T2 preparation method for measuring hyperemic myocardial O2 consumption: in vivo validation by positron emission tomography




To validate a new T2-prepared method for the quantification of regional myocardial O2 consumption during pharmacologic stress with positron emission tomography (PET).

Materials and Methods

A T2 prepared gradient-echo sequence was modified to measure myocardial T2 within a single breath-hold. Six beagle dogs were randomly selected for the induction of coronary artery stenosis. Magnetic resonance imaging (MRI) experiments were performed with the T2 imaging and first-pass perfusion imaging at rest and during either dobutamine- or dipyridamole-induced hyperemia. Myocardial blood flow (MBF) was quantified using a previously developed model-free algorithm. Hyperemic myocardial O2 extraction fraction (OEF) and consumption (MVO2) were calculated using a two-compartment model developed previously. PET imaging using 11C-acetate and 15O-water was performed in the same day to validate OEF, MBF, and MVO2 measurements.


The T2-prepared mapping sequence measured regional myocardial T2 with a repeatability of 2.3%. By myocardial segment-basis analysis, MBF measured by MRI is closely correlated with that measured by PET (R2 = 0.85, n = 22). Similar correlation coefficients were observed for hyperemic OEF (R2 = 0.90, n = 9, mean difference of PET – MRI = −2.4%) and MVO2 (R2 = 0.83, n = 7, mean difference = 4.2%).


The T2-prepared imaging method may allow quantitative estimation of regional myocardial oxygenation with relatively good accuracy. The precision of the method remains to be improved. J. Magn. Reson. Imaging 2011;33:320–327. © 2011 Wiley-Liss, Inc.

CORONARY ARTERY DISEASE (CAD) manifests as an imbalance between supply and demand of myocardial oxygen. Myocardial O2 consumption (MVO2) represents an important physiological index of CAD and can provide direct evidence of ischemic heart disease. MR blood oxygen level-dependent (BOLD) imaging has been employed as a noncontrast agent approach to determine myocardial oxygenation extraction fraction (OEF) during hyperemia (1), using a segmented multicontrast turbo spin-echo (TSE) sequence. MVO2 can then be determined by Fick's law: MVO2 = OEF × MBF, where MBF is the microvascular myocardial blood flow. In the prior study, an imaging acquisition dark-blood turbo-spin-echo (DB-TSE) sequence was employed to measure myocardial T2 within a single breath-hold. By acquiring two T2 images with different echo spacing (τ) at rest to calculate model parameters, hyperemic T2 can be used to calculate hyperemic OEF through the two-compartment model (1). However, previously each T2 was calculated using only three different TEs due to limited acquisition time. Because the saturation of blood signal using DB-TSE relies on steady electrocardiogram (ECG) triggering, arrhythmias that caused irregular ECG triggering often induced flow artifacts and deteriorated image quality of the T2 maps. Intrascan motion between the two T2 acquisitions at rest might have significant impact in the calculation of model parameters. These factors have contributed to the inaccuracy of OEF estimation (2).

T2-prepared sequences (T2prep) have long been used to assess myocardial edema (3, 4) and reveal the T2-weighted BOLD effect in hearts (5). The bright-blood T2prep sequence is relatively less sensitive to the irregular ECG triggering in terms of flow artifacts. Because the T2prep method employs Hahn spin-echo or 3–4 180° pulses to achieve T2-weighting, relatively larger τ can be used in the T2prep. This may improve the sensitivity of the T2 method to oxygenation changes in the myocardial microvasculature, in which 90% of myocardial blood volume (MBV) resides in the capillaries (6). With fast gradient-echo acquisition, cardiac motion artifacts could be reduced and acquisition speed improved. The latter may help to increase the number of TEs with which to calculate myocardial T2. The aim of this study was to evaluate and validate a newly developed T2prep BOLD imaging sequence for the quantification of hyperemic myocardial MVO2 using positron emission tomography (PET) imaging as the gold standard method. The study was performed in a canine model with and without coronary artery stenosis.



