PURPOSE

Magnetization transfer (MT) is a magnetic resonance imaging (MRI) contrast mechanism that enables the detection of less mobile protons whose transverse relaxation times (T₂) is too short to be detected directly. Conventionally, a magnetization transfer ratio (MTR), defined by Eq. (1), has been used to analyze the degree of signal attenuation due to the MT effect,
$$ MTR(Δω,ω_1 )=(S_0-S(Δω,ω_1 ))/S_0 $$ (1)

where S₀ and S(Δω, ω₁) are the water signals acquired without/with off-resonance RF saturation, respectively, Δω is the RF frequency offset (Hz), and ω₁ is the RF saturation power (rad/sec). However, although the MTR value depends on the MT effect, it is a complex combination of various parameters, including MT parameters and water relaxations. As a result, although the MTR value has been linked to the extent of myelin damage ¹, it can also be influenced by edema and inflammation ^{2, 3}, and its variation in multiple sclerosis is often small ⁴. In this work, we developed a new MT data analysis and acquisition method with improved sensitivity and specificity to the MT effect compared to the conventional MTR.

METHODS

We define a new method, termed the double offsets and powers magnetization transfer ratio (dopMTR) in Eq. (2), which is the inverse subtraction of two MT signals acquired at two sets of sequence parameters including (Δω, ω₁) and (cΔω, cω₁) together with T_1w normalization (c is a constant). $$dopMTR(Δω,ω_1)=|S_0/S(Δω, ω_1 ) -S_0/S(cΔω, cω_1 ) | R_{1w} $$ (2)
Simulations of two-pool coupled Bloch equations, with varied macromolecular pool concentration (f_m), MT rate between macromolecular and water pools (k_mw), and water relaxations (T_1w, T_2w), were performed to study the specificity and sensitivity of the MTR and dopMTR to MT effect. Experiments on a series of cross-linked bovine serum albumin (BSA) samples and 4 healthy rat brains were also performed to evaluate the two methods. Three samples of 5% (sample #1), 10% (sample #2), and 15% (sample #3) (w/w) BSA in phosphate-buffered Saline (PBS) with pH of 7.0 were prepared. 0.075mM MnCl2 was added to these three samples to vary their water relaxations. These three samples have different f_m values (and thus also different R_1w values) and were used to study the dependence of the two MT analysis methods on f_m. Three additional samples of 0.05 mM (sample #4), 0.075 mM (sample #5), and 0.1 mM (sample #6) MnCl2 in PBS with pH of 7.0 were prepared. BSA was added to these samples until they reached 10% (w/w). These three samples, which have different R_1w but the same f_m, were used to study the dependence of the two MT analysis methods on R_1w. All animal experiments were approved by the local Animal Care and Usage Committee. Experiments were performed on a Varian 4.7T magnet with a 38-mm RF coil. The MT sequence contains a 5s continuous wave saturation pulse followed by a SE-EPI readout. For comparison with the two MT analysis methods, quantitative MT (qMT) using a selective inversion recovery (SIR) method was also performed ⁵.

RESULTS

Fig.1 shows simulations of MTR as a function of several sample parameters. The MTR values are nonlinearly dependent on f_m, k_mw, and T_1w. When f_m was increased from 5% to 15% (3 times), the MTR value was increased only from 64% to 83% (1.3 times), suggesting that the sensitivity of MTR to f_m was low. The MTR values have a weak dependence on T_1w. However, given its weak dependence on f_m this weak dependence on T_1w may still reduce the specificity of the MTR to the MT effect.
Fig.2 shows the simulated dopMTR as a function of several sample parameters. The dopMTR value is proportional to f_m, increases nonlinearly with k_mw, and is roughly independent of T_1w and T_2w, showing improved specificity and sensitivity to the MT effect when compared with the MTR method.
Fig. 3 shows images of the MTR, dopMTR, and qMT-determined f_m, R_1w, k_mw on BSA samples #1-6. The contrast among samples #1-3 in the dopMTR images is similar to the image of qMT-determined f_m. The images of the MTR, on the other hand, show nearly no contrast. Additionally, contrast among samples #4-6 can be observed in the image of the qMT-determined R_1w, but not in the image of the dopMTR.
Fig. 4 shows images of the MTR, dopMTR, and qMT-determined f_m, R_1w, k_mw in a representative rat brain. The white matter (WM) can be clearly found in the images of the dopMTR and qMT-determined f_m, but not the MTR.

DISCUSSION AND CONCLUSION

MT signals usually have contributions from both MT and direct water saturation (DS) effects. The MTR subtracts a control signal acquired without saturation from an MT signal which cannot fully remove DS and induces a complex dependence on T_1w. According to Bloch equations, the DS, but not MT effect, remains constant when Δω and ω₁ are changed in the same proportion. Thus, the MT signal measured with cΔω and cω₁ can be used as a reference signal, whereas the MT signal measured with Δω and ω₁ can be used as a label signal, to isolate the MT from DS effect. After removing the DS effect, R_1w can be easily removed by normalizing T_1w ⁶.

Acknowledgements

The author acknowledges grant support from NIH (R21 AR074261, R03 EB029078, and R01 EB029443)