Magnetic Susceptibility and B0 Inhomogeneity

Susceptibility differences between diamagnetic tissue water and paramagnetic material cause local magnetic field changes. Such changes occur on a subvoxel spatial scale, for example with red blood cells (a 5 mm), ferritin (20 nm), and MRI contrast agents, and present an important contrast mechanism (see below). Conversely, susceptibility inhomogeneity on a spatial scale of several voxels (1-10 mm) near air/ tissue interfaces produces severe image artifacts, including geometric distortions and signal loss. Susceptibility field inhomogeneity AB0(x, y, z) can be described analytically for simple geometries, whether numerically modeled (Li et al., 1996; Truong et al., 2002), or experimentally measured (Spielman et al., 1998; Truong

Fig. 3.58. (A) Numerically computed saggital susceptibility field map in human brain. (Truong et al., 2002). (B) Gradient echo image of the same subject using TR/TE 600 ms/ 12 ms, nominal flip 22.5°, FOV 18 cm, 1024 x 512, 2 mm slice thickness. (C) Matching flip angle map and (D) receive sensitivity map,

Fig. 3.58. (A) Numerically computed saggital susceptibility field map in human brain. (Truong et al., 2002). (B) Gradient echo image of the same subject using TR/TE 600 ms/ 12 ms, nominal flip 22.5°, FOV 18 cm, 1024 x 512, 2 mm slice thickness. (C) Matching flip angle map and (D) receive sensitivity map, computed from axial gradient echo images with nominal flip angles of 60°/120° using the relationship S2a/Sa = sin 2a/sin a = 2 cos a. Note that in the ventricles, the receive sensitivity map is contaminated by proton density (Truong et al., 2006).

computed from axial gradient echo images with nominal flip angles of 60°/120° using the relationship S2a/Sa = sin 2a/sin a = 2 cos a. Note that in the ventricles, the receive sensitivity map is contaminated by proton density (Truong et al., 2006).

et al., 2006). For ultrahigh field MRI, B0 inhomogeneity is especially problematic because field changes scale linearly with the magnetic field strength. For example, the numerical simulation in Figure 3.58A shows variability of 2 ppm near the sphenoid. Under typical image conditions, resultant image distortions are less then one pixel at 1.5 Tand thus negligible, whereas they extend over several voxels at 8 T. Misregistration artifacts occur in both spin and gradient echoes in the frequency and slice encode direction. In addition, susceptibility inhomogeneity creates severe signal loss in gradient echo images, where the size of the signal void increases with increasing echo time and field strength. For example, areas of signal voids are seen near the sphenoid sinus with banding patterns extending up to the corpus callosum and in the cerebrospinal fluid (CSF) spaces anterior to the brainstem (Fig. 3.58B). Consequently, susceptibility artifact correction is of paramount importance for ultrahigh field MRI. Several approaches have been proposed for this purpose, including gradient compensation (Yang et al., 1998; Glover, 1999), tailored radiofrequency (RF) pulses (Cho and Ro, 1992; Stenger et al., 2002), active and pas sive shimming (Wilson et al., 2002), and post-processing (Irarrazabal et al., 1996; Kadah and Hu, 1998).

Was this article helpful?

0 0

Post a comment