| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
infoRAD |
1 From the Cancer Research UK Clinical MR Research Group, Institute of Cancer Research and Royal Marsden NHS Foundation Trust, Downs Road, Sutton, Surrey, United Kingdom SM2 5PT (J.A.D., D.J.C., S.W.S., M.O.L.); the Department of Radiology, Mount Vernon Hospital, North-wood, Middlesex, England (A.R.P.); and the Department of Psychiatry, University of Cambridge, Addenbrooke’s Hospital, Cambridge, England (J.S.). Presented as an infoRAD exhibit at the 2003 RSNA Annual Meeting. Received July 19, 2005; revision requested October 13 and received November 11; accepted December 6. J.A.D., D.J.C., and M.O.L. may benefit financially from the licensing and commercialization of the MRIW software; A.R.P. has used the MRIW software to analyze results of commercially sponsored clinical drug trials; the other authors have no financial relationships to disclose. Supported by grant C1060/A808 from Cancer Research UK. Address correspondence to J.A.D. (e-mail: jamesd{at}icr.ac.uk).
 |
Abstract |
|---|
 Top Abstract IntroductionContrast Agent KineticsContrast Agent ConcentrationFunctional DescriptionConclusionsTAKE-HOME POINTSReferences |
|---|
© RSNA, 2006
|
Introduction |
|---|
|
TopAbstractIntroductionContrast Agent KineticsContrast Agent ConcentrationFunctional DescriptionConclusionsTAKE-HOME POINTSReferences |
|---|
Current methods of assessing angiogenesis can be considered as either direct or indirect. The most frequently used direct method is microvessel density counting after immunostaining with panendothelial cell antibodies (6). Microvessel density counting usually requires tumor tissue from surgical specimens and is unable to provide information on the functional state of the vasculature. More recently, indirect biomarkers of angiogenesis such as the blood level of angiogenic factors and imaging methods have been used. Advantages of indirect methods are that they are non-invasive, can be performed with the tumor in situ, and may be used to monitor response to treatment. Indirect techniques are quantitative, and in the case of imaging, the functional status of the vasculature can be assessed. It is important to note that implanted tumor xenograft data show that there is a discrepancy between perfused and visible microvessels: A variable 20%–85% of microvessels are perfused at any given time; this results in a difference between histologic and functional vascular density (7).
Dynamic contrast-enhanced magnetic resonance (MR) imaging following the administration of low molecular weight contrast media (<1 kDa) is the most popular imaging method for evaluating human tumor vascular function in situ (8). Data reflecting tissue perfusion, microvessel permeability surface area product, and extracellular leakage space can be obtained, depending on the technique used. Insights into these physiologic processes are obtained qualitatively by characterizing kinetic enhancement curves or quantitatively by applying complex compartmental modeling techniques. Advantages of adopting a quantitative approach include derivation of kinetic parameters that are manufacturer and measurement sequence independent, providing pathophysiologic insights into tissue contrast agent kinetic behavior, and the ability to compare parameters acquired serially in the same patient or to compare data obtained from different imaging centers. Therefore, quantitative dynamic contrast-enhanced MR imaging techniques are preferred for antiangiogenesis and antivascular clinical trials (9).
In this article, we present cross-platform software for analysis of dynamic contrast-enhanced MR imaging data that yields functional parameters for tissue and allows exploration of these results. Specific topics discussed are contrast agent kinetics, contrast agent concentration, functional description, and clinical applications.
|
Contrast Agent Kinetics |
|---|
|
TopAbstractIntroductionContrast Agent KineticsContrast Agent ConcentrationFunctional DescriptionConclusionsTAKE-HOME POINTSReferences |
|---|
The "first pass" describes the initial passage of the bolus of contrast medium and lasts for a few cardiac cycles. In most tissues except the brain, testes, and retina, the contrast agent rapidly passes into the extravascular-extracellular space (EES) (also called the leakage space [ve]) at a rate determined by the permeability of the microvessels and their surface area and by blood flow. In tumors, typically 12%–45% of the contrast medium leaks into the EES during the first pass (10). The transfer constant (Ktrans) describes the transendothelial transport of low molecular weight contrast medium from the vascular space to the interstitial space.
