INTRODUCTION

Quantitative T1ρ mapping has gained increasing attention due to its high sensitivity to low-frequency molecular motion process that allows characterization of macromolecular composition and proton exchange in tissue.¹ Standard T1ρ preparation uses a composite continuous-wave (CW) RF pulse for spin locking, but it is relatively sensitive to B₀ and B₁ imperfections. Self-compensating spin lock pulses and 180° refocusing pulses can be used in the preparation to partially compensate for B₀ and B₁ inhomogeneities to reduce banding artifacts, but the off-center slices are still subject to quantification inaccuracy. 3D Magnetization-Prepared Angle-Modulated Partitioned k-Space Spoiled Gradient Echo Snapshots (3D MAPSS) is a fast MP-GRE technique using RF phase cycling and a variable flip angle train to achieve fast T1ρ mapping with high accuracy and spatial fidelity.² Its quantification accuracy however is compromised by the CW-T1ρ preparation. Instead, adiabatic RF pulses with both amplitude and frequency modulations, such as the hyperbolic secant pulses (HSn, n = 1, 2, 4, 8), have a wider bandwidth and are much less sensitive to B₀ and B₁ inhomogeneities, as shown in some previous studies.^3,4 The purpose of this study was to implement a 3D MAPSS sequence with adiabatic T1ρ preparation to achieve fast, accurate quantification with high spatial resolution. The sequence was tested on phantom, human knee and brain at 3T.

METHODS

Figure 1 illustrates the 3D MAPSS T1ρ sequences with continuous-wave (Figure 1a: CW-T1ρ) and adiabatic (Figure 1b: Adb-T1ρ) RF pulses used for spin locking. CW-T1ρ preparation was implemented as a composite RF pulse consisting of 90°-TSL/2-180°-TSL/2-90°, where TSL is the time of spin locking. Adiabatic full passage (AHP) RF pulses with HS8 amplitude and frequency modulations were used for Adb-T1ρ preparation. A group of four AHP pulses following MLEV phase cycling were used for spin locking. For CW-T1ρ, the last 90° hard pulse was used for phase cycling by reversing its phase for two paired scans. For Adb-T1ρ, phase cycling was achieved by reversing the frequency modulation for the second half of the last AFP pulse. Both cases lead to a pair of scans with same TSL but opposite longitudinal magnetization (M_z+ or M_z-) immediately before data acquisition. A complex subtraction of data from the paired scans can eliminate the contaminating signal from T1 relaxation.² CW-T1ρ and Adb-T1ρ experiments were performed on phantom (three pairs of tubes with 2%, 3% and 4% homogeneous agarose concentrations), human knee and brain (male healthy volunteer, 45 years old) on a 3T clinical scanner (Ingenia Elition, Philips Healthcare, The Netherlands). Table 1 summarizes the sequence parameters for each experiment. T1ρ maps were calculated using a mono-exponential two-parameter fitting algorithm. The center slice (slice 24) and two off-center slices (slices 8 and 41) of the phantom were selected for quantification using a custom analysis tool developed in IDL (Exelis VIS, Boulder, CO). The study was approved by local institutional review board and consent was received from the participating subject.

RESULTS

Figure 2 shows T1ρ maps of the phantom at the center slice (slice 24) and two off-center slices (slices 8 and 41) where frequency offset is present. The T1ρ maps of center slice look similar and are free of artifacts (Figure 2, top row) for both CW-T1ρ and Adb-T1ρ. Severe artifacts can be observed in the two off-center slices for CW-T1ρ (middle and bottom left), indicating that Adb-T1ρ is less sensitive to B₀ field inhomogeneities. Quantitative results further confirmed that Adb-T1ρ provides more consistent T1ρ measurements in the off-center slices compared to CW-T1ρ. For the center slice (slice 24), both CW-T1ρ and Adb-T1ρ provided very consistent T1ρ values (largest difference < 2%) for tubes with the same agarose concentration. However, for slice 8, CW-T1ρ values were (38.21±1.28 ms, 34.33±1.19 ms), (53.23±2.28 ms, 43.59±1.25 ms), (83.41±7.85 ms, 65.62±3.17 ms) for the three pairs of tubes, indicating significant variation of the T1ρ values (largest difference = 27%) for tubes with same agarose concentration. In contrast, the T1ρ values from Adb-T1ρ were more consistent as follows, (49.00±1.72 ms, 51.45±1.96 ms), (67.55±2.45 ms, 63.16±2.09 ms), (100.32±5.22 ms, 95.89±2.95 ms) for three pair of tubes (largest difference = 7%). Similar observations were found for slice 41. Figure 3 shows good delineation of the knee cartilage for both CW-T1ρ and Adb-T1ρ maps without apparent artifacts or boundary blurring. Figure 4 illustrates example CW-T1ρ and Adb-T1ρ maps with isotropic resolution of the brain in sagittal (top row), coronal (middle row), and axial (bottom row) views.

DISCUSSION

3D MAPSS T1ρ mapping with adiabatic RF pulses was successfully implemented and the phantom experiment demonstrated that Adb-T1ρ is less sensitive to frequency offset compared to CW-T1ρ, suggesting Adb-T1ρ may provide more accurate T1ρ measurement in conditions where magnetic field is inhomogeneous. The Adb-T1ρ sequence also provided good image quality for human knee and brain T1ρ mapping. Future study is warranted to quantitatively compare the performance of Adb-T1ρ to conventional CW-T1ρ in the presence of B₀­ and/or B₁ inhomogeneities in human knee and brain applications.

CONCLUSION

3D MAPSS Adb-T1ρ offers fast and high resolution T1ρ mapping as 3D MAPSS CW-T1ρ and is more robust to B₀ inhomogeneities to potentially obtain more accurate T1ρ quantification.

Acknowledgements

None.