Phase imaging has become an alternative method to magnitude imaging for anatomical information (1). The classical approach is using gradient-recalled-echo (GRE) due to the linear phase profile to resonance frequency shift. In a previous research, phase imaging with transition-band balanced steady-state free precession (bSSFP) was explored to show refined anatomy of human brain at 7T (2). Pass-band bSSFP images have been studied for various purposes (3-5), but phase images of pass-band bSSFP have not been investigated well. In this study, we investigated phase images of pass-band bSSFP at multiple phase cycling (PC) angles at ultrahigh field of 9.4T. The result was compared with phase images from the conventional gradient-recalled-echo (GRE).

Sprague-Dawley rats were studied with approval from the Institutional Animal Care and Use Committee (N=7). Pass-band bSSFP and GRE scans were conducted using 9.4T Varian animal scanner (Palo Alto, CA). The common scan parameters were matrix size=192×256 (phase-encoding×readout), TR/TE=20/10ms, single slice with thickness=2mm, field of view=2.4×2.4cm². Only for bSSFP, additional scans of TR/TE=10/5 ms were performed. The flip angles were set at 16˚ in bSSFP and 8˚ in GRE acquisition. In bSSFP study, 4 different phase cycling angles of 0˚, 90˚, 180˚, and 270˚ (denoted as PC0, PC1, PC2, and PC3, respectively) were applied to get separate data for both TR/TE=20ms/10ms and 10ms/5ms. Acquisition of the same slice was repeated 24 times (TR/TE=20 ms/10ms) or 48 times (TR/TE=10/5ms), maintaining the same scan time of 92s for each of the 9 sequences (1 GRE and 8 bSSFP at 2 TR values and four PC angles). The 9 scans composed one full dataset and acquisition of datasets was repeated 15 to 25 times varying depending on subjects. Eventually all the data from within each scan and across different datasets were averaged for each of the 9 sequences.

To obtain the image in isotropic inplane resolution of 93.75μm, the raw dataset was sequentially zero-padded and Fourier-transformed. The large-scale phase variation was mitigated by applying a high-pass filter (generated by a 2D Gaussian filter with FWHM of 60 voxels) to the original data.

The phase images from the 9 sequences were compared in terms of phase contrast between white matter (WM) and gray matter (GM) by calculating the average of phase values within each region of interest (ROI). The ROIs of WM and GM were manually set in corpus callosum and in cortical cortex respectively (Fig. 1).

Figure 2 shows the quantitative ROI analysis of phase values from the bSSFP and GRE scans. In bSSFP, the phase values of WM varied with respect to PC angle, while the phase values of GM changed little with PC angle. The WM to GM phase ratio was the highest in PC3 and was twice as high as that in GRE, indicating phase images of pass-band bSSFP with a specific PC angle can present WM better than the phase image of conventional GRE method.

Figure 3 shows the phase images of bSSFP acquisition in 8 different conditions and 1 GRE acquisition. The qualitative comparison between bSSFP phase images also demonstrated that the phase contrast varied significantly with PC angles (Fig. 3a). Interestingly, bSSFP PC3 in 20ms of TR phase image offered a noticeable visibility of the WM tissues and small structures presumed to be fiber bundles, such as corpus callosum, anterior commissure, and the neural tracts of striatum (Fig. 3a). Vascular structures were mostly visible as dark signals (i.e., negative phase values), whereas WM regions and the small structures presumed to be neuronal fiber bundles were mostly presented as bright signals (i.e., positive phase values).

In ultrahigh field, the high susceptibility effects induce remarkable phase contrast between tissues and small microstructures. This makes bSSFP promising to be conducted in high field. In addition, bSSFP showed the capability of phase imaging for advanced anatomy through the multiple phase cycling approach, which enables the inspection of complex brain microstructures. Anterior commissure has been easily observed by the conventional MRI imaging techniques, since the WM tissues exist in bundles of nerve fiber. However, substructures of striatum have been difficult to be detected in routine MRI protocol. From this reason, previous MRI studies were limited to investigating the severity of striatal damage by staining (6) or utilizing magnetic resonance spectroscopy (7).The phase images of pass-band bSSFP can provide advanced anatomical information by using multiple phase cycling at ultrahigh field. The phase image of bSSFP can visualize the substructure of striatum, which cannot be imaged by conventional methods. bSSFP will provide a new insight into phase imaging for anatomical information.