In T2-weighted TSE or spin-echo (SE) images with an interecho spacing τ (the time difference between two consecutive 180° pulses in TSE or between the 90° pulse and subsequent echoes in SE), the signal in a myocardial tissue voxel can be approximated in a biexponential form from a two-compartment model as follows:

equation image(1)

where Svoxel is the mean signal intensity of the voxel at echo time TE and S0 is a variable related to the proton density of the voxel, receiver gain, and T1 of the tissue. T2apparent is the apparent T2 of the myocardium. T2b and T2t are the T2 values of blood and tissue, respectively. Using the van Zijl intravascular component model (7), intravascular T2b can be derived as:

equation image(2)

where α, β, and γ are functions of magnetic susceptibilities, interecho spacing (τ), oxygenation-dependent T2 of erythrocytes and plasma, TE, arterial oxygen saturation (Ya), and hematocrit. These constants can be derived with experimental data obtained at 1.5 T (8). The extravascular T2t can be approximated using a diffusion model (9, 10):

equation image(3)

where R20t is the intrinsic myocardial tissue transverse relaxation rate, and R21t is a function of the diffusion constant (D), the susceptibility difference between the blood vessel and the surrounding tissue, the geometry of the heart relative to the static B0 field, and the size of capillary and venous vessels. Increasing τ would improve the sensitivity of this method to the changes in myocardial oxygenation. Both parameters are subject-specific and independent of myocardial perfusion status. With knowledge of R20t, R21t, apparent myocardial T2, and MBV, hyperemic myocardial OEF can be calculated through Eqs. [1–3].

Magnetic Resonance Imaging (MRI) Method for the Calculation of OEF and MVO2

A T2prep sequence was implemented in a 1.5 T Siemens Sonata system (Siemens Medical Solutions, Malvern, PA). The sequence diagram is shown in Fig. 1. This is similar to a cardiac T2prep method reported previously (11), except that the data acquisition module is a gradient-echo readout, rather than the TrueFISP (true fast imaging with steady state precession) readout. A previous study (12) has shown that a greater number of 180° pulses in the T2 preparation module can reduce both the B1 and B0 inhomogeneities. If a TE is fixed, then less interecho spacing τ will be allowed in order to accommodate more 180° pulses. A simple simulation was performed to study the effect of the number of 180° pulses on the myocardial T2 sensitivity relative to the changes in myocardial OEF. Using approximate R20 and R21 values observed in the animal study, the simulation result is demonstrated in Fig. 1. It is apparent that fewer 180° pulses, or longer τ, may increase the sensitivity of the sequence to the changes in myocardial oxygenation. We have found that a T2prep module with three 180° pulses yielded minimal inhomogeneity artifacts with the echo times selected.

Figure 1.

Diagram of the T2prep sequence with gradient-echo readout (left) in one cardiac cycle and simulation result (right). In the T2prep module, three 180° pulses were used. TE1 was acquired in one cardiac cycle and TE2 with segment #1 was acquired in next cycle, and so on. The simulation shows T2prep with three 180° pulses has better sensitivity to T2 vs. OEF than the one with one 180° pulse.

To calculate myocardial OEF during hyperemia by a two-compartment model (1), two model parameters, R20t and R21t, need to be determined at rest. Rather than using T2 to calculate OEF from Eqs. [1–3], a regression method was developed to use the dataset at rest, {Svoxel, TE, α, β, and γ}, as shown in Eqs. [1–3], for the determination of R20t and R21t. Then using a similar T2-weighted hyperemic data set, a regression can be repeated again to calculate the hyperemic OEF. The procedure for this is detailed in the Appendix. Table 1 shows the TE and the model parameters used to fit Eqs. [1–3] in this method. Their derivations were reported previously (1), which was based on the theoretical model (13). Increasing τ will increase α, β, and γ, and thereby increase the sensitivity of T2 to the change in OEF (Eqs. [2, 3]). MBV at rest and during hyperemia were calculated independently by first-pass perfusion imaging.