Four major factors determine the behavior of low molecular weight contrast media in tissues during the first few minutes after injection. These are the characteristics of the bolus and vascular delivery, the blood perfusion, the transport of contrast agent across vessel walls, and the diffusion of contrast medium in the interstitial space. If the delivery of the contrast medium to a tissue is insufficient (flow-limited situations or where vascular permeability is greater than inflow), then blood perfusion will be the dominant factor determining tissue enhancement and Ktrans approximates to tissue blood flow per unit volume (11); this condition is commonly found in extracranial tumors due to high microvessel permeability. If tissue perfusion is sufficient and transport out of the vasculature does not deplete intravascular contrast medium concentration (non–flow limited or permeability limited), then transport across the vessel wall is the major factor that determines tissue enhancement (Ktrans then approximates to the permeability surface area product). The latter circumstance occurs in areas of radiation fibrosis and in the presence of an intact or partially intact blood-brain barrier but can also occur in extracranial tumors, usually after treatment. The mixed situation occurs most commonly; for low molecular weight gadolinium-containing chelates, there is a tendency for the influence of flow to outweigh that of permeability surface area product in tumors.
As low molecular weight contrast media do not cross cell membranes, the volume of distribution is effectively the EES (ve). Contrast medium also begins to diffuse into tissue compartments further removed from the vasculature, including areas of necrosis and fibrosis. Over a period typically lasting several minutes to an hour, the contrast agent diffuses back into the vasculature (described by the rate constant or kep), from where it is excreted (usually by the kidneys, although some extracellular fluid contrast media have significant hepatic excretion). When capillary permeability is very high, the return of contrast medium is typically rapid, resulting in faster washout as plasma contrast medium concentrations fall.
MR imaging sequences can be designed to be sensitive to the vascular phase of contrast medium delivery (so-called susceptibility-weighted or T2*-weighted dynamic contrast-enhanced MR imaging, which reflects tissue perfusion and blood volume) (12,13). Alternatively, MR imaging sequences can be designed to be sensitive to the presence of contrast medium in the EES and thus reflect microvessel perfusion, permeability, and extracellular leakage space (relaxivity-weighted or T1-weighted dynamic contrast-enhanced MR imaging). These two dynamic contrast-enhanced MR imaging methods are compared in Table 1; both can be analyzed by using the Magnetic Resonance Imaging Workbench (MRIW) software, whose functional description is given later in this article. By using the capabilities of this software, it is possible to become familiar with the steps for quantifying contrast agent kinetics, to understand the pathophysiologic basis of dynamic contrast-enhanced MR imaging, and to appreciate methods of displaying kinetic data.
|
|
Contrast Agent Concentration |
|---|
|
TopAbstractIntroductionContrast Agent KineticsContrast Agent ConcentrationFunctional DescriptionConclusionsTAKE-HOME POINTSReferences |
|---|
Contrast Agent Concentration in Tissue
During measurement of contrast agent uptake, the T1 relaxation rate of a pixel at a given time during the dynamic imaging period is calculated from the ratio of the T1-weighted signal intensity to the proton-density–weighted signal intensity acquired before contrast agent injection. Accurate quantitation of the T1 relaxation times and ultimately the contrast agent concentration requires a prior T1 calibration experiment performed with the proton-density– and T1-weighted measurement sequences. The relationship between the ratio of the proton-density– and T1-weighted signal intensities and the known relaxation times of the test phantoms is usually a single exponential, for the coil and acquisition parameters used.
The concentration, C, of contrast agent in the voxel is calculated from the observed changes in T1 relaxation rate from the precontrast value, T10:
 | (1) |
where R1 is the longitudinal relaxivity of the contrast agent at the field strength of the MR imaging unit.
Vascular Contrast Agent Concentration
In a perfusion measurement, the change in R2* (
R2*) with time caused by the presence of contrast agent can be quantified from a single-echo T2*-weighted dynamic series by using the following formula (16):
 | (2) |
where S0 is the mean preenhancement signal intensity, S(t) is the signal intensity as a function of time, and TE is the echo time.
However, the absolute value of R2* can be calculated from dual-echo data by using the following expression (17):
 | (3) |
Here, R2* is calculated by offsetting R2* by its mean preenhancement value.
