Three dimensional first-pass myocardial perfusion imaging at 3T: feasibility study
© Shin et al; licensee BioMed Central Ltd. 2008
Received: 21 June 2008
Accepted: 11 December 2008
Published: 11 December 2008
In patients with ischemic heart disease, accurate assessment of the extent of myocardial perfusion deficit may be important in predicting prognosis of clinical cardiac outcomes. The aim of this study was to compare the ability of three dimensional (3D) and of two dimensional (2D) multi-slice myocardial perfusion imaging (MPI) using cardiovascular magnetic resonance (CMR) in determining the size of defects, and to demonstrate the feasibility of 3D MPI in healthy volunteers at 3 Tesla.
A heart phantom was used to compare the accuracy of 3D and 2D multi-slice MPI in estimating the volume fraction of seven rubber insets which simulated transmural myocardial perfusion defects. Three sets of cross-sectional planes were acquired for 2D multi-slice imaging, where each set was shifted along the partition encoding direction by ± 10 mm. 3D first-pass contrast-enhanced (0.1 mmol/kg Gd-DTPA) MPI was performed in three volunteers with sensitivity encoding for six-fold acceleration. The upslope of the myocardial time-intensity-curve and peak SNR/CNR values were calculated.
Mean/standard deviation of errors in estimating the volume fraction across the seven defects were -0.44/1.49%, 2.23/2.97%, and 2.59/3.18% in 3D, 2D 4-slice, and 2D 3-slice imaging, respectively. 3D MPI performed in healthy volunteers produced excellent quality images with whole left ventricular (LV) coverage. Peak SNR/CNR was 57.6 ± 22.0/37.5 ± 19.7 over all segments in the first eight slices.
3D performed better than 2D multi-slice MPI in estimating the size of perfusion defects in phantoms. Highly accelerated 3D MPI at 3T was feasible in volunteers, allowing whole LV coverage with excellent image quality and high SNR/CNR.
Ischemic heart disease (IHD) is a leading cause of mortality in the world, accounting for over 7 million deaths in 2002 and over 9 million expected deaths by the year 2030 . Coronary catheter-based X-ray angiography (coronary angiography) is currently the "gold standard" modality for assessing disease of the coronary arteries that is the basis for IHD. However, coronary angiography is an invasive approach requiring insertion of a catheter into the coronary artery and administration of contrast agent to demonstrate the coronary artery disease. Considerable attention has been paid to non-invasive myocardial perfusion imaging methods that provide a means to assess the distribution of blood flow within the myocardium. Presently, single photon emission computed tomography (SPECT) is routinely used in clinical myocardial perfusion imaging (MPI) studies, but suffers from attenuation artifacts, low spatial resolution and exposure to ionizing radiation .
First-pass cardiovascular magnetic resonance (CMR) MPI is an ideal tool for the evaluation of myocardial perfusion, due to its high spatial resolution, lack of ionizing radiation, and relatively short scan time. First-pass CMR MPI tracks the passage of a gadolinium-based T1 contrast agent by time resolved T1 weighted imaging . Two dimensional (2D) multi-slice imaging is typically used as a means for data acquisition in CMR MPI, and has demonstrated high sensitivity (>85%) and specificity (>75%) [4–7]. Besides its diagnostic utility, MPI is also an important approach for assessment of prognosis in symptomatic patients. Patient follow-up studies have shown that the extent of abnormal myocardial perfusion is useful for the prediction of cardiac outcomes [8–10]. However, its accurate estimation is restricted in conventional 2D multi-slice MPI, due to the limited tradeoff between spatial coverage and inter-slice spacing.
Three dimensional (3D) CMR MPI is an advantageous alternative to 2D multi-slice CMR MPI since it supports contiguous spatial coverage and therefore has the potential to more accurately estimate the extent (volume) of abnormal perfusion. Moreover, 3D encoding inherently retains high signal-to-noise ratio (SNR) efficiency and high capacity for parallel imaging acceleration , and has been demonstrated at 1.5 T . The purpose of the present study was (i) to evaluate the performance of 3D MPI and 2D multi-slice MPI for estimating defect size in a phantom study, and (ii) to demonstrate the feasibility of in-vivo 3D MPI accelerated by sensitivity encoding (SENSE) encompassing the entire left ventricle (LV) at 3 Tesla.
All experiments were performed on a General Electric 3 Tesla scanner with peak gradient amplitude of 40 mT/m and peak slew rate of 150 T/m/s, using a body-coil for RF transmission and an eight-channel cardiac coil array for signal reception.