Table 1. Fitting Parameters in Eq. [1] for the Calculation of Myocardial Hyperemic OEF
TE (msec)αβγ

Using Fick's law, MVO2 can be calculated by:

equation image(4)

where [Hb] is the hemoglobin concentration in the blood (g/mL) and 1.39 represents the maximum binding capacity of oxygen per unit mass of hemoglobin (mL/g). In our dogs, the average [Hb] was approximately 0.14 g/mL and arterial oxygen saturation Ya was 0.98.

Animal Preparation

A total of six healthy beagles (n = 6, weight = 9.4 ± 2.4 kg) were examined with the approval of the Animal Studies Committee at our institution. The six dogs were randomly selected either to undergo coronary artery narrowing surgery in the proximal left anterior descending arteries (LAD) or to be left intact. On the days of the imaging sessions the dogs were sedated with 1–2 mg/kg body weight of morphine subcutaneously and anesthetized with 12.5 mg/kg of sodium thiopental intravenously. The dogs were then intubated with an endotracheal tube and ventilated with 100% oxygen at a tidal volume of 12 mL/kg and a rate of 15–20 breaths/minute. Anesthesia was maintained with ventilated 2%–3% isoflurane. The heart rate and blood pressure were monitored continuously and hard copies of the record were obtained every 5 minutes using an MRI-compatible hemodynamic monitor (Vital Signs Monitoring System, Invivo Research, Orlando, FL).

For stenosis creation, a thoracotomy was performed in the fourth intercostal space and the pericardium was incised. The LAD was dissected free distal to the first diagonal branch. The artery was instrumented in a proximal-distal order, with a Doppler flow probe, a pneumatic occluder, and an MRI-compatible, adjustable stenosis clamp. Several 20-second occlusions were performed to determine the typical hyperemic flow response. After tightening the stenosis clamp, another occlusion was performed to assess the change in hyperemic flow response. After attaining the desired level of stenosis, as determined by the reduction in hyperemic flow response (14), the occluder and flow probe were removed and the chest was closed for imaging. In this study the stenosis severities were 0% (n = 2), 75% (n = 2), and 95% (n = 2). These dogs were randomly assigned to receive either dipyridamole or dobutamine stress. Dipyridamole is a potent coronary vasodilator and dramatically increases MBF (and thus decreases OEF) without significantly increasing MVO2. The ionotropic agent dobutamine, on the other hand, increases MVO2, and through β2 adrenoreceptor-mediated vasodilation of resistance vessels, moderately increases MBF. Therefore, the study was designed to create a large range of hyperemic OEF and MVO2.

MRI and PET Protocol

The MR scans were performed on a 1.5T system equipped with a fast gradient system (maximal gradient strength = 40 mT/m, maximal slew rate = 200 mT/m/ms). A four-element phased array coil was used for signal reception. Each beagle was placed supine within a plastic bed and a three-lead ECG patch was attached to its chest for ECG-monitoring and pulse-sequence triggering. All MR scans were performed during a single breath-hold (≈15–20 sec), achieved by turning off ventilation at the end of expiration. After scout imaging, a short axis plane of the left ventricle (LV) at the mid-cavity level was obtained with a T2prep sequence, both at rest and during dipyridamole- or dobutamine-induced hyperemia. Dipyridamole was infused at 0.14 mg/min/kg for 4 minutes with an MRI-compatible syringe pump (Harvard Apparatus, Holliston, MA). Dobutamine was started at a dose of 5 μg/min/kg body mass, and titrated at 5 μg/min/kg intervals every 3 minutes until the rate-pressure product was approximately doubled compared to rest. Dobutamine doses were typically the same during MR and PET scans; the average dobutamine dose was 20 μg/min/kg during both MR and PET scans.

