Cardiovascular cine imaging and flow evaluation using Fast Interrupted Steady-State (FISS) magnetic resonance

Existing cine imaging techniques for the cardiovascular system rely on balanced steady-state free precession (bSSFP) or spoiled gradient-echo (sGRE) readouts, each of which has significant limitations. For instance, fat appears bright with cine bSSFP. As a result, small-caliber embedded structures such as the coronary and internal mammary arteries are obscured by artifacts from fat/water chemical shift. Cine bSSFP is also susceptible to artifact from rapid through-plane flow and banding artifact from off-resonance effects. Alternatively, cine sGRE suffers from inferior signal-to-noise ratio and temporal resolution, as well as considerable in-plane flow saturation.

We hypothesized that a prototype cine imaging technique, radial fast interrupted steady-state (FISS) [1], could overcome these limitations. The technique was compared to standard Cartesian cine bSSFP in a small group of healthy subjects to evaluate cardiac function, coronary artery conspicuity, and aortic valve morphology. Given its advantageous properties, we further hypothesized that the cine FISS technique, in combination with arterial spin labeling (ASL), could provide an alternative to phase contrast for visualizing and quantifying in-plane flow within the aorta and branch vessels.

The study was IRB-approved, and subjects provided written informed consent. Eight healthy subjects (5 male, age = 24 to 54 years) were imaged at 1.5 T (MAGNETOM Avanto, Siemens Healthineers, Erlangen, Germany). Following standard localizer scans, breath-hold cine images were acquired in the left ventricular (LV) 2-chamber, 3-chamber, and 4-chamber views, as well as obliquely through the aortic valve.

The RR intervals during the CMR examinations ranged from approximately 685 msec to 1225 msec. There was no significant difference between Cartesian cine bSSFP and cine FISS in the calculated biplane LV ejection fraction (67.5% ± 4.3% vs. 68.3% ± 3.6%, p = NS) nor in qualitative image ratings for the heart (4.0 ± 0.0 vs. 4.0 ± 0.0, p = NS). Cine FISS showed much greater suppression of epicardial fat signal in all subjects, as well as reduced signal from subcutaneous fat (RV-to-epicardial fat blood pool CNR = 40.6 ± 11.4 (mean ± standard deviation) for cine FISS vs 12.7 ± 10.5 for cine bSSFP, p = 0.002; RV-to-subcutaneous fat blood pool CNR = 42.5 ± 10.8 for cine FISS vs 0.7 ± 8.4 for cine bSSFP, p < 0.001). Banding artifacts in the subcutaneous tissues were consistently less apparent with cine FISS compared with cine bSSFP (Fig. 1; see Additional file 1: Figure S1).

Compared with cine bSSFP, cine FISS significantly improved visualization of the coronary arteries (coronary artery conspicuity = 4.0 ± 0.0 for cine FISS vs. 2.6 ± 0.5 for cine bSSFP, p = 0.019) (Fig. 2; see Additional file 1: Figure S2).

For imaging of the aortic valve using a 6-mm slice thickness, image quality was not significantly different for cine FISS, Cartesian and radial cine bSSFP (4.0 ± 0.0 vs. 3.5 ± 0.5 vs. vs. 3.6 ± 0.5, respectively, p = NS). Conspicuity of the aortic valve leaflets was maximized by imaging with 2-mm thick slices using cine FISS, whereas thin-slice imaging resulted in increased artifacts using either Cartesian or radial cine bSSFP (Fig. 3; see Additional file 1: Figure S3). Aortic valve conspicuity values were 4.0 ± 0.0 for cine FISS versus 2.3 ± 0.5 for radial cine bSSFP and 2.4 ± 0.7 for Cartesian cine bSSFP (p = 0.001).

Dynamic flow patterns were well shown in the aorta, coronary and renal arteries using cine FISS ASL (Figs. 4, 5 and 6; see Additional file 1: Figure S4 and S5). The labeled bolus could be reliably visualized in subtracted images over the entire cardiac cycle as it traversed the length of the vessel. In contrast, the bolus could only be reliably visualized over a few cine frames in non-subtracted images (see Additional file 1: Figure S6).

Flow velocity measurements in the pulsatile flow phantom showed excellent correlation (r² = 0.997, p = 0.001) between cine FISS ASL and 2D cine phase contrast (Fig. 7; see Additional file 1: Figure S7).

In healthy subjects, there was excellent correlation between maximal aortic flow velocities measured by cine FISS ASL and 2D phase contrast (r² = 0.959, p < 0.001). Mean coronary flow velocity, measured with cine FISS ASL over a ≈ 209 ± 97 msec (mean ± sd) span of diastole was 11.7 ± 3.0 cm/s. The cine FISS ASL contrast-to-noise ratio between the coronary artery and background was 16.5 ± 6.1.

FISS is a recently described steady-state imaging technique that has several potential advantages for cine imaging of the cardiovascular system, including: (1) pronounced suppression of fat signal; (2) elimination of through-plane flow artifacts; and (3) reduction in banding artifacts caused by off-resonance effects.

Conventional fat suppression techniques are not useful for cine bSSFP, since repeatedly interrupting the steady-state signal to apply a chemical shift-selective RF pulse can produce severe ghosting artifacts [3]. A more promising approach for fat-suppressed cine imaging is to use a periodically interrupted steady-state pulse sequence like S5FP [4] or FISS [1]. These imaging techniques demonstrate similar signal to bSSFP for on-resonant spins, but wider notches of suppression for off-resonant spins. The wider notches result in improved fat suppression, reduced banding artifact and decreased signal from off-resonant tissues located at the edges of the field of view caused by imperfect shimming.

