MRI Sequence Type

As discussed throughout the chapter, bolus tracking MRI can be performed using different types of MR sequences. The selection of the optimal sequence depends on a compromise between the different (sometimes competing) factors, as well as the particular application, and the tissue under study. This section discusses some of the main issues associated with the selection of the MR sequence for bolus tracking MRI, and their effects on quantification of perfusion

Gradient-Echo vs. Spin-Echo

As discussed in Sect. 4.1, bolus-tracking MRI can be performed using either gradient-echo (through the T2* effect) or spin-echo (through the T2 effect). Although both MR sequences are sensitive to the susceptibility gradients induced by the contrast agent, it has been shown that gradient-echo is sensitive to the total vas-culature, while spin-echo is sensitive primarily to the microvasculature (Weisskoff et al. 1994b; BoxermaN et al. 1995). Therefore, the question arises as to the best sequence type for the investigation of perfusion in oncology. Although it was originally thought that the microvascular sensitivity of spin-echo sequences was favourable (since microvascular density has been used as the standard marker to predict tumour angiogen-esis), recent studies have suggested that gradient-echo approaches are better (Donahue et al. 2000; Sugahar a et al. 2001). This may not be totally unexpected since large, tortuous vessels are often present in tumours. Sugahara et al. (2001), using successive gradient-echo and spin-echo sequences (acquired in random order), detected significant differences between high-grade and low-grade gliomas only with the gradient-echo technique. Similarly, Donahue et al. (2000), using a sequence to simultaneously acquire gradient-echo and spin-echo images, found correlation between CBV and tumour grade only for the CBV calculated from gradient-echo images. The simultaneous acquisition of gradient-echo and spin-echo images also allows the calcu-

a As previously mentioned, the gamma-variate function is not always an accurate model for the first pass of the tracer. Since the quantification of rR relies on the identification of the first pass by using a gamma-variate fit, this might lead to potential errors (Kassner et al. 2000). b Since part of the recirculation phase represents true contrast recirculation, the area under C(t) in the recirculation phase is expected to correlate with CBV [see Eq. (3)].

Fig. 4.9a-f. a Example of post-contrast CBV, K2, and ratio maps from a patient with a malignant lymphoma (large arrow) and venous angioma (small arrow). While the post contrast Tj-weighted image (a) demonstrates signal enhancement in the lymphoma region, neither the gradient-echo CBV (b), spin-echo CBV (e), or ratio maps (d) show increases relative to surrounding brain. However, permeability factor (K2) increases are apparent on the gradient-echo K2 map (c) but not on the spin-echo K2 map (f). The venous angioma does demonstrate increases on the gradient-echo CBV and ratio maps, a finding consistent with a larger blood volume and increased average vessel diameter in this region. [Image kindly provided by Dr. K. Donahue-Schmainda, and previously published in Donahue et al. (2000)]

Fig. 4.9a-f. a Example of post-contrast CBV, K2, and ratio maps from a patient with a malignant lymphoma (large arrow) and venous angioma (small arrow). While the post contrast Tj-weighted image (a) demonstrates signal enhancement in the lymphoma region, neither the gradient-echo CBV (b), spin-echo CBV (e), or ratio maps (d) show increases relative to surrounding brain. However, permeability factor (K2) increases are apparent on the gradient-echo K2 map (c) but not on the spin-echo K2 map (f). The venous angioma does demonstrate increases on the gradient-echo CBV and ratio maps, a finding consistent with a larger blood volume and increased average vessel diameter in this region. [Image kindly provided by Dr. K. Donahue-Schmainda, and previously published in Donahue et al. (2000)]

lation of the maps of the ratio "AR27AR2" (see Fig. 4.9), which has been suggested as a potential marker of average vessel diameter due to the different sensitivity to vessel size of the images (Dennie et al. 1998; Donahue et al. 2000). This parameter was found to be strongly correlated to brain tumour grade (Donahue et al. 2000).

Spatial Coverage vs. Time Resolution

Since a proper characterisation of the transient signal drop is required in order to accurately quantify DSC-MRI data, the time resolution is limited by the transit time of the bolus through the tissue. Typically, a repetition time (TR) of 2 s or less is necessary to adequately sample the fast changes during the first pass. Therefore, the spatial coverage will be limited by the maximum number of slices that can be acquired within this TR. The use of EPI sequences typically allows 10-15 slices to be acquired during this time. However, as mentioned in Sect. 4.2, EPI suffers from large image distortions in areas of interface between different susceptibility properties (e.g. tissue-air interfaces), and it is not suitable for imaging in many organs. Therefore, other imaging modalities (e.g. segmented EPI, FLASH, etc.) are sometimes used. In these cases, the 1.5-2 s TR typically allows only a couple of slices to be acquired. In many cases (see for example Rempp et al. 1994; Schreiber et al. 1998; Vonken et al. 1999), one of these slices is positioned through a major artery to estimate the AIF, leaving only one slice for measuring perfusion in the tissue of interest. In summary, the selection of the spatial coverage and time resolution must usually be done as a compromise, according to the particular application and tissue under study.

2D vs. 3D Sequences

The use of 3D sequences has been reported in studies of bolus tracking (see for example van Gelderen et al. 2000 and Kassner et al. 2000). They allow increased coverage, although typically at the expense of an increased TR, or a decreased spatial resolution. The increased coverage is necessary in cases where the location of the abnormality is not known a priori, or when the information from multiple areas is required. However, the time of acquisition is not well defined for each image in 3D MRI (cf. the time of acquisition of each slice in multi-slice 2D MRI), and this effect will introduce a smoothing of the time changes during the first pass (due to the signal intensity changes during the MR volume acquisition). If the time resolution is of the order of several seconds, this can lead to an erroneous characterisation of the passage of the bolus. Therefore, 2D sequences are usually preferable, since this effect generally can be neglected.

Single-Echo vs. Dual-Echo Sequences

As discussed in Sect. 4.2.2, the use of dual-echo sequences allows the quantification of changes in R2* without assumption regarding the T behaviour (Miyati et al. 1997; Barbier et al. 1999). This method provides a simple way to remove the relaxivity effects. However, the sampling of two echoes introduces considerable demands on the sampling time, reducing the maximum number of slices that can be acquired (and therefore, coverage) in a given TR. Since the use of a pre-enhancement approach has been shown to be very effective in many situations (Kassner et al. 2000), single-echo sequences with a pre-dose of contrast may be advantageous when spatial coverage is important.

0 0

Post a comment