For T2prep scans, five TEs were used: 24, 36, 48, 60, and 72 msec all in one breath-hold. Other imaging parameters included: field of view (FOV) = 220 × 130 mm2, slice thickness = 8 mm, k-space matrix = 256 × 76. For gradient-echo data acquisition, TR/TE = 2.8/1.2 msec, segmentation number = 19, flip angle = 12°, and total data acquisition time = 22 × RR-interval (≈600 msec) or 13.2 sec. The T2 scans were repeated three times at rest to evaluate the repeatability of the cardiac T2 measurements.

To calculate hyperemic OEF, another technique is required to measure MBV for Eqs. [1, 3]. In addition, in order to calculate MVO2 from Fick's law (Eq. [1]), MBF data is needed; thus, both MBF and MBV were acquired using quantitative MR first-pass perfusion imaging techniques. This dynamic imaging was performed during a bolus injection of 0.02 mmol/kg Multihance (Bracco Diagnostic, Princeton, NJ), an extracellular contrast agent, using a saturation-prepared turbo fast low-angle shot (FLASH) sequence. Imaging was started after T2prep imaging at rest and during hyperemia. The duration between resting and hyperemic imaging was ≈30 minutes. The same short-axis slice of the LV was acquired during mid-diastole, triggered by the R-wave of the ECG. Sixty-eighty dynamic images were gathered sequentially at every RR interval (typically ≈600 msec after the R-wave). Other imaging parameters included: TR = 2.5 msec; TE = 1.2 msec; TI = 90 msec; flip angle = 18°; FOV = 220 × 138 mm2; matrix size = 128 × 64; slice thickness = 8 mm; BW = 675 Hz/pixel; and image acquisition time window per cardiac cycle = 150 msec.

After the MRI study the dogs were immediately moved to the PET suite. PET imaging was performed on a Focus 220 microPET scanner (Concorde Medical Systems, Knoxville, TN). A transmission scan was first performed with a rod source to ensure proper positioning and to correct for photon attenuation. Then 15O-water (ave. 7.5 ± 1.8 mCi) was administered intravenously and dynamic PET data were acquired for 5 minutes. After allowing 10 minutes for decay of the 15O-water, 11C-acetate (ave. 7.8 ± 2.4 mCi) was injected intravenously and dynamic PET data were acquired for 30 minutes. Approximately 40 minutes after the resting 11C-acetate scan (to allow for radionuclide decay), the pharmacologic stressor was started, set to the appropriate dose to approximately match the hemodynamics during MRI, and the 15O-water and 11C-acetate protocols were repeated for stress imaging.

Image Analysis

One blinded observer analyzed the MRI and PET images. First-pass perfusion images were first analyzed with a JAVA program (Java V5.0, Sun Microsystems, Santa Clara, CA) reported previously (15). Images were denoised by a wavelet method and then analyzed by a new model-independent perfusion algorithm (16). This algorithm rapidly generated both MBF and MBV maps on which regions of interest (ROIs) could be drawn. Due to inappropriate injection of contrast agent during hyperemia, hyperemic MBF data from the first dog, with 0% stenosis, were discarded. Two resting MBF data (LAD and LCx) from another dog were also excluded due to motion artifacts induced by inappropriate anesthesia during first-pass perfusion imaging at rest. With six dogs at two segments (LAD and LCx) and two different flow states (resting and hyperemia), excluding the four data points mentioned, there were a total of 20 MBF data points.

To calculate hyperemic OEF, signal intensities from two large ROIs on the anterior (LAD-perfused) and the posterior (LCx-perfused) regions on the baseline T2-weighted images were used to calculate model parameters R20t and R21t. The signal intensities on the stress T2-weighted images were then fitted to Eq. [1] to calculate hyperemic OEF (Appendix). To assess repeatability of T2 measurements, the mean T2 values on the posterior remote myocardial regions were measured in the three resting T2 maps. The mean and standard deviation (SD) of three T2 values were obtained and the repeatability in each dog was expressed as the ratio of SD over the mean.

