Macromolecular proton fraction as an ultimate source of brain tissue contrast in ultra-high magnetic fields
Anna V Naumova1,2, Andrey E Akulov3, Alexandr V Romashchenko 3, Oleg B Shevelev 3, Marina Yu Khodanovich 2, and Vasily L Yarnykh 1,2

1Radiology, University of Washington, Seattle, WA, United States, 2National Research Tomsk State University, Tomsk, Russian Federation, 3Institute of Cytology and Genetics, Novosibirsk, Russian Federation


This study provides methodological background for neuroimaging applications of fast 3D MPF mapping in ultra-high magnetic fields and demonstrates that MPF presents the most effective source of high-field brain tissue contrast. Unique contrast features, high spatial resolution, and the capability to provide quantitative information about myelination make MPF mapping an ultimate tool for small animal neuroimaging in high magnetic fields.


T1-weighted imaging in ultra-high magnetic field suffers from the convergence of T1 relaxation times with an increase of field strength1, which makes difficult or even impossible generation of sufficient positive contrast between white (WM) and gray matter (GM) for structural neuroanatomical applications. Macromolecular proton fraction (MPF) mapping2,3 is a method potentially capable to mitigate this problem due to strong myelin-based WM-GM contrast4-6 and independence of this parameter of magnetic field strength. MPF is a key parameter of the two-pool model of magnetization transfer (MT) defined as a relative amount of immobile macromolecular protons involved into magnetization exchange with water protons7. A recent time-efficient single-point 3D MPF mapping method3 allows reconstruction of MPF maps from three source images with high spatial resolution and excellent tissue contrast. However, this approach is based on constraining certain two-pool model parameters and their combinations, which are field-dependent and need to be determined for specific field strength. The objective of this study was to adopt fast 3D MPF mapping as anatomical and quantitative neuroimaging modality for small animal applications in ultra-high magnetic fields based on the characterization of the two-pool model parameters in brain tissues.


Image Acquisition. Four adult male Wistar rats were imaged under isoflurane anesthesia on a 11.7T animal MRI scanner (Bruker BioSpec, Germany). Brain Z-spectroscopic images were obtained using a 3D MT-prepared spoiled gradient echo (GRE) sequence with TR/TE=25/2.7 ms and excitation flip angle α=9°. For off-resonance saturation Gaussian pulse was used with duration 10 ms, 12 offset frequencies (Δ) in a range 0.75-48 kHz and 3 effective flip angles FAMT = 500, 1000, and 1500º. Complementary R1=1/T1 maps were obtained using the variable flip angle (VFA) method with a 3D GRE sequence (TR/TE=25/2.7 ms, α=3, 12, 20, 25, 30°). All images were acquired with resolution of 200x200x400 µm3 and whole-brain coverage. Scan time for each Z-spectral and VFA data point was 1 min 41 s. To correct for field heterogeneities, 3D B0 and B1 maps were acquired using the dual-TE8 and AFI9 methods, respectively. Additionally, 2D T2 mapping was performed using multiple spin-echo sequence with TR=5 s and 16 echoes with 10 ms echo spacing. To demonstrate the feasibility of high-resolution whole-brain MPF mapping, 3D MPF maps were obtained from three source images (MT-, PD-, and T1-weighted) with isotropic 170 µm resolution using the single-point method with the synthetic reference image3.

Image Reconstruction and Analysis. To comprehensively characterize tissue contrast, the following parametric maps were compared: MPF, k, T2F, and T2B maps obtained by fitting Z-spectroscopic data to the pulsed MT model2; MPF map reconstructed using the single-point method with synthetic reference image3 from a data point with Δ= 5 kHz and FAMT = 500º; maps of combinations of the two-pool model parameters used as constraints in the single-point MPF reconstruction algorithm, R=k(1-MPF)/MPF and R1T2F 2; R1 and proton-density (PD) maps reconstructed from VFA images; and T2 and PD maps reconstructed from multi-echo images. Parameter measurements were performed in regions-of-interest (ROIs) for a series of WM and GM structures. These data were used to determine parameter constraints in the algorithm2, compare MPF obtained from Z-spectra fit and single-point reconstruction, and compare WM-GM contrast between parameters.