The vascular contrast agent concentration, Cv(t), can then be calculated by scaling R2* by the contrast agent relaxivity, r2, in plasma. However, despite being a low molecular weight agent, gadopentetate dimeglumine cannot occupy the entire vascular compartment due to the impermeability of cell membranes of blood cells, so Cv(t) must be scaled by the hematocrit fraction, (1–Hct), giving the blood plasma concentration, Cp(t) (16):
 | (4) |
Unless measured directly from the patient, Hct is assumed to be equal to 0.4, as measured by Just (18).
|
Functional Description |
|---|
|
TopAbstractIntroductionContrast Agent KineticsContrast Agent ConcentrationFunctional DescriptionConclusionsTAKE-HOME POINTSReferences |
|---|
MRIW can perform contrast agent uptake pharmacokinetic analysis on both T1-weighted and first pass bolus modeling of T2*-weighted dynamic contrast-enhanced MR imaging data. The quantitative imaging protocol for T1-weighted imaging uses a proton-density–weighted image acquisition, later used for signal conversion to contrast agent concentration, followed by dynamic T1-weighted image acquisitions of the identical section(s) for a period to allow circulation and uptake of contrast agent (19). The period of acquisition of dynamic data typically varies from 3 to 8 minutes. The bolus of contrast agent is injected intravenously 30 seconds after the start of the T1-weighted acquisition. The contrast agent is preferably administered at an injection rate of 4–5 mL/sec by using an automated MR imaging power injector. The T2*-weighted imaging protocol uses a T2*-weighted acquisition and does not require the initial proton-density–weighted acquisition. Injection methodology is the same as for the T1-weighted protocol. A combined acquisition yielding both T1- and T2*-weighted images is also supported; the combined acquisition is a T1-weighted acquisition with the addition of a second echo to provide T2*-weighted images (20).
Once pharmacokinetic data processing is completed, MRIW provides the ability to view the results, either for the whole ROI or on a pixel-by-pixel basis, in the postprocessing window. The average concentration-time curve (CTC) and model fit for the ROI, the histogram of the current functional parameter, or the individual pixel CTC and model fit may be chosen for display. The postprocessing window also allows export of results to formatted text files for more in-depth statistical analysis in other software packages. Histograms, average CTC, and pixel-by-pixel results can be exported in this manner. The currently displayed image and color overlay may be saved to the common portable network graphics (PNG) file format to provide a graphical record of the parameter distribution in the ROI.
Modeling of Contrast Agent Uptake
MRIW enables user-defined calibration curves (single and bi-exponential) to be entered for specific sites and coils. This is typically performed by using a test object with a variety of known T1 relaxation times, such as the Eurospin test object (Diagnostic Sonar, Livingston, West Lothian, Scotland). The interval between calibration measurements depends on the stability of the MR imaging unit being used; they should be performed on a regular basis, particularly after major software or hardware upgrades and after each cryogen fill.
The evolution over time of the contrast agent concentration is modeled by using the model of Tofts and Kermode (21), which describes the leakage of contrast agent from the blood plasma to the extravascular, extracellular space (EES) through permeable capillary walls. The model provides estimates of physiologic parameters of tissue. The concentration of contrast agent in the blood plasma, Cp, is derived theoretically to be a bi-exponential decay (22):
 | (5) |
The two terms in this plasma washout curve correspond to the equilibration of contrast agent between the plasma and extracellular space (fast) and the removal of contrast agent from the plasma by the kidneys (slow). D is the dose of gadopentetate dimeglumine in millimoles per kilogram of body weight. The fast component is represented by a1 and m1, while a2 and m2 represent the slow component. By using this plasma curve, the equation for contrast agent concentration in the tumor tissue is obtained:
 | (6) |
where kep = Ktrans/ve and a1, a2, m1, and m2 are the parameters of the plasma curve.
In the nomenclature reported by Tofts et al (11), Ktrans is the transfer constant describing the leakage of contrast agent from the blood plasma to the EES, ve is the fraction of tissue volume comprising the EES, and kep is the rate constant describing the return of contrast agent from the EES to blood plasma. As noted earlier, Ktrans is a measure of the permeability surface area product of the capillary walls in the tissue, although measured values may reflect poor delivery of contrast agent in perfusion-limited situations.