For PET images, the nongated, attenuation-corrected images were reconstructed with filtered back-projection and transferred to a Sun workstation (Sun Microsystems, Menlo Park, CA). Three image slices that approximately matched the MR slice were selected and ROIs matching the MR image ROIs were drawn to generate blood and myocardial time–activity curves. MBF was determined by a previously validated compartmental modeling method (17). MVO2 was then determined by a one-compartment kinetic model to estimate the rate at which 11C-acetate was converted to 11CO2 to determine MVO2 (18). Finally, regional OEF was estimated by application of Fick's law (Eq. [4]).

In this PET study there should be 12 myocardial segments with hyperemic MVO2 data points generated by 11C-acetate imaging. However, data from three segments in the second and third dogs were discarded due to isotope spillover effects or severe motion artifacts. Because MVO2 calculation in MRI relies on MBF and OEF and the MBF values in the first dog cannot be used, the comparable number of data points in OEF and MVO2 were 9 and 7, respectively.

Data Analysis

All data are reported as mean ± SD. Statistical analysis was performed with paired t-test to compare the regional parameters (MBF, OEF, and MVO2) during the hyperemia between PET and MRI. A P < 0.05 was considered statistically significant.


Hemodynamic and T2 Data

Hemodynamic parameters of all the dogs at rest and during peak hyperemia were recorded. Figure 2 shows the correlation of rate pressure product (RPP) between PET and MRI, as well as the Bland–Altman plot for the RPP data. It is clearly shown that the PET RPP was slightly lower than the MRI RPP, but there was no statistically significant difference. Repeatability of myocardial T2 at rest was 2.3% ± 2.1%, which appears to be lower than 3.5% obtained using a T2-weighted turbo spin echo sequence in dogs (1), although the difference was not significant (P = NS).

Figure 2.

Correlation of rate-pressure-product between MRI and PET imaging sessions. Except for one outlier observed in the Bland–Altman plot, data indicates fairly comparable hemodynamic conditions between PET and MRI acquisitions.

MBF, Hyperemic OEF, and MVO2

Figure 3 demonstrates MRI and PET acetate images in one dog with 95% stenosis during dipyridamole induced hyperemia. Figure 4 shows the correlation and Bland–Altman plots between PET and MRI measurements in MBF (n = 20). MRI measurements slightly underestimated MBF by 2.3% with limits of agreement between −22.6% and 27.5%.

Figure 3.

Examples of T2prep images at rest (a) and during dipyridamole vasodilation (b) in a dog with severe LAD stenosis. One first-pass perfusion source image during the vasodilation is shown in (c). The MBF map (d) shows the ROIs drawn in the LAD and LCx subtended myocardial segments. Black arrows point to the hypointensity in the LAD region, even at rest. The corresponding 11C-acetate PET image (e) demonstrates markedly reduced acetate uptake.

Figure 4.

Correlation of MBF measurements between PET and MRI (left) and corresponding Bland–Altman plot (right). Different symbols represent different stenosis and stress states. There was no systematical error, but relatively large limits of agreement. The arrow in the left plot indicates two overlapping data points.

Figure 5 demonstrates the correlations in OEF (n = 9) and MVO2 (n = 7) between PET and MRI. Another dataset was added in each correlation plot for assuming the resting OEF data used to calculate the hyperemic OEF is equal to the resting OEF measured by PET, rather than using a fixed value of 0.6. In the Bland–Altman plots, only the datasets using the fixed resting OEF of 0.6 were compared between PET and MRI. There were no significant differences in hyperemic OEF and MVO2 observed between using either set of resting data. MRI methods slightly overestimated hyperemic OEF by 2.4% and underestimated hyperemic MVO2 by 4.2%. However, the limits of agreement for both measurements were relatively large, −33.9% to 29.1% (OEF) and −12.9% to 21.4% (MVO2), respectively.

Figure 5.

Correlation and Bland–Altman plots of OEF (top) and MVO2 (bottom) between PET and MRI. In the Bland–Altman plots, except for one outlier, other data between the two imaging modalities agreed very well, except for one outlier. For the correlation plots the filled circles represent the MRI hyperemic OEF calculated using an assumed resting OEF of 0.6. The open circles represent the MRI hyperemic OEF calculated using the resting OEF measured by PET. The Bland–Altman plots use only OEF values calculated from the assumed resting OEF value.


