Myocardial perfusion imaging by first-pass contrast-enhanced cardiovascular magnetic resonance was introduced in 1990, when Atkinson et al. first used inversion recovery gradient-echo imaging after injection of a bolus of gadolinium-based contrast (Gd-DTPA) to observe contrast agent transit through the cardiac chambers and myocardium.¹ Myocardial perfusion CMR was examined in animal experiments during pharmacologic vasodilation and showed a good correlation with microspheres for the assessment of blood flow.²,³ and ⁴ Early data from small studies showed that CMR can be a useful tool for the noninvasive assessment of myocardial perfusion reserve (MPR) and allows an accurate detection of significant myocardial ischemia when compared with x-ray coronary angiogram.⁵,⁶ The technique has ongoing rapid development along with experimental validation and clinical evaluation. Recently, it has been validated in different large single and multicenter clinical trials against nuclear myocardial perfusion imaging.⁷,⁸ and ⁹ Technical developments have also occurred in terms of gradient systems, field strength, radiofrequency coil arrays, pulse sequence design, contrast agents, and perfusion analysis methods.

With the advent and widespread availability of faster MRI scanners with very high spatial resolution (approximately 1 × 1 mm), CMR has become an established method for myocardial perfusion imaging. Compared with SPECT and PET, CMR perfusion offers the advantages of a noninvasive and radiation-free assessment, as well as higher spatial and temporal resolution images. In this chapter, we review the different CMR techniques used in myocardial perfusion imaging.

It is important to inform patients about the steps of the test and that they need to hold breath when asked to do by the examiner and the possible side effects of the vasodilator agent and to report symptoms when they feel them. Also, it is important to notify patients before coming to the test that no food and drinks containing xanthines (tea, coffee, chocolate, and cola) may be consumed for at least 12 to 24 hours prior to the examination. Aminophylline, theophylline, and other xanthine medicines are competitive adenosine antagonists and should be avoided for 24 hours prior to the examination. Guide to intervention with adverse reactions (Table 19.2).

Dynamic contrast-enhanced first-pass perfusion is ideally acquired as dynamic series of images and displayed as a movie. It is a very challenging technical process to make this dynamic series of images demonstrating a motion-free cross section of cardiac muscle whose signal intensity rises and falls over time as the bolus of contrast agent passes through the myocardial tissue.

Standardized first-pass perfusion protocol has been proposed by the society of cardiac MRI (SCMR) as follows¹⁰ (Fig. 19.3):

Scout imaging as per LV structure and function module aiming to plan a true short-axis (SA) view of the LV. For ischemia evaluation, data must be acquired for every heartbeat to visualize the flow of the contrast agent through the LV cavity and the myocardium. A minimum of three SA slices (apical, mid, and basal) should be acquired every heartbeat to ensure coverage of all except the most apical LV segment. Additional SA or long-axis slices may also be useful. The slice locations are timed to different points of time (i.e., phases) of the cardiac cycle, but each location is acquired repeatedly at the same phase. To obtain three SA, the four- and twochamber views are used. First, align five slices on the four- and twochamber views and adjust the slice gap so that the first slice is in the plane of the mitral valve and the last slice is at the apex. Then, remove the first and fifth slices by changing the number of slices from five to three. Slice thickness should be 8 mm, in-plane resolution ≤3 mm, and readout temporal resolution between 100 and 125 milliseconds or shorter as available (Fig. 19.4).

First-pass perfusion images must be T1 weighted in order to maximize the effect of the contrast agent on signal intensity. The fast (turbo-) gradient-echo and hybrid echo-planar imaging sequences described below employ small flip angles and very short time of repetition (TR), in order to reduce acquisition time, resulting in poor T1 contrast. While the balanced steady-state free precession sequence uses a higher flip, angle is weighted by the ratio of T2/T1. To overcome all of this, a preparation pulse is applied prior to the readout pulse sequence with an adequately long preparation pulse delay to establish a high T1 contrast prior to the readout sequence is employed.

To accelerate perfusion imaging and improve spatial resolution, parallel imaging approaches such as sensitivity encoding (SENSE) are used.¹⁴,¹⁵ Parallel imaging depends on the spatial difference of radiofrequency coil sensitivities to fill the k-space and accelerate the image reconstruction process. Nevertheless, practically, this is limited to twofold acceleration due to associated artifact and noise drawback.¹⁶ Another parallel k-t acceleration technique, known as kt-BLAST factor 5, with 11 training profiles has been shown to provide greater acceleration than SENSE and have been proposed more recently as a useful technique to improve the spatial resolution of perfusion imaging even further while preserving temporal resolution and cardiac coverage.¹⁷

Recently, novel techniques have been proposed for rapid acquisition of data. The main idea is to collect data in the k-space in radial spokes or spiral approach rather than the Cartesian approaches. A multislice “real-time” ungated data acquisition method is an example of the radial pulse sequence and has been proposed by Harrison et al.¹⁸ The advantage of this method is the fast acquisition of data providing a number of perfusion images adequate for diagnosis without using ECG gating signals and during free breathing. Therefore, it can be potentially used in patients who have arrhythmias, have difficulties of gating, or are unable to perform adequate breath holding. In this
method, a saturation recovery radial turboFLASH sequence is used, and 20 or 24 radial k-space radial rays in four subsets of five to six rays each are obtained for each slice. Four to five slices, according to the number of rays used, can be acquired after a single saturation pulse and a 40-millisecond delay. Each image is acquired in a few milliseconds (40 to 60 milliseconds) and is repeated continuously, roughly four times per second without gating and during free breathing. This means that each slice is acquired at different cardiac phases in every heartbeat. Image reconstruction from undersampled radial k-space data occurs by repetitive compressed sensing tool via spatial and temporal total variation constraints.¹⁹ A retrospective image-based gating method is used, which is similar to a regional sum method applied to self-gate cardiac cine data.²⁰ Therefore, after reconstruction, the ungated images can be processed to self-gate into systole and diastole image sets. Subsequently, deformable image registration is used to correct for cardiac and respiratory motion remaining in systolic and diastolic images.²¹ In comparison to conventional ECG-gated perfusion techniques, this method is efficient in terms of providing maximal information quickly enough for dynamic contrast-enhanced first-pass perfusion imaging; however, it has been assessed over a small number of patients, and larger trials on patients with ischemic heart disease are required to validate this technique for CMR perfusion. Linogram and PROPELLER are variable methods of radial acquisition. The explanation of these methods falls outside the scope of this chapter, but the interested reader is referred to the respective bibliographical references.²²,²³

Spiral k-space scanning is another fast data acquisition technique that allows a substantial speed of perfusion imaging compared to conventional techniques. In a spiral acquisition, each k-space line follows a spiral trajectory with variable number of interleaves (Fig. 19.7); this is valuable in terms of achieving more coverage of k-space with fewer shots than Cartesian. Spiral acquisition fills a circular area of k-space and neglects the corners of k-space, which are collected with a Cartesian acquisition. Simply, the technique consists of defining the required number of interleaves, sampling time, number of points per trajectory, nominal field of view, maximum gradient and slew rate parameters, and a function w(t) describing the relative density as a function of time where (one corresponds to Nyquist sampling density) to determine the desired k-space trajectory. The details of this technique fall outside the scope of this chapter, but the interested reader is referred to the respective bibliographical references.²⁴,²⁵ It is thought that spiral pulse sequences produce high-quality first-pass perfusion images with minimal dark-rim and off-resonance artifacts, high SNR and contrast-to-noise ratio (CNR), and good delineation of resting perfusion abnormalities.²⁶