Time

Prog #

Egbert Jan Willem Bleeker¹, M. A. van Buchem¹, M. J.P. van Osch¹

¹Leiden University Medical Center, Leiden, Netherlands

Obtaining quantitative values for perfusion from DSC-MRI requires a correct AIF for deconvolution. Commonly the AIF is obtained in the vicinity of the MCA using the δR₂^* derived from magnitude data, but δ φ, derived from the phase of the MR signal, proves to be a suitable candidate as well. This study provides the optimal locations for AIF measurements based on the phase signal by numerical simulations validated by phantom experiments.

Peter Gall¹, Esben Thade Petersen², Xavier Golay², Valerij Kiselev¹

¹University Hospital Freiburg, Freiburg, Germany; ²National Neuroscience Institute, Singapore, Singapore

CBF estimates are blurred by delay and dispersion of a bolus from an artery to the tissue. A cylinder with laminar flow conditions can approximate a blood vessel. Obeying a scaling rule, a tree composed of such cylinders can represent the vasculature. For each path within the tree a vascular transfer function can then be derived. ASL techniques are capable of measuring the transport function that describes the trespass trough major cerebral arteries. The theoretical vasculature model for the path evaluated in the ASL measurement in a volunteer can predict the measured transport function.

Greg Zaharchuk¹, Roland Bammer¹, Matus Straka¹, Rexford D. Newbould¹, Jean-Marc Olivot¹, Michael Mlynash¹, Maarten G. Lansberg¹, Greg W. Albers¹, Michael E. Moseley¹

¹Stanford University, Stanford, California , USA

Reasonable relative CBF images that correlate well with gold standard techniques such as O-15 water PET can be created from PWI bolus tracking experiments.  A large problem is that the global CBF scaling factor varies significantly from patient to patient, such that absolute PWI-based CBF measurements are not reliable.   This study examines whether corrections for AIF partial voluming and nonlinear bulk blood contrast relaxivity in single shot GE EPI PWI lead to improved CBF quantitation, by comparing with a gold standard xeCT method in patients with cerebrovascular disease.

Doerthe Ziegelitz¹, Linda Knutsson², Göran Starck³, Stig Holtås², Mats Tullberg³

¹Sahlgrenska University Hospital, Gothenburg, Sweden; ²Lund University Hospital, Sweden; ³Sahlgrenska University Hospital, Sweden

To improve the reproducibility of AIF registration and absolute CBF quantification in DSC-MRI at 1.5T, we rescaled the time integral by use of a venous output function. CBF estimates of 21 healthy, elderly volunteers, obtained by CT-perfusion and DSC-MRI on 2 consecutive days, were compared. DSC-MRI, calculated without the AIF time-integral correction, did not result in any significant correlation with CT-perfusion. The rescaled DSC-MRI showed fair to moderate correlation with CT-perfusion for the central grey matter and the whole brain. Our results indicate that the used method for correction of partial volume effects in DSC-MRI experiments is effective at 1.5T.

Kyrre Eeg Emblem¹, ², Baard Nedregaard¹, John K. Hald¹, David Scheie¹, Atle Bjornerud¹, ²

¹Rikshospitalet, Oslo, Norway; ²University of Oslo, Oslo, Norway

Studies have shown that cerebral blood volume (CBV) maps derived from MR perfusion can improve glioma grading. A problem with any MR perfusion based glioma grading method is the need to exclude large vessels infiltrating the tumor region in order to avoid false positives due to the artificial CBV increase. We propose a novel vessel segmentation technique which uses several properties of the DSC response to automatically segment and remove macrovascular structures from CBV maps. Preliminary results suggest that this technique improves the diagnostic efficacy of glioma grading, especially for gliomas located in the middle cerebral artery (MCA) regions.

Yeng-Peng Liao¹, Ing-Tsung Hsiao, Ho-Ling Liu

¹ Chang Gung University, Tao-Yuan, Taiwan

This study is aimed to evaluate the influence of AIF shape and onset time in estimating the physiological parameters from T1-weighted DCE-MRI kinetic analysis in a reference region model (RRM), and compare its results to that from a simple Kety model (SKM) with sampled AIF.  Conditions combining different SNRs, onset times, injection durations were simulated.  The mean errors and coefficient of variation (CoV) of K^trans,TOI were calculated for both methods.  The results show that with the availability of stable K^trans,RR , the RRM produces stable and smaller mean errors of K^trans,TOI, and similar CoV as compared to those from SKM.

Hai-Ling Margaret Cheng¹, ²

¹The Hospital for Sick Children, Toronto, Canada; ²University of Toronto, Toronto, Canada

DCE-MRI quantification of pharmacokinetic parameters is subject to various sources of measurement errors and physiological unknowns. One unknown that has not been studied in DCE-MRI is the influence of a finite capillary transit time. Simulations show that longer transit times give rise to significant error in parameter estimates. A modification to the Tofts model is presented and is shown to substantially improve parameter accuracy in the presence of both transit time and bolus arrival delay. Accuracy is maintained in the transfer constant (K^trans) and interstitial space (v_e), and underestimation of the plasma volume (v_p) is greatly reduced.

Michael Gerhard Kaul¹, Gerhard Adam¹

¹University Medical Center Hamburg-Eppendorf, Hamburg, Germany

For pharmacokinetic analysis dynamic contrast enhanced (DCE)-MRI data is transformed from a signal enhancement to a Gadolinium related concentration scale. This work aims to minimize the error of ignoring the influence of T2* relaxation in spoiled T1-weighted gradient sequences without measuring T2*. This is accomplished by an easy to implement iterative algorithm. Its effectiveness is proved by simulated data for a first pass of Gd-DTPA in blood and tissue. Within several iteration steps a sufficient correction is achieved.

Edward Ashton¹

¹VirtualScopics, Inc., Rochester, New York, USA

It is commonly assumed that tracking of changes in vascular parameters using DCE-MRI requires conversion of the signal changes seen in tissues to Gd concentration.  Recent studies have indicated that such conversion may be unnecessary using certain protocols, and may in fact increase measurement noise.  Questions have also been raised regarding the advisability of measuring an arterial input function (AIF) from DCE-MRI data as opposed to assuming a single generalized population AIF.  In this work we have assessed the impact of each of these decisions on the statistical confidence in study outcome using data from a Phase I clinical study.

C. Chad Quarles¹, John C. Gore¹, Thomas E. Yankeelov¹

¹Vanderbilt University, Nashville, Tennessee, USA

We recently proposed a dual echo DSC-MRI approach that removes EES T₁ and T₂* extravasation effects and enables the computation of R1 time curves. In this study we demonstrate that CBF and CBV measurements corrected for EES T₁ and T₂* leakage effects strongly correlate to those derived from a DSC study using an intravascular contrast agent.  This correlation was much greater than that derived from uncorrected and T₁ only corrected measurements. We also show that DSC and standard DCE measured R1 time curves are very similar and yield a high correlation between the derived K^trans values.