In this study a new T2prep sequence was developed to measure myocardial T2 within a single breath-hold. Although a dark-blood TSE sequence can measure myocardial T2, the limited echo number and flow artifacts in the left ventricle may reduce the accuracy and precision of the T2 measurement. Because the sensitivity of myocardial T2 to the change of myocardial oxygenation can be adjusted by the interecho spacing in the TSE sequence, it is better to employ a relatively large interecho spacing. However, this is limited by the data acquisition window during mid-diastole. Our T2prep sequence can provide variable interecho spacings and uses a bright blood approach to reduce flow artifacts. Since the data acquisition was a gradient-echo module, the echo number, segmentation number, and data acquisition window can be adjusted to allow for up to five echoes to be collected within one breath-hold. The resting myocardial T2 measurements show good repeatability. The resulting hyperemic OEF and MVO2 data measured by MRI appear to correlate well with the PET measurements.

With the previous technique for estimating model parameters in hyperemic OEF calculation, the resting myocardial T2 must be measured twice, at two different echo spacing values. In the new implementation, myocardial T2 was measured using different echoes, thus different echo spacing values were used in individual measurements. The model parameters could be extracted using a multivariate regression approach and signal intensities at different echo times. This removes the intrascan motion problem associated with the previous approach and simplifies the procedure to calculate OEF. Furthermore, using signal intensity to perform regression can be less sensitive to imaging noise, compared to using T2 values, since regression of T2-weighted signal intensities would introduce additional noise. While gradient-echo acquisitions have intrinsically low signal-to-noise ratios (SNR), the beginning of the data acquisition module followed immediately after the T2prep module, allowing acquisition of the most T2-weighted signal. Another acquisition option is the TrueFISP sequence. However, the requirement for a steady state in this sequence separates the T2prep module and acquisition module by a series of RF pulses, which would reduce the T2-weighting effect. Other methods using transient TrueFISP signal intensity (11), ie, without steady state preparation pulses, may be alternative to improve SNR and warrants further investigation.

One notable issue in the T2 measurement of myocardium in ischemia or infarction is the impact of edema on the calculation of myocardial OEF. Edema would likely increase resting T2, relative to normal myocardial T2 (4). However, the calculation of hyperemic OEF actually relies on the difference of myocardial T2 between hyperemic and resting states (ie, the BOLD effect). Such differences should still reflect the oxygen supply and demand of underlying myocardial tissue, with or without edema. It is thus unlikely that edema would have any major effect on the calculation of myocardial OEF and MVO2 using this method. Regardless, we did not observe any edema with our animal model and further study on this issue is still needed to confirm our theoretical prediction.

The design of the validation protocol was intended to create a wide range of myocardial perfusion and oxygen consumption rates. Either vasodilatory or ionotropic agents were applied. Regional MBF ranged from 1.1 to 3.0 mL/min/g, whereas regional OEF and MVO2 varied from 0.27 to 0.99, and from 6.7 to 17.1 μmol/g/min, respectively. While the measurements in OEF and MVO2 achieved less than 5% error, the limits of agreement remains to be improved. Such variation may be partially swayed by the small sample size. Another reason may be due to some variations in hemodynamics between PET and MRI, as seen in Fig. 1. If the one outlier data point is removed, the limits of agreement would be −24% to 10.6% for OEF, and −11% to 15% for MVO2. Due to the small number of subjects, the intraobserver variability was not assessed.

Using conventional extracellular contrast agents in the first-pass perfusion imaging, MBF is usually quantified with a Fermi deconvolution. Compared to this method, our newly developed deconvolution method has the advantages of fewer parameters and assumptions, which in turn leads to lower noise susceptibility and faster calculation in MBF mapping. MRI slightly underestimated MBF when compared to PET, but the correlation was excellent and no apparent systematic error was observed. This is consistent with other findings using an intravascular contrast agent (12), likely due to noise contamination that leads to wider impulse response with a lower amplitude (lower MBF). The Bland–Altman limits (−23% to 27% in Fig. 3) were relatively high, mainly due to a couple of severely underestimated MBF data points. Further optimization in contrast dosage and imaging parameters to improve SNR may be needed to improve the precision of MBF measurements.

