest cartilage dimension and a transient signal decay k-space filter retained T1rho contrast. Measurement of T1rho using the T1rho-prepared b-GRE sequence matches standard T1rho-prepared SE in the medial patellar and lateral patellar cartilage compartments. T1rho-preparedb-GRE T1rho was found to have low interscan variability between four separate scans. Mean patellar cartilage T1rho was elevated compared to femoral and tibial cartilage T1rho.
Conclusion: The T1rho-prepared b-GRE acquisition rapidly and reliably accelerates T1rho quantification of tissues offset partially by a TSL-dependent point spread function.
Copyright (c) 2008 Wiley-Liss, Inc.