With cine GRE, multiple images at the same slice position are generated to capture different time points in the cardiac cycle. The terminology to describe these images can be confusing. For clarity, the individual images are referred to as frames, rather than “cardiac phases” (Figure III4-1).

The multiple frames of a cine GRE sequence are played in a cinematic loop to show a video of the beating heart. The temporal resolution and number of frames define how smooth or jerky the cine loop appears. The temporal resolution of the cine GRE acquisition is usually defined as the duration of the cardiac cycle that each frame represents. For example, if each frame is acquired in 50 msec intervals after the R wave, then the temporal resolution is 50 msec. If MR data are acquired for 700 msec of each heart beat, then the number of frames will be 14. Higher-temporal-resolution images are needed for more accurate assessment of cardiac motion and function, particularly during systole. As will be discussed later in this chapter, some images can be generated by interpolation (view sharing or echo sharing in k-space). The true temporal resolution of interpolated data is more complicated to determine.

ECG-gated sequences acquire gradient echoes repeatedly after each R wave for the entire duration of the acquisition window. The manner in which the echoes are partitioned or segmented into different k-spaces determines the temporal resolution of the images and also the total acquisition time. Each k-space corresponds on a one-to-one basis to a specific frame (Figure III4-2). Generally, only a small proportion of the data needed to fill each k-space is collected in a single heartbeat. The amount of data needed to fill each k-space determines the spatial resolution of the images (the number of phase-encoding steps). The number of heartbeats needed to fill a given k-space determines the total acquisition time for the sequence.

Because the k-space data for each frame are collected across a number of heartbeats, the image quality is strongly dependent on the consistency of cardiac motion and gating from beat to beat.

The quality of ECG-gated images, particularly cine GRE images, relies not only on having a reasonable temporal resolution, but also, importantly, on the consistency of data acquisition with respect to the ECG. For example, the MR system assumes that the cardiac position immediately after the R-wave is the same for all heartbeats. Therefore, when echoes are acquired at this time point over multiple beats, the resulting image should be a perfect representation of the heart at that point in time. The system similarly assumes that the heart position at, for example, 240 msec and 560 msec after the R wave is the same from beat to beat as well. However, with irregular rates or premature beats, this may not be true (Figure III4-3). Combining data that reflects the heart in different states of contraction will degrade image quality substantially.

For cardiac MR pulse sequences, acquisition times are usually a multiple of heartbeats rather than of seconds. The acquisition times therefore depend on a subject’s heart rate. Because most cardiac MR examinations require cine acquisitions in multiple planes of the heart, it is vital that cine gradient echo acquisition times be short enough for a comfortable breath hold. A subject’s breath-holding capacity is usually thought of in seconds (and not heartbeats!). Most healthy individuals can suspend respiration for 20 sec. Those with moderate cardiac or respiratory ailments are usually still able to hold their breath for 10 sec, particularly if supplemented with oxygen via nasal cannula. If acquisition times exceed the breath-holding capacity of subjects, then the ECG-gated images should be performed during free breathing with 3-4 signals averaged, thereby lengthening the acquisition times by a factor of 3-4.

For all cine GRE images, acquisition parameters can be modified to balance the tradeoffs between temporal resolution, spatial resolution, and acquisition times (Figure III4-4). Fundamentally, the parameter that constrains all three is the minimum time needed to generate each gradient echo, reflected in the minimum TR, and this varies depending on the MR system and the pulse sequence used.

In general, ideal temporal resolution of sequences performed for assessment of cardiac function should be 50-60 msec or less, depending on the subject’s heart rate. The faster the heart rate, the better the temporal resolution needed to resolve the motion of the heart during systole. If temporal resolution is inadequate, the cine images are likely to underestimate left ventricular contractility.