3D k-space for a given spatial resolution and FOV was undersampled at a net acceleration factor of six. Since the readout encoding line (k x ) is always fully sampled, undersampling in the other two encoding lines (k y , k z ) can be viewed in the 2D ky-kz plane (Fig. 1). Six-fold undersamped data were obtained by three-fold undersampling in ky and two-fold undersampling in k z . Aliasing-free 3D images were reconstructed by using SENSE, and the resulting temporal resolution was one heart beat. The position of sampled k-space encoding lines was shifted in a cyclic pattern in the ky-kz plane such that acquired data from six neighboring cardiac cycles could be combined to form a coil sensitivity map for SENSE reconstruction (Fig. 1).
Imaging parameters of 3D and 2D multi-slice scans
100 × 66 × 10
120 × 80
30 × 30 × 10 cm3
30 × 30 cm2
The borders of defects in reconstructed images were manually outlined, and the volume of a defect was expressed as the percent of the whole myocardium. To accurately estimate partial inclusion of the insets in a voxel, the phantom image with a defect was normalized by the image without a defect, and the fraction of defect was linearly interpolated in each normalized voxel.
In-vivo experiments using 3D MPI were performed in three volunteers. Written informed consent was obtained from all participants. Imaging parameters were the same as those in Table 1 except T SR = 130 ms and matrix size = 100 × 66 ~69 × 10. Data acquisition was located at the center of diastole, and the acquisition time was 304 ms. A proton density weighted data set was obtained using 4° flip angle with the saturation pulse turned off during the first six cardiac cycles. Contrast media (0.1 mmol/kg Gd-DTPA, Magnevist) was injected at a rate of 5 ml/s followed by 20 ml saline flush at the same rate. Subjects were instructed to hold their breath as long as possible.
All image reconstruction was performed off-line using MATLAB (Mathworks, Natick, MA). Proton density weighted images were obtained by simply combining the k-space data from the first six cardiac cycles. SENSE was used for the image reconstruction from six-fold undersampled 3DFT perfusion data [11, 17]. SENSE reconstruction of undersampled data from the k th cardiac cycle utilized coil sensitivity maps derived from the (k-2) th to (k+3) th cardiac cycles (see Fig. 1).
Raw perfusion images were normalized by proton density images to remove variations in the receiver coil sensitivities. Corrected images were then segmented into six (basal and mid-short axis levels) or four (apical level) myocardial sectors, according to the American Heart Association 17-segment model . Within each myocardial sector, a time-intensity-curve (TIC) was generated and the corresponding upslope value was computed by linear fitting of the data during signal enhancement.
Noise standard deviation is typically calculated from the background region-of-interest (ROI) for the computation of SNR, assuming spatially uniform noise distribution [19, 20], but this assumption is no longer valid in SENSE reconstruction due to spatially varying reconstruction noise. Hence in this work, SNR was calculated by the "difference method" , where noise standard deviation was estimated from the difference in two consecutive images.
The heart phantom study has demonstrated the superiority of 3D CMR MPI over 2D multi-slice CMR MPI in sizing perfusion defects. The limitation of this phantom study was that only transmural defects were tested, due to the lack of a rubber model for a subendocardial defect. The capability of sizing various types of defects will be investigated by perfusion scans in patients with known coronary artery disease (CAD).
When viewed in a video format (Additional files 1, 2 and 3), the in-vivo perfusion images showed flickering artifacts due to errors in sensitivity map used for SENSE reconstruction. This artifact appeared only during the start of RV and LV enhancement, which barely affected the upslope of myocardial TIC. The flickering can be reduced by employing variable density k-space undersampling. The artifacts will be reduced and diffused over the FOV by taking central region of k-space only from current cardiac cycle, and combining outer part of k-space from neighboring cardiac cycles. More actively, strong quadratic regularization can be used for the separate reconstruction of each coil map from a smaller number of neighboring k-space data sets, which will involve less signal change in blood pools.
Six-fold acceleration was achieved by applying SENSE in the in-vivo studies, but further acceleration will be explored for the following reasons. First, spatial resolution of 3.0 × 4.3~4.5 × 10 mm3 used in this study may be insufficient for visualizing subendocardial defects. Higher spatial resolution may also help to avoid the dark rim artifacts which were seen on most volunteer images. Gibbs ringing is one of critical reasons for the artifacts, and the use of high spatial resolution has been shown to restrict its transmural extent [22, 23]. Second, the data acquisition time (304 ms) need to be shortened to avoid potential motion artifacts in subjects with high heart rates. An eight-channel receiver coil was used in this study, and using larger array of coil elements would be one way to increase acceleration rate. Recently introduced compressed sensing concept is another promising source for further acceleration . The addition of a regularization of l1-norm such as total variation to SENSE reconstruction will enable image reconstruction from more highly undersampled k-space data.