MRIW provides default parameters for the T1 calibration curve, the plasma washout curve (22), and the contrast agent relaxivity, but it allows users to enter, save, and load calibrations for their site-specific parameters, plasma curves, and the relaxivity of the contrast agent in use. This flexibility allows MRIW to handle a wide variety of input data in an appropriate way as long as suitable input parameters are provided. Temporal resolutions of between 1 and 15 seconds are typically used for T1-weighted dynamic studies, and the total acquisition time may be between 3 and 8 minutes. Model fitting is performed by using a nonlinear Levenburg-Marquart method.
The pharmacokinetic modeling is performed on a pixel-by-pixel basis within a user-selected ROI. Not all contrast agent behavior in tissue can be accurately represented with a two-compartment pharmacokinetic model. Where a model fit cannot be achieved, the shape of the uptake curve still provides useful information about tissue properties to the radiologist (23,24). The model-independent integrated area under the gadolinium curve (IAUGC) for each pixel is calculated to permit analysis of tumors where the assumptions underlying the models break down or where a model-independent approach is preferred (9,25).
Perfusion Modeling
The decrease in signal intensity corresponding to the passage of a bolus of contrast agent is manifested as a peak or series of peaks in Cp(t), corresponding to the first, second, and subsequent passes of the bolus through local vasculature following circulation through the rest of the body. By fitting a model function such as a gamma variate, 
(t), to the first-pass peak of Cp(t), properties such as relative blood volume (rBV), mean transit time (MTT), and relative blood flow (rBF) can be measured. The gamma variate has the following form (17):
 | (7) |
where K, 
, ß, and AT are fitted parameters: K is a normalization constant, and ß describe the ascending and descending edges of the first-pass peak, respectively, and AT is the arrival time of the bolus (taken from the midpoint of the bolus injection). AT is estimated by using an algorithm suggested in reference 26. To isolate the first-pass peak, images acquired 30 seconds or more after the arrival time of the bolus are not fitted.
A numerical model derived by Thompson (27) is used to estimate suitable starting values for each of the parameters in Equation (7). This model uses the ratio r, which is given by the following formula:
 | (8) |
where p is the time to peak and p1/2 is the time to reach half peak intensity on the ascending slope, both of which can easily be measured from Cp(t)., ß, and K are then given by the following formulas:
 | (9) |
 | (10) |
and
 | (11) |
where Cp( p) is the value of Cp(t) at t = p. The estimates for K, , ß, and AT are used as input parameters to the nonlinear model-fitting algorithm used to fit Equation (7) to the R2* data.
According to tracer dilution theory, rBV is the area under the fitted curve, MTT is the full width half maximum of the fitted peak, and, according to the central volume theorem, rBF is the ratio rBV/MTT. However, rBV can be estimated by taking the area under the curve (AUC) of Cp(t), using the trapezoidal approximation, over a region corresponding to the first-pass peak. This is taken as t = AT to AT + 30 seconds.
UNFOLD (unaliasing by Fourier-encoding the overlaps using the temporal dimension), described by Madore et al (28), can be used to reduce the effects of noise and transient image artifacts such as the ghosting caused by flow or motion during Fourier transform imaging. This is implemented in MRIW’s processing to improve the quality of the data being input to the fitting procedures to enhance the reliability and quality of the fit. For each pixel, the one-dimensional time series data were Fourier transformed to produce the frequency spectrum of the temporal data. These temporal frequencies were then filtered by using a Fermi filter, as given in Equation (12):
 | (12) |
The frequency, 
, is normalized such that ±max = ±1. The width and sharpness of the filter can be varied by adjusting the values of the shape parameters EF and kT. The values used are EF = 0.25 and kT = 0.04. We have optimized the filter coefficients to ensure best data modeling.