The spatial resolution of cine GRE images depends on the structures that need to be resolved. For studies of overall cardiac function, an in-plane spatial resolution of 2-2.5 mm is probably sufficient. Higher spatial resolution, such as 1-2 mm, helps to define finer structures such as cardiac valves, a patent foramen ovale, or smaller vessels such as coronary arteries or bypass grafts. Typical imaging fields of view are about 250 mm × 350 mm (allowing for rectangular field of view for short axis views). Thus for a standard 100-128 × 256 matrix, the in-plane spatial resolution is usually approximately 2-2.5 mm × 1.4 mm. A 192 × 256 matrix provides superior resolution of 1.3 × 1.4 mm.

Most vendors preset sequences with a minimum TR based on the desired image quality and contrast of a sequence. After TR is fixed, the next consideration in setting up a cine GRE sequence is usually the acquisition time, constrained by the subject’s breath-holding capacity. For a given TR and acquisition time, parameters of a cine GRE sequence are selected based on the desired balance between spatial and temporal resolution (Figure III4-4).

Cine GRE sequences are designed to evaluate flow and function of the myocardium and vessels. They are typically referred to as bright-blood sequences, because the signal intensity of the blood is bright relative the myocardium and vessel wall. How this image contrast is achieved varies with the type of pulse sequence. With standard spoiled gradient echo sequences, the image contrast is based on the time-of-flight phenomenon, similar to that discussed in Chapter II-2. Repeated RF pulses cause saturation of the stationary tissues, while the inflow of fresh, unsaturated protons in moving blood leads to its relatively high signal intensity. To allow time for the inflow of moving blood, the TR of these gradient sequences is on the order of 8 msec or more. Even with a TR of 8 msec, if blood flow is slow, signal of the blood pool can become saturated and the contrast between myocardium and blood pool reduced.

Because of the need to allow time for inflow, these sequences are not able to take advantage of the ultrashort TRs possible with newer MR systems. Steady-state free precession sequences, as described in Chapter I-4, produce images with high contrast between blood and myocardium based on their T2 (and T1) differences (Figure I4-19), independent of flow. These sequences benefit from shorter TRs and TEs, which in turn translate into shorter acquisition times or higher-spatial-resolution images in equivalent acquisition times.

More details about the different methods used to produce cine GRE images are outlined in the following sections and summarized in Table III4-1.

With conventional cine GRE, the temporal resolution is equal to the TR for the sequence. TR is usually about 50-60 msec, so that 10-15 frames are produced. The flip angle is selected based on image contrast and the TR. A typical flip angle of 15-20° gives a good balance between the amount of signal in the image and the image contrast between blood and myocardium. Because the image acquisition time is directly proportional to the number of phase-encoding steps, rectangular field of view is valuable for reducing acquisition times without sacrificing spatial resolution.

The choice of views per segment offers an additional flexibility in planning a cine GRE acquisition. It is selected based on the desired relationship between spatial
resolution, temporal resolution, and acquisition time. Typically, a temporal resolution of 50-60 msec is desired, corresponding to about 10-15 frames. If temporal resolution is too low (for example, if only 6-8 frames are acquired at 100 msec intervals), then the information about the heart’s motion might be inadequate for interpretation. To show fine detail such as cardiac valves, ideally, the spatial resolution should be less than 2 mm. Finally, acquisition time is usually kept less than 20 sec for breath holding. With ECG-gated sequences, acquisition time depends on number of heartbeats, and so the actual acquisition time depends on the subject’s heart rate.

Given the interplay of these parameters, how should a sequence be set up based on a subject’s heart rate and breath-holding capacity?

Consider the following two sample subjects: one with a heart rate of 60 beats per minute (A) and another with a heart rate of 90 beats per minute (B). In both cases, the same sequence is used—a prospectively gated segmented k-space cine GRE sequence with true TR = 10 msec. The desired imaging matrix for an axial view is 100 × 256 (assuming recFOV is used). What is the relationship between temporal resolution and acquisition time in these two cases when views per segment is varied?