Aspects on the accuracy of cerebral perfusion parameters obtained by dynamic susceptibility contrast MRI: a simulation study

Abstract

Several studies have indicated that deconvolution based on singular value decomposition (SVD) is a robust concept for retrieval of cerebral blood flow in dynamic susceptibility contrast (DSC) MRI. However, the behavior of the technique under typical experimental conditions has not been completely investigated. In the present study, cerebral perfusion was simulated using different temporal resolutions, different signal-to-noise ratios (S/Ns), different shapes of the arterial input function (AIF), different signal drops, and different cut-off levels in the SVD deconvolution. Using Zierler's area-to-height relationship in combination with the central volume theorem, calculations of regional cerebral blood volume (rCBV), regional cerebral blood flow (rCBF), and regional mean transit time (rMTT) were accomplished, based on simulated DSC-MRI signal curves corresponding to artery, gray matter (GM), white matter (WM), and ischemic tissue. Gaussian noise was added to the noise-free signal curves to generate different S/Ns. We studied image time intervals of 0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 s, as well as different degrees of signal decrease. The singular-value threshold in the SVD procedure and the shape of the AIF were also varied. Increased rCBF was seen when noise was added, especially for rCBF in WM at the larger image time intervals. The rCBF showed large standard deviations using a low threshold value. A prolonged time interval led to a lower absolute value of rCBF both in GM and WM, and a low/broad AIF also underestimated the rCBF. When a larger maximal signal decrease was assumed, smaller standard deviations were observed. No systematic change of the average rCBV was observed with increasing noise or with increasing image time interval. At S/N = 40, a low cut-off value resulted in an rCBF that was closer to the true value. Furthermore, at low S/N it was difficult to differentiate ischemic tissue from WM.

Introduction

Dynamic susceptibility contrast (DSC) MRI can potentially be used for the determination of cerebral perfusion parameters such as regional cerebral blood volume (rCBV), regional cerebral blood flow (rCBF), and regional mean transit time (rMTT), using the theory of intravascular tracers [1]. High spatial resolution, no patient exposure to ionizing radiation, and the possibility to combine morphological and functional information during one single imaging session make the method an attractive concept in a clinical environment. However, the procedure is not completely straightforward; numerous factors come into play in the calculation of cerebral perfusion. The deconvolution algorithm, the location and registration of the arterial input function (AIF), the temporal resolution, different signal decrease, and signal-to-noise ratio (S/N) all influence the quantification of the cerebral perfusion.

Several authors have discussed the importance of an appropriate deconvolution algorithm to be used in DSC MRI [2], [3], [4], [5], [6], [7], and, currently, the most common technique is based on singular value decomposition (SVD) [3]. The SVD deconvolution algorithm is a nonparametric technique that allows the estimation of perfusion parameters even at low S/Ns. The technique is also fairly independent of underlying vascular structure and volume. By eliminating the singular values in the diagonal matrix (often denoted W) that are smaller than a certain threshold, i.e., by choosing a cut-off value, the influence from noise is reduced [3]. Liu et al. discussed the importance of an appropriate cut-off value [8] and presented functions for choosing the optimal threshold based on the S/N of the concentration time curve. Calamante et al. [9] recently presented simulations of the effects of delay and dispersion in the measured AIF when using SVD deconvolution and found that a significant underestimation of rCBF and overestimation of rMTT were obtained when the two effects were introduced. Furthermore, Sorensen and Reimer stated that the preferred temporal resolution is between 0.8 and 2.2 s and that the optimal temporal resolution is 1 s using an SE EPI sequence [10]. Another important parameter is the echo time (TE) due to its effect on the baseline signal level and the signal decrease. An optimum signal drop is obtained when TE equals T₂*, i.e., when the ratio of minimum signal S_min to baseline signal S₀ equals 1/e, but a larger decrease leads to a decrease in quality [11].

Experimental optimization investigations have been performed [12] as have theoretical studies of the relationships among different parameters by computer simulations [13], [14], [15]. The results of our simulations constitute a relevant supplement to these, since practical perfusion measurements by use of DSC MRI are carried out with very different pulse sequences and imaging protocols, leading to difficulties in the comparison of results from different studies. Furthermore, quantification of rCBF by DSC MRI is likely to become a powerful tool, for example, in tumor therapy monitoring and for determining tissue at risk in acute ischemic stroke, provided that the technique can yield reproducible absolute values with a well-defined accuracy and precision.

In this study, quantification of cerebral perfusion parameters in GM, WM, and ischemic penumbra under various conditions was simulated using the SVD deconvolution technique. The parameters that were varied in the present simulation study were the temporal resolution, the maximal signal drop, the S/N, the shape of the AIF and the W-matrix cut-off value in the SVD deconvolution.