There are a few limitations to the T2prep method in this study. First, image inhomogeneity was occasionally observed in the LV myocardium in T2prep images, causing spatial variations in the calculated OEF and MVO2. This may be induced by imperfections in the 180° RF pulses (12). To address this, the number of 180° RF pulses can be adjusted in our T2prep implementation. When more 180° pulses were applied, the images showed less inhomogeneity, but this method was less sensitive to changes in myocardial oxygenation. While three 180° pulses were used as the optimal case, they were not always adequate when severe cardiac motion presented. Second, resting OEF of 0.6 was assumed in our study in order to estimate hyperemic OEF. Using the resting OEF obtained by PET in our model calculation failed to show significant improvement. This may be explained by similar resting OEF measured by PET (0.67 ± 0.2) and assumed MRI data (0.6), as well as by minimal hemodynamic differences between PET and MRI studies. Finally, the SNR in our gradient-echo sequence was relatively low. One possible solution is to use a transient TrueFISP sequence, which may be the subject of our future research.

In conclusion, a T2prep sequence with our new algorithm to determine myocardial OEF and MVO2 was validated in a canine model by the PET study. MBF quantified by our newly developed model-free deconvolution method was also validated against PET measurements. Compared to the TSE method, T2prep is more robust in patient study with easy setup and, more important, may have better precision and sensitivity. Although the completely noninvasive arterial spin labeling (ASL) technique (19) allows measurement of MBF, it cannot measure MBV for the OEF calculation. The first-pass perfusion imaging approach remains the best choice for measuring MBF and MBV needed for calculating OEF and MVO2. The proposed technique was performed in this large animal model, aiming for future clinical application detecting myocardial ischemia in human subjects. The simple two-breath-hold imaging technique allows hyperemic OEF and MVO2 measurements in one image slice, if MBF and MBV in the first-pass perfusion imaging can be quantified. Such measurements may help monitor myocardial function and metabolism in patients with either regional or global ischemia, and may allow consecutive assessments of dose-responses in the myocardium to various therapeutic interventions.


We thank Pamela Baum, Susie Grathwohl, and Terry Sharp for animal preparation and assistance with PET imaging. We also thank Pilar Herrero and Zulfia Kisrieva-Ware for assisting with PET data analysis, as well as Laura Schollmeyer for assisting with editing of the article.


Myocardial T2prep images were analyzed with the T2-weighted signal intensities at rest and during the stress were obtained at different TEs. Theoretically, R20t and R21t can be determined through Eqs. [1–3] by applying multivariate nonlinear regression. However, it was found that nonlinear fitting was not stable since three parameters needed to be calculated from only five data points. We have thus developed the following “linear” procedures to calculate these parameters and hyperemic myocardial regional OEF.

  • 1Using a single-variable nonlinear regression, apparent T2apparent values are first calculated at rest and during the stress; S0 is obtained as well, but not used in the calculation of OEF.
  • 2With an assumed or PET-measured resting OEF, R2b is calculated at rest for each TE using Eq. [2], and the R2t is calculated using Eq. [1] with MBV obtained from the perfusion measurement at rest.
  • 3In Eq. [3], R2t is linearly correlated with R20t and R21t by only the single independent variable term OEF2MBV2τ2, where τ = TE/3; R20t (the intercept) and R21t (slope) can be obtained by linear fitting of this equation.
  • 4Having determined R20t, R21t, and hyperemic MBV, the relationship (calibration curve) of hyperemic T2apparent and hyperemic OEF can be derived using Eq. [1]. Based on T2apparent during the stress from procedure 1), myocardial OEF during the stress can be derived.