With cine FISS, coronary artery conspicuity is greatly improved throughout the cardiac cycle because the suppression of fat signal negates chemical shift artifact at the boundary between the vessel wall and surrounding epicardial fat. The ability to rapidly image the coronary arteries throughout the cardiac cycle might add diagnostic value to conventional static navigator-gated 3D coronary CMR angiography [5] in certain cases. For instance, it might be used to dynamically asses the severity of kinking at the origin of a potentially malignant coronary anomaly, or to demonstrate phasic narrowing of a coronary bridge.

Cine bSSFP is prone to artifacts from rapid through-plane flow, which can be attributed to: (1) failure of rapidly flowing spins to attain a steady-state in the brief interval they reside within the thin slice; and (2) mislocalized signal due to steady-state magnetization persisting after the flowing spins have left the slice [6, 7]. For cross-sectional imaging of the aortic valve, we found that bSSFP flow artifacts were negligible for 6-mm thick slices but became significant when the slice thickness was reduced to 2-mm, resulting in decreased conspicuity of the valve leaflets. In contrast, the aortic valve leaflets were sharply delineated using thin-slice cine FISS. While such thin slices are not routinely used for cardiac imaging, they might be helpful in situations where high spatial resolution is needed, e.g. for precise multi-phase CMR measurements of the aortic valve apparatus in patients who are scheduled for transcatheter aortic valve replacement (TAVR) [8].

Flow imaging in the cardiovascular system is currently performed using phase contrast MRI [9]. Although 4D approaches are under active investigation, breath-hold 2D cine phase contrast is the mainstay of clinical practice and allows rapid, through-plane flow quantification in the heart and great vessels. A mechanistically-distinct alternative approach for flow imaging involves the use of ASL [10], which provides a useful quantitative tool for measuring cerebral blood flow and other perfusion indices without the need for contrast infusion [11]. In addition, ASL techniques can be used to evaluate arterial flow patterns, particularly when a cine readout is incorporated [12]. In recent years, most research efforts using cine ASL have focused on the brain, e.g. to evaluate collateral flow in the circle of Willis or flow dynamics in arterio-venous malformations [13‐15]. However, cine ASL has not to our knowledge been used to measure arterial flow velocities.

The basic principle of flow velocity measurement using cine FISS ASL is fundamentally different from phase contrast. Whereas phase contrast measurements depend on flow-induced phase shifts, cine FISS ASL relies on the frame-to-frame bulk displacement of labeled spins, which translates in a straightforward way into in-plane flow velocity. Unlike phase contrast, flow velocity measurements with cine FISS ASL are relatively free from partial volume effects – a benefit of the extreme level of background signal suppression and resistance to off-resonance effects. Moreover, they are largely unaffected by gradient-induced eddy currents or Maxwell field effects that are sources of measurement error with phase contrast [16]. In principle, accurate in-plane flow velocity measurements are obtainable so long as temporal resolution and arterial conspicuity are sufficiently high. In the current study, high radial undersampling factors were used to achieve cine frame rates as high as 50 Hz in a single breath-hold acquisition. In-plane cine FISS ASL flow velocity measurements correlated well with through-plane phase contrast measurements in a pulsatile flow phantom and the aorta. Moreover, the use of a thick slice (up to 48-mm in our study) with cine FISS ASL allowed semi-projective imaging of blood flow, which is not possible with 2D cine phase contrast techniques due to partial volume averaging of flow-induced phase shifts in the artery with background phase shifts in static tissues. Using cine FISS ASL, one can dynamically visualize blood flow along the entire length of both renal arteries including intra-renal branches with a single thick-slice breath-hold acquisition (Fig. 5; see Additional file 1: Figure S5). This allows direct comparison of flow patterns and velocities in the left and right renal arteries, which might be useful for determining the hemodynamic significance of a renal artery stenosis. In addition to renal artery stenosis, the semi-projective approach has potential clinical utility in a variety of other vascular disorders. For instance, using cine FISS ASL one could rapidly evaluate flow patterns in the pulmonary arteries (see Additional file 1: Figure S8) to identify occluded branches in a patient with suspected pulmonary embolism.

Our initial empirical experience with semi-projective imaging suggests that using an excitation RF pulse with a large time-bandwidth product [17] is key to preserving arterial detail in thick-slice acquisitions, presumably because doing so maximizes the slice-select gradient and helps to overcome intravoxel dephasing from local field inhomogeneities. Alternatively, one can create a maximum intensity projection from several overlapping thin-slice cine ASL acquisitions, although this approach is less efficient and risks misregistration artifact.

In conclusion, cine FISS demonstrates several benefits for cardiovascular imaging compared with cine bSSFP, including better suppression of fat signal and reduced artifacts from through-plane flow and off-resonance effects. The main drawback is a slight (~ 20%) decrease in temporal resolution. In addition, preliminary results suggest that cine FISS ASL provides a potential alternative to phase contrast techniques for in-plane flow quantification, while enabling an efficient, visually-appealing, semi-projective display of blood flow patterns throughout the course of an artery and its branches.