SNR in a 3D encoded image varies along the partition encoding direction in proportion to the shape of excitation profile (refer to Fig. 6). With TBW of slab excitation = 4 used in this study, the lowest peak SNR in the basal slice (slice #1) was 64% of the highest SNR in the mid short axis slice (slice #4). Evenly high SNR over partition slices should be desired for consistent diagnostic capacity. Variable-rate selective excitation (VERSE) is a technique that modifies the original RF and gradient waveform such that either peak RF value or total RF duration can be reduced with the slice profile unchanged [25, 26]. We will explore the use of VERSE technique to increase TBW of slab excitation RF for a sharper profile without increase in RF duration.
In addition to high SNR and contiguous coverage, the large convenience of image registration is a potential merit of 3D encoding. The registration of time-resolved perfusion images is important for subsequent semi-quantitative and/or absolute quantitative perfusion analysis such as TIC upslope. Typically, the registration of 2D images should be performed by non-rigid methods that allow image deformation to accurately model through-plane motion. Since through-plane motion can be registered as accurately as in-plane motion in 3D images, respiratory motion can be corrected for by relatively simple 3D rigid-body model.
We have demonstrated that 3D CMR MPI is superior to 2D multi-slice CMR MPI in sizing transmural perfusion defects. Mean/standard deviation of errors in estimating the volume fraction across the seven defects were -0.44/1.49%, 2.23/2.97%, and 2.59/3.18% in 3D, 2D 4-slice, and 2D 3-slice imaging, respectively. We have also demonstrated the feasibility of in-vivo 3D MPI at 3T accelerated by sensitivity encoding. 3D CMR MPI at 3T can provide complete LV coverage with excellent image quality and high peak SNR and CNR. Evaluation of this technique in patients with known CAD, and the use of regularized parallel imaging reconstruction for higher spatial resolution and speed, are currently under investigation.
The authors thank Michael K. O'Connor from the Mayo Clinic for providing the heart phantom.
- Mathers CD, Loncar D: Projections of global mortality and burden of disease from 2002 to 2030. PLoS Medicine. 2006, 3: e442-10.1371/journal.pmed.0030442.PubMed CentralView ArticlePubMedGoogle Scholar
- Hunold P, Schlosser T, Barkhausen J: Magnetic resonance cardiac perfusion imaging- a clinical perspective. Eur Radiol. 2006, 16: 1779-1788. 10.1007/s00330-006-0269-3.View ArticlePubMedGoogle Scholar
- Kellman P, Arai AE: Imaging sequences for first pass perfusion-A review. J Cardiovasc Magn Reson. 2007, 9: 525-537. 10.1080/10976640601187604.View ArticlePubMedGoogle Scholar
- Wolff SD, Schwitter J, Coulden R, Friedrich MG, Bluemke DA, Biederman RW, Martin ET, Lansky AJ, Kashanian F, Foo TK, Licato PE, Comeau CR: Myocardial First-Pass Perfusion Magnetic Resonance Imaging A Multicenter Dose-Ranging Study. Circulation. 2004, 110 (6): 732-737. 10.1161/01.CIR.0000138106.84335.62.View ArticlePubMedGoogle Scholar
- Plein S, Radjenovic A, Ridgway JP, Barmby D, Greenwood JP, Ball SG, Sivananthan MU: Coronary artery disease: myocardial perfusion MR imaging with sensitivity encoding versus conventional angiography. Radiology. 2005, 235 (2): 423-430. 10.1148/radiol.2352040454.View ArticlePubMedGoogle Scholar
- Bernhardt P, Engels T, Levenson B, Haase K, Albrecht A, Strohm O: Prediction of necessity for coronary artery revascularization by adenosine contrst-enhanced magnetic resonance imaging. Int J Card. 2006, 112: 184-190. 10.1016/j.ijcard.2005.08.050.View ArticleGoogle Scholar
- Klem I, Heitner JF, Shah DJ, Sketch MH, Behar V, Weinsaft J, Cawley P, Parker M, Elliott M, Judd RM, Kim RJ: Improved detection of coronary artery disease by stress perfusion cardiovascular magnetic resonance with the use of delayed enhancement infarction imaging. J Am Coll Cardiol. 2006, 47: 1630-8. 10.1016/j.jacc.2005.10.074.View ArticlePubMedGoogle Scholar
- Brown KA, Boucher CA, Okada RD, Guiney TE, Newell JB, Straus HW, Pohost GM: Prognostic value of exercise Thallium-201 imaging in patinets presenting for evaluation of chest pain. J Am Coll Cardiol. 1983, 1 (4): 994-1001.View ArticlePubMedGoogle Scholar