Example experimental and fitted Z-spectra from ROIs corresponding to the corpus callosum and cortex are presented in Figure 1. Parameter constraints for the single-point algorithm were determined as follows: R = 23 s-1, R1T2F = 0.013, and T2B =10 µs. There was an excellent agreement between MPF maps reconstructed from Z-spectra fit and single-point synthetic reference method (Figure 2). MPF values obtained with both methods correlated with r=0.99 and no significant bias (Figure 3). MPF showed notably superior tissue contrast compared to other parametric maps, as indicated by percentage differences between WM and GM (Figure 4). Illustration of the rat brain anatomy on a high-resolution 3D MPF map is presented in Figure 5.

Discussion and Conclusion

This study demonstrates the first application of fast 3D MPF mapping for high-resolution quantitative neuroimaging in ultra-high magnetic fields and indicates that MPF provides the most effective source of brain tissue contrast. MPF values in WM and GM structures at 11.7T (Figure 3) were similar to those at lower field strengths4-6, thus confirming field independence of MPF. Among constrained parameter combinations, only the product R1T2F exhibited apparent field dependence, being about two-fold smaller compared to 3T 2. Unique contrast features, high spatial resolution, and quantitative information about myelination make MPF mapping an ultimate tool for neuroimaging in high magnetic fields.


Russian Science Foundation (project #14-45-00040), NIH grant R21EB016135, National MS Society grant RG4864A1.


1. Balchandani P, Naidich TP. Ultra-High-Field MR Neuroimaging. AJNR Am J Neuroradiol 2015; 36(7): 1204-15.

2. Yarnykh VL. Fast macromolecular proton fraction mapping from a single off-resonance magnetization transfer measurement. Magn Reson Med 2012;68:166-178.

3. Yarnykh VL. Time-efficient, high-resolution, whole brain three-dimensional macromolecular proton fraction mapping. Magn Reson Med 2015 Jun 22.

4. Underhill HR, Rostomily RC, Mikheev AM, at al. Fast bound pool fraction imaging of the in vivo rat brain: association with myelin content and validation in the C6 glioma model. Neuroimage 2011; 54:2052-2065.

5. Samsonov A, Alexander AL, Mossahebi P, et al. Quantitative MR imaging of two-pool magnetization transfer model parameters in myelin mutant shaking pup. Neuroimage 2012;62:1390–1398.

6. Yarnykh VL, Bowen JD, Samsonov A, et al. Fast whole-brain three-dimensional macromolecular proton fraction mapping in multiple sclerosis. Radiology 2015;274:210-220.

7. Morrison C, Henkelman RM. A model for magnetization transfer in tissues. Magn Reson Med 1995; 33:475–482.

8. Yarnykh VL. Actual flip-angle imaging in the pulsed steady state: a method for rapid three-dimensional mapping of the transmitted radiofrequency field. Magn Reson Med 2007;57:192–200.

9. Skinner TE, Glover GH. An extended two-point Dixon algorithm for calculating separate water, fat, and B0 images. Magn Reson Med 1997;37:628–630.


Figure 1. Experimental (points) and fitted (lines) Z-spectra in ROIs corresponding to the corpus callosum (white matter) matter and cortex (gray matter) of the rat brain obtained at the 11.7T magnetic field strength. The fitted parameters are as follows: MPF =12.9%, k=1.8 s-1, T2F = 21.9 ms, and T2B=12.0 µs for WM, and MPF = 6.1%, k=1.9 s-1, T2F = 33.3 ms, and T2B=10.0 µs for GM.

Figure 2. Example parametric maps of the rat brain obtained at the 11.7T: MPF maps reconstructed by Z-spectra fit (MPF-Zsp) and single-point method (MPF-1p), their difference (ΔMPF), maps of direct and reverse cross-relaxation rate constants (k and R), T2 maps for the free and bound protons (T2F and T2B), T2 map reconstructed from multi-echo data (T2), R1 map reconstructed from VFA data (R1), product R1T2F, and proton density maps reconstructed from VFA (PFvfa) and multi-echo (PDme) data.

Figure 3. Linear correlation between MPF measurements in brain structures obtained from maps reconstructed by Z-spectra fit (MPF-Zsp) and single-point method (MPF-1p). WM structures include cingulum, corpus callosum, external capsule, internal capsule, fornix, and cerebellar WM. GM structures include olfactory bulb, caudate putamen, hippocampus, superior colliculus, inferior colliculus, motor cortex, visual cortex, thalamus, and cerebellar GM.

Figure 4. Percentage difference between quantitative MRI parameters in WM and GM averaged across brain structures and describing tissue contrast on corresponding parametric maps (Figure 2).

Figure 5. Reformatted orthogonal sections of a 3D MPF map of the rat brain obtained using the single-point synthetic-reference method with isotropic resolution of 170 µm.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)