Synopsis

MRI techniques that utilize the magnetization transfer (MT) effect are highly desirable as they potentially can map the myelin content. However, MT measurements produced by conventional MT-MRI techniques, such as MT ratio imaging, vary with the pulse sequence and the scan parameters used. The MT effect and the intrinsic spin–lattice relaxation, both a component of the MR longitudinal relaxation, must be separated to improve MT-MRI accuracy and precision. To solve the problem, a method for MT-MRI is developed in this work. In patients of multiple sclerosis, the present MT-MRI shows improved specificity to demyelinating lesions in the brain.

INTRODUCTION

Multiple sclerosis (MS) is an inflammatory, demyelinating condition of the central nervous system (CNS), which affects 2.3 million people worldwide. MRI is the preferred imaging modality for MS to detect demyelinating lesions in the CNS. For MS, MRI methods that can detect myelin content are highly desirable, such as magnetization transfer (MT) MRI, in which the myelin content is inferred from the exchange of magnetization between water and myelin and conventionally quantified by the MT ratio (MTR). MTR is useful but is strongly dependent on the pulse sequence and the scan parameters. Consequently, comparison between MTR measurements is rather difficult [1].

A more reliable approach is to measure the MT rate constant k, and more sophisticated MT-MRI methods have been developed [2,3]. However, another important problem is that the conventional MT measurement also includes a component of the spin–lattice relaxation (intrinsic T₁) and thus does not accurately reflect the myelin content. To solve the problem, a novel MRI method is developed that is feasible for routine clinical application and can produce two isolated quantitative images: the MT R_k map and the R₁ map. Different from the conventional approach, the R₁ map in the present method is not separately measured because such measurement is contaminated by the MT effect.

METHODS

Recall that MT is a relaxation mechanism of the longitudinal magnetization. The differential equation governing the longitudinal relaxation, including MT and the spin–lattice relaxation [4], is given by $$ \frac{M_z}{dt} = ( R_1 + Q R_k ) M_\mathrm{eq} + (R_1 + R_k) M_z ,$$ where M_eq is the magnetization at thermal equilibrium and R_k is the MT constant but is expressed in the same physical dimension of R₁. The present MT-MRI method quantifies MT by recording and analyzing the evolution of the longitudinal magnetization (M_z), as in a T₁ measurement. The pulse sequence utilizes 2D interleaved slice acquisition with a spiral trajectory [5] using an RF excitation pulse train, starting with an inversion pulse. An MT pulse is played out before each excitation pulse. There are hence multiple MT pulses between the acquisitions of a same slice. The macromolecular pool is therefore in a repeated steady-state of excitation and recovery. Q is a dimensionless function of time that describes such dynamic steady-state of the macromolecular pool driven by the MT pulses. The equation is solved numerically [6]. In our MT-MRI method, one scan is performed without and one performed with the MT pulses, keeping the RF transmit amplitude and the receiver gain the same. The image data of both scans are combined to fit Eq. (1); R₁ and R_k are obtained simultaneously and are isolated from each other.

RESULTS AND DISCUSSION

A sample slice from our MT MRI method for an MS patient is shown in Fig. 1. To facilitate discussion, an example lesion is labeled with a red arrow. Figure 1A is the MT rate constant R_k map and Fig. 1B the intrinsic R₁ map; they are generated together as described above. The labeled lesion is highly visible and has a sharp boundary in the R_k map but not so in the intrinsic R₁ map. Having sharp boundaries is a histopathological hallmark of MS demyelinating lesions. In addition, grey matter has significantly lower R_k relative to white matter because myelin is the dominant source of MT in the CNS. These imaging features confirm that separating R_k and R₁ in MT-MRI improves the MT imaging specificity to myelin, which our technique is designed for.

Figure 1C is the general/apparent R₁ map as obtained from a separate R₁ mapping; the apparent R₁ value is basically a mixture of the intrinsic R₁ and R_k. This apparent R₁ is the contrast mechanism of the conventional clinical T₁-weighted imaging shown in Fig. 1D. The labeled lesion is still visible in these unspecific images but the lesion boundary is not as prominent as the R_k map, which again demonstrates that the MT effect when isolated from the apparent longitudinal relaxation can improve the MRI lesion specificity.

Figure 1E is reconstructed following the conventional definition of the MTR. Because the MTR map is a mixture of R_k and R₁, it resembles the apparent R₁ map (Fig. 1C) and the T₁-weighted image (Fig. 1D).

CONCLUSION

It is important in MT-MRI to separate the intrinsic R₁ from the MT effect but the R₁ map is not to be measured individually from MT in the longitudinal relaxation measurement. The present MT-MRI can easily be completed in a clinically feasible scan time (<10 min, including the flip-angle measurement) and the MT R_k map can properly reflect the MS demyelinating lesions.

Acknowledgements

References

1. Ropele, S. & Fazekas, F. Magnetization transfer MR imaging in multiple sclerosis. Mult. Scler. Part II Nonconv. MRI Tech. 19, 27–36 (2009).

2. Gochberg, D. F. & Gore, J. C. Quantitative magnetization transfer imaging via selective inversion recovery with short repetition times. Magn. Reson. Med. 57, 437–441 (2007).

3. Kim, T., Shin, W. & Kim, S. Fast magnetization transfer and apparent T1 imaging using a short saturation pulse with and without inversion preparation. Magn. Reson. Med. 71, 1264–1271 (2014).

4. Caines, G. H., Schleich, T. & Rydzewski, J. M. Incorporation of magnetization transfer into the formalism for rotating-frame spin-lattice protonNMR relaxation in the presence of an off-resonance-irradiation field. J. Magn. Reson. 95, 558–566 (1991).

5. Glover, G. H. Simple analytic spiral K-space algorithm. Magn. Reson. Med. 42, 412–415 (1999).

6. Hsu, J.-J. Flip-angle profile of slice-selective excitation and the measurement of the MR longitudinal relaxation time with steady-state magnetization. Phys. Med. Biol. 60, 5785 (2015).