Clinical Applications
In practice, MRIW hides the complexity of pharmacokinetic modeling behind a user-friendly interface, allowing a clinician to easily interact with the data. Once image data series are selected, along with suitable calibration and timing data, the clinician can choose the ROI for analysis. The clinician may choose to scroll through the images to review where the contrast agent is taken up. Optionally, this may be performed with subtraction images to further highlight the effects of contrast agent. Figure 1 shows the main window of MRIW after ROI selection and data fitting have been performed on a dual-echo data set that provides both T1- and T2*-weighted images. This gives an overview of the results, which can then be reviewed in more detail by using the postprocessing window. The ROI outline in pale blue clearly demarcates pixels that are inside the ROI.
|
|
|
Patient motion is a problem in any dynamic measurement, and dynamic contrast-enhanced MR imaging data are no exception. If an enhancing structure moves within the image to the location of an adjacent nonenhancing structure during a measurement, this will cause that pixel CTC to exhibit enhancing behavior. Naive observation of the pixel CTC alone, without consideration of the motion, would incorrectly identify the pixel as enhancing. MRIW assumes that input data are registered, and thus the clinician must review the data for unacceptable motion. This is best achieved in MRIW during ROI selection, optionally by using the subtraction tool. Once an ROI is selected, the clinician can move backward and forward through the dynamic image set and observe the excursions of structures relative to the ROI outline.
Perfusion information is important in assessing both the viability of a tumor and the blood supply, which carries therapeutic agents to treat the tumor. There is a marked difference in the vasculature of normal tissue and of tumors. Normal vasculature forms an ordered structure with continuous, steady flow from arterial supply to venous drainage. Tumor vasculature is typically chaotic, leaky, and tortuous with uneven flow, pooling of blood, and even flow direction reversal. This difference can be demonstrated with MRIW. Figure 3 shows a well-perfused tumor, the same tumor and section shown in Figure 2. The first CTC, shown in Figure 3a, is of normal tissue vasculature; a sharp first-pass peak is visible with little or no contrast agent remaining in the tissue after passage. The CTC for tumor tissue vasculature shown in Figure 3b has a long tail, which indicates that contrast agent remains in the tumor tissue after first passage. This is due to a combination of tortuous vessels, leakiness of vascular walls allowing contrast agent to leak into the EES, and blood pooling within the tumor. To prevent this affecting the first-pass modeling of perfusion, MRIW uses only 30 seconds of data after the arrival of the contrast agent in the pixel.
|
|
A summary of the features of MRIW is provided in Table 2.
|
|
Conclusions |
|---|
|
TopAbstractIntroductionContrast Agent KineticsContrast Agent ConcentrationFunctional DescriptionConclusionsTAKE-HOME POINTSReferences |
|---|
Quantitative maps allow serial measurements of the same subject to assess response of tumors to treatment. Pixel-by-pixel analysis prevents the loss of information inherent when analyzing the ROI as a whole. Heterogeneity of a tumor and changes in the heterogeneity over time are important indicators of tumor physiology and response to therapy. MRIW has been used to analyze data for clinical drug trials to assess the efficacy of the particular treatment (29) and compare dynamic contrast-enhanced MR imaging results with results of histologic assessment (30). The reproducibility of dynamic contrast-enhanced MR imaging measurements has also been investigated (26,31). MRIW demonstrates the increased utility of computers and software in medical research and clinical practice.
MRIW is available under license from the Institute of Cancer Research through the Enterprise Business Unit. Contact details are available at http://www.icr.ac.uk/enterprise_unit.
|
TAKE-HOME POINTS |
|---|
|
TopAbstractIntroductionContrast Agent KineticsContrast Agent ConcentrationFunctional DescriptionConclusionsTAKE-HOME POINTSReferences |
|---|
|
Footnotes |
|---|
Abbreviations: CTC = concentration-time curve, DICOM = Digital Imaging and Communications in Medicine, EES = extravascular-extracellular space, MRIW = Magnetic Resonance Imaging Workbench, ROI = region of interest
|
References |
|---|
|
TopAbstractIntroductionContrast Agent KineticsContrast Agent ConcentrationFunctional DescriptionConclusionsTAKE-HOME POINTSReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  F. Chen, Y. Suzuki, N. Nagai, X. Sun, H. Wang, J. Yu, G. Marchal, and Y. Ni Microplasmin and Tissue Plasminogen Activator: Comparison of Therapeutic Effects in Rat Stroke Model at Multiparametric MR Imaging Radiology, August 1, 2007; 244(2): 429 - 438. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOGRAPHICS | RADIOLOGY | RSNA JOURNALS ONLINE |
