Although the specificity of T1 and T2 measurements are generally perceived as being poor, several investigators have recently shown that the T2 signal decay in neural tissue is multi-exponential with echo time (Menon and Allen 1991; MacKay et al. 1994; Whittall et al. 1997). Further investigation has shown that different water tissue compartments each have distinct T2 characteristics, and may be separated (see Fig. 24). In white matter, the water signal compartments are believed to originate from components of free water (e.g., edema, CSF, which have long T2 > 120ms), extracellular water (T2 ^ 60-90 ms) and water within the myelin membranes of axons (T2 ~ 10-40 ms) (MacKay et al. 1994; Beaulieu et al. 1998; Stanisz and Henkelman 1998; Vavasour et al. 1998; Laule et al. 2004). The T2 of the extracellular fraction can be used to identify inflamed neural tissues (Stanisz et al. 2004), and the latter component is of significant interest because it is specific to myelin, which is critical for signal conduction in the brain. Consequentially, a potentially important biomarker is the myelin water fraction, which is the total signal
from the short T2 signal component relative to the total signal from both the short and intermediate tissue signal components. In healthy adult WM, the myelin water fraction (MWF) is typically 6-15% dependent upon the region (Whittall et al. 1997). A representative map of MWF is shown in Fig. 25.
Measurements of MWF are usually obtained using a 2D multiple spin echo sequence, which consists of a train of equally spaced 180° refocusing pulses (Poon and Henkelman 1992; Poon and Henkelman 1995; Whittall et al. 1997). T2 measurements are highly sensitive to errors in the RF magnetic field, which are problematic for typical slice-selective RF refocusing pulses. Consequently, non-selective refocusing pulses are often used, which limits the acquisition to a single 2D slice. Variable amplitude crusher gradient pulses are typically placed around each refocusing pulse to suppress the signal from stimulated echoes. The fitting of the T2 model is also highly sensitive to image noise; consequently, long scan times are typically required to achieve sufficient SNR. Different strategies exist for fitting the T2 signal decay to a multi-exponential function (e.g., Stanisz and Henkelman 1998; Webb et al. 2003; Jones et al. 2004) although the non-negative least squares (NNLS) method is probably most commonly used (Whittall et al. 1997). The slow acquisition time (typically > 10 minutes) for a single 2D slice has ultimately limited the application of this approach. However, one consideration is that the 2D imaging times are in line with MR spectroscopy. Further, the MWF is one of the more specific measures of white matter tissue properties, which makes it promising for correlations with measures of brain connectivity. Careful selection of echos in conventional pulse sequences may provide reasonable myelin maps (Vidarsson et al. 2005; Oh et al. 2006), although the option to acquire such data is not available routinely on most clinical scanners. Future
Fig. 25. Maps from a myelin water fraction experiment. The image on the left is a proton-density weighted image obtained from the first TE (8 ms) in the CPMG echo train. The map on the right is the estimated myelin water fraction image at the same slice location. Note that the myelin water fraction is much higher in regions of white matter
Fig. 25. Maps from a myelin water fraction experiment. The image on the left is a proton-density weighted image obtained from the first TE (8 ms) in the CPMG echo train. The map on the right is the estimated myelin water fraction image at the same slice location. Note that the myelin water fraction is much higher in regions of white matter developments are clearly needed to improve both the acquisition speed and spatial coverage of the technique, which are somewhat at odds with one another. Imaging at higher magnetic field strengths, with better RF coils, parallel imaging and 3D pulse sequences may ultimately improve the utility of the method.
To date, no studies have been performed which have related MWF measurements to measures of brain connectivity. However, MWF measurements in WM have been shown to be affected in brain diseases with aberrant brain connectivity behavior including schizophrenia (Flynn et al. 2003) and multiple sclerosis (Vavasour et al. 1998; Gareau et al. 2000; Whittall et al. 2002; Laule et al. 2004; Tozer et al. 2005).
Was this article helpful?