- Hachamovitch R, Berman DS, Shaw LJ, Kiat H, Cohen I, Cabico JA, Friedman J, Diamond GA: Incremental prognostic value of myocardial perfusion single photon emission computed tomography for the prediction of cardiac death. Circulation. 1998, 97: 535-543.View ArticlePubMedGoogle Scholar
- Marwick TH, Case C, Sawada S, Rimmerman C, Brenneman P, Kovacs R, Short L, Lauer M: Prediction of mortality using dobutamine echocardiography. J Amer Coll Cardiol. 2001, 37: 754-760. 10.1016/S0735-1097(00)01191-8.View ArticleGoogle Scholar
- Weiger WM, Pruessmann KP, Boesiger P: 2D SENSE for faster 3D MRI. Magma. 2002, 14: 10-19. 10.1007/BF02668182.View ArticlePubMedGoogle Scholar
- Kellman P, Zhang Q, Larson AC, Simonetti OP, McVeigh ER, Arai AE: Cardiac first-pass perfusion MRI using 3D TrueFISP parallel imaging using TSENSE. Proc, ISMRM, 12th Annual Meeting, Kyoto. 2004, 310-Google Scholar
- Staewen RS, Johnson AJ, Ross BD, Parrish T, Merkle H, Garwood M: 3-D FLASH Imaging Using a Single Surface Coil and a New Adiabatic Pulse, BIR-4. Invest Radiol. 1990, 25 (5): 559-567. 10.1097/00004424-199005000-00015.View ArticlePubMedGoogle Scholar
- Kim D, Gonen O, Oesingmann N, Axel L: Comparison of the effectiveness of saturation pulses in the heart at 3T. Magn Reson Med. 2008, 59: 209-215. 10.1002/mrm.21423.View ArticlePubMedGoogle Scholar
- Knesaurek K, Machac J: Comparison of 18F SPECT with PET in myocardial imaging: A realistic thorax-cardiac phantom study. BMC Nucl Med. 2006, 6: 5-10.1186/1471-2385-6-5.PubMed CentralView ArticlePubMedGoogle Scholar
- Kim D, Axel L: Multislice, dual-imaging sequence for increasing the dynamic range of the contrast-enhanced blood signal and CNR of myocardial enhancement at 3T. J Magn Reson Imaging. 2006, 23: 81-86. 10.1002/jmri.20471.View ArticlePubMedGoogle Scholar
- Kellman P, Epstein FH, McVeigh ER: Adaptive sensitivity encoding incorporating temporal filtering TSENSE. Magn Reson Med. 2001, 45: 846-852. 10.1002/mrm.1113.View ArticlePubMedGoogle Scholar
- Cerqueira MD, Weissman NJ, Dilsizian V, Jacobs AK, Kaul S, Laskey WK, Pennell DJ, Rumberger JA, Ryan T, Verani MS: Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: a statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association. Circulation. 2002, 105 (4): 539-542. 10.1161/hc0402.102975.View ArticlePubMedGoogle Scholar
- Henkelman RM: Measurement of signal intensities in the presence of noise in MR images. Med Phys. 1985, 12: 232-233. 10.1118/1.595711.View ArticlePubMedGoogle Scholar
- Constantinides CD, Atalar E, McVeigh ER: Signal-to-noise measurements in magnitude images from NMR phased arrays. Magn Reson Med. 1997, 38: 852-857. 10.1002/mrm.1910380524.PubMed CentralView ArticlePubMedGoogle Scholar
- Reeder SB, Wintersperger BJ, Dietrich O, Lanz T, Greiser A, Reiser MF, Glazer GM, Schoenberg SO: Practical approaches to the evaluation of signal-to-noise ratio performance with parallel imaging: application with cardiac imaging and 32-channel cardiac coil. Magn Reson Med. 2005, 54: 748-754. 10.1002/mrm.20636.View ArticlePubMedGoogle Scholar
- EVR Di Bella DP, Sinusas A: On the dark rim artifact in dynamic contrast-enhanced MRI myocardial perfusion studies. Magn Reson Med. 2005, 54: 1295-1299. 10.1002/mrm.20666.View ArticleGoogle Scholar
- Plein S, Ryf S, Schwitter J, Radjenovic A, Boesiger P, Kozerke S: Dynamic contrast-enhanced myocardial perfusion MRI accelerated with k-tSENSE. Magn Reson Med. 2007, 58: 777-785. 10.1002/mrm.21381.View ArticlePubMedGoogle Scholar
- Lustig M, Donoho DL, Pauly JM: Sparse MRI: The Application of compressed sensing for rapid MR imaging. Magn Reson Med. 2007, 58: 1182-1195. 10.1002/mrm.21391.View ArticlePubMedGoogle Scholar
- Conolly S, Nishimura DG, Macovski A, Glover G: Variable-rate selective excitation. J Magn Reson. 1988, 78 (3): 440-458.Google Scholar
- Hargreaves BA, Cunningham CH, Nishimura DG, Conolly SM: Variable-rate selective excitation for rapid MRI sequences. Magn Reson Med. 2004, 52: 590-597. 10.1002/mrm.20168.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.