Intervertebral disc T1rho relaxation mapping with spin-lock 3D b-FFE imaging at 3T

Introduction: Quantitative T1ρ imaging (1) is a non-invasive MRI technique that has been shown to be sensitive to interactions between motion restricted water molecules and their surrounding macromolecular environment. T1ρ-weighted contrast is produced by transverse magnetization relaxation in the presence of an on-resonance continuous wave (cw) RF pulse (spin-lock pulse). T1ρ quantification should be helpful in identifying early interverterbral disc degeneration (2). In this study, a time-efficient spin-lock 3D balanced fast field echo (bFFE) imaging was developed for rapid in vivo interverterbral disc T1ρ relaxation mapping at 3T. It shows that median T1ρ value of the nucleus pulposus is greater than that of the annulus, consistent with results from previous studies (3). Significant reduction of T1ρ relaxation was found to be an indication of early interverterbral disc degeneration. Methods: The proposed spin-lock 3D bFFE sequence was implemented and applied for T1ρ quantification in three healthy volunteers on a 3T Philips Achieva MRI scanner with a spine coil. The diagram for the sequence is shown in Fig. 1. The unwanted bright fat signal was firstly suppressed by a spectrally selective RF pulse and the associated dephase gradient. The following T1ρ relaxation consists of an excitation pulse (90x), two spin-lock cw pulses (+y, -y, spin-lock frequency: 500Hz) and a restoration pulse (90y). Spin-lock times (TSLs) of 1, 10, 20, 30 and 40ms were used for T1ρ relaxation mapping. A half-alpha pulse and four startup pulses with dummy-scans were used to approach the steady-state but with T1ρ-weighted preparation maintained. The subsequent normal phase alternating bFFE readout was used for acquisition. The delay time (TI) after acquisition was set as 4000ms to restore equilibrium magnetization prior to the next T1ρ preparation. TE and TR were 2.5ms and 5.0ms respectively. The matrix size was 448x448x6 and the flip angle was 50. The T1ρ-weighted images with different TSLs were processed using an IDL program on a pixel-by-pixel basis to obtain the T1ρ relaxation map using the equation of S(TSL)=Aexp(T1ρ/TSL). Results: Scan time for each TSL was 46 seconds and the total scan time for T1ρ mapping with five TSLs took around five minutes. T1ρ weighted lower lumbar images with different TSLs and the calculated T1ρ relaxation map for one subject were shown in Fig. 2. It was seen that the median T1ρ value of the nucleus pulposus was greater than that of the annulus. The mean T1ρ values of the nucleus for each IVD were 103ms (L1L2), 133ms (L2L3), 54ms (L3L4) 100ms (L4L5) and 70ms (L5S1), respectively. A trend of reducing T1ρ values from L2L3 to L5S1 was evident except L3L4, consistent with previous report in (3). The significant reduction of T1ρ value at L3L4 may indicate early disc degeneration. This degenerative change was confirmed by a radiologist’s diagnostic reading who was blinded to T1ρ results. Discussion: Spin-lock 3D bFFE imaging allows acquisition for full coverage of spine without slice gap in sagittal orientation within 5 minutes with high spatial resolution. The considerable time savings make it feasible for a clinical T1ρ quantification scan protocol. Scan time could be further reduced by reducing the delay time for magnetization restoration. bFFE provides a time efficient acquisition with high signal-to-noise ratio under steady state. However, the acquisition under steady state eliminates parts of T1ρ-weighted contrast, so the T1ρ-weighted preparation has to be acquired during the transient state prior to the steady state (4). Thus, the multi-shot segmented k-space acquisition should be employed. bFFE imaging is also sensitive to B0 imhomogeneity and prone to banding artifact, so good shimming performance is critically required. Although the T1ρ map shows a little bit noisy due to the fine spatial resolution and small pixel size, it could be improved by reducing the matrix size or employing smoothing filters during post-processing. Acknowledgement: This work is supported by HK GRF grant CUHK464807. References: [1] Redfield AG, Phys Rev, 98:1787-1809(1955); [2] Auerbach JD et al, Eur Spine J 15:S338-344(2006); [3] Walter RT et al, JMRI 28:744-754(2008); [4] Blumenkrantz G et al, MRI 24:1001-1007(2006)