Dynamic Contrast Enhanced Imaging

Though T1 measurements may be made in a matter of seconds (Freeman et al. 1994; Tong and Prato 1994), temporal or spatial constraints usually preclude their use for tracking the passage of a contrast agent bolus (especially for 3D measurements). As discussed above, a common experimental approach is to measure T1 before contrast agent administration then image the tissue rapidly during uptake using a fast Trweighted sequence (Brookes et al. 1999). The pre-contrast T1 measure provides estimates of T10 and the lumped constant, g.p.exp(-TE/T2*) (Eq. 5.3). Thus, by substitution, T following contrast agent administration may be estimated directly from signal intensity measurements (Zhu et al. 2000). To reduce any error associated with RF miscalibration a final, bookend, T measurement may also be made (Cron et al. 1999). Selection of an imaging methodology begins with a series of basic choices such as an appropriate RF coil, imaging plane and sequence to avoid issues such as flow artefacts and field of view aliasing. Subsequently, the choice of Trweighted sequence for bolus tracking must fulfil a long list of both generic and study-specific criteria and a wide range of methods have been used in the field (see many of the other chapters in this book). High in the list of generic criteria are: temporal resolution, Tj sensitivity and dynamic range, spatial coverage, and resolution; each of these competes against noise for the limited MR signal. Perhaps the principal decision to be made when selecting a Trweighted sequence for bolus tracking is whether or not an arterial input function (AIF) will be measured (see Chap. 6).

It has been shown that accurate characterisation of the AIF requires a temporal resolution in the order of a second (Henderson et al. 1998). Furthermore, following bolus injection of a typical clinical dose of contrast agent the T1 of the blood may decrease by more than an order of magnitude (Fritz-Hansen et al. 1996). Monitoring such large changes in relaxation rate requires an imaging sequence with a good dynamic range (Fig. 5.2). Competing directly with this requirement is the need to monitor much smaller changes in T1 at the level of the tissue. The location of the AIF, in relation to the tissue of interest, and the extent of that tissue dictates the requirements for spatial coverage. Finally, it is rare to identify a local feeding artery to provide an AIF but the closer the AIF is to the true tissue arterial supply the more accurate the subsequent modelling (Calamante et al. 2000). However, the spatial resolution of the images places a minimum diameter on the artery to be imaged. It is often the case that measurement of the AIF proves to be impossible or inappropriate. Typically this may be due to difficulties in choosing an appropriate artery (e.g. in studies of breast cancer) or the necessity to use an imaging sequence that lacks the necessary temporal resolution or fails to saturate incoming arterial water and thereby makes AIF estimation impossible (Fritz-Hansen et al. 1996). Most MR studies to date have been performed without measurement of

100 150

Fig. 5.2. To measure both the signal in the artery and tissue an imaging sequence selected for a bolus tracking experiment requires a large dynamic range. In this example mean values of the raw signal intensity from regions of interest placed in the external iliac artery (filled squares) and in muscle (internal obturator; open squares) are shown on a semi-logarithmic plot. Note the order of magnitude difference in the level of the signals obtained from each region

100 150

Fig. 5.2. To measure both the signal in the artery and tissue an imaging sequence selected for a bolus tracking experiment requires a large dynamic range. In this example mean values of the raw signal intensity from regions of interest placed in the external iliac artery (filled squares) and in muscle (internal obturator; open squares) are shown on a semi-logarithmic plot. Note the order of magnitude difference in the level of the signals obtained from each region the AIF with either a population averaged AIF used in the subsequent data analysis or the AIF has been neglected altogether (see Chap. 6). In such studies the constraints on the required temporal resolution are relaxed and the imaging sequence used may be more T sensitive with a reduced dynamic range. This provides greater opportunity for improving the spatial coverage, resolution or signal to noise ratio of the images acquired.

Finally, with the images acquired and the signal changes converted to changes in contrast agent concentration, quantitative analysis may proceed. Measurement of contrast agent concentration changes in the artery (providing the AIF) are typically made using user-defined regions or automated vessel identification procedures (Rijpkema et al. 2001; Parker et al. 2003). However, the selection of appropriate tissue for analysis remains the subject of considerable debate. Traditionally, a region of interest is defined in, for example, tumour tissue. Once analysed, the characteristics of this region are estimated but are these characteristics, particularly in a heterogeneous cancer, representative of the tumour as a whole? Much research has gone into comparing the results of whole tumour region definition versus the selection of small, highly enhancing, sub-regions or semi-automated region selection (Mussurakis et al. 1997) (Mussurakis et al. 1998). Furthermore, images can be analysed on a per-pixel basis providing tissue characterisation at significantly improved spatial resolutions. Nevertheless, how are the parameters estimated in each pixel combined to provide simple representative values for the tissue as a whole? Mean or median values may not describe the range of characteristics observed. Histogram analysis of the measured uptake parameters present one route for describing the heterogeneity observed (Hayes et al. 2002; Checkley et al. 2003) but a consensus on the best approach for describing heterogeneous contrast enhancement remains the subject of continuing research.

0 0

Post a comment