Introduction

In MR, tissue with anisotropic micro- or molecular structure may yield signal variations (i.e. signal anisotropy) that depend on the direction of a magnetic field applied to the tissue. One of the well-known signal anisotropy is a diffusion signal in diffusion tensor imaging measured in white matter of the brain (1). In this example, a voxel with white matter fibers, which have anisotropic microstructure, generates signal variations that depend on the relative orientation between the primary direction of the fibers and the applied diffusion gradient.

Signal anisotropy has been observed in several MRI contrasts applied in various structures in the body (2). Magnetic susceptibility is another well-explored source for the signal anisotropy. For example, a long cylinder that has magnetic susceptibility different from a surrounding medium generates magnetic field both within and outside of the cylinder that changes by the relative angle between the long axis of the cylinder and B0 field (3,4). If a voxel contains the cylinder, the gradient echo signals (both magnitude and phase) of the voxel will change as a function of the angle. Hence, a biological structure (e.g. vein) that has a preferential orientation with magnetic susceptibility different from surrounding tissue (e.g. paramagnetic deoxy-hemoglobin vs. diamagnetic parenchyma) will demonstrate signal anisotropy in gradient echo measurements. Similarly to the vein, other cylindrical structures such as axon, skeletal muscle fiber, myofiber exhibit similar or more complex signal anisotropy. In this talk, I will explain the origins of the signal anisotropy in white matter, heart and muscle.

Origins of Signal Anisotropy

[White Matter] Different from a simple cylindrical shape of a vein, a white matter fiber can be modeled as a hollow cylinder that divides water space into axonal (innermost cylinder), myelin (annulus), and extracellular water space (outside of hollow cylinder) (5,6). The magnetic susceptibility source is located in the annulus where diamagnetic myelin exists. This structural anisotropy with magnetic susceptibility of myelin induces a unique magnetic field pattern that changes with the relatively orientation of the cylinder and B0 field (5-7). Interestingly, the signal anisotropy of white matter is more complex than that from the simple hollow cylinder because of the magnetic susceptibility anisotropy of myelin (i.e. magnetic susceptibility of myelin changes its value depending on the direction of the B0 field relatively to the orientation of lipid bilayers in myelin (8,9)). Additionally, a shorter T2 or T2* in myelin water as compared to axonal and extracellular water further complicates signal evolutions (10,11). When all of these factors are combined, the magnitude and phase of each water pool show unique signal characteristics that various as a function of the angle between the fiber and B0 field. The resulting GRE magnitude and phase signals from a voxel that encloses the myelinated fibers contain rich information of such sources (6,12). It has been demonstrated that some of the information (e.g. myelin susceptibility value and myelin water volume) can be extracted from the signal anisotropy measurements (6,9,13).

[Heart and Muscle] Similarly to the myelin in white matter, muscle, both skeletal muscle (14) and myocardium (15), has been demonstrated to possess susceptibility anisotropy. The muscle contains an alpha-helix peptide bond structure, which is diamagnetic and has magnetic susceptibility anisotropy. This anisotropy creates orientation dependent magnitude and phase evolution and has been exploited to generate fiber orientation mapping of the heart (15).

Applications and Complications

In white matter of the brain, the magnetic susceptibility induced signal anisotropy has been utilized for fiber orientation mapping (e.g. susceptibility tensor imaging using phase data (9), T2* orientation mapping using magnitude data (16), and frequency difference mapping using phase data (17)). It also helped to visualize structures outside of the brain (e.g. myofiber orientation in the heart (15) and nephron structure in the kidney (18)). These methods, however, require multiple data acquisition at different object (e.g. head) positions and, therefore, may have limited applications (e.g. high-resolution fiber orientation mapping for ex-vivo samples). Recently, novel approaches that combine the anisotropy measurement (e.g. T2* map) at a single orientation and the fiber orientation map from DTI have been proposed to generate orientation dependent anisotropy information (19,20). These approaches do not require multiple orientation data and still utilize the signal anisotropy information and, therefore, may have potentials for in-vivo clinical applications.

In addition to the direct applications of the anisotropy, the knowledge gained in understanding the signal anisotropy has been utilized. One example is myelin water imaging using gradient echo data (10,13). The conventional gradient echo myelin water imaging exploited T2* difference between myelin water pool and axonal/extracellular water pool and had limitation in accuracy and reliability due to proximity of T2* values between the pools (10). The phase characteristics revealed by the signal anisotropy studies of white matter have suggested distinct phase features of the myelin water pool (5,6) and has helped to improve accuracy and reliability of the method (13).

When the signal anisotropy is not properly modeled in susceptibility mapping, it becomes a source for errors in estimating susceptibility values in quantitative susceptibility mapping (ignoring susceptibility anisotropy and microstructural anisotropy) and susceptibility tensor imaging (ignoring microstructural anisotropy) (12). These effects may influence the reproducibility of the methods.