**3. Advanced hemodynamic assessment in AF using 4D-flow**

Four-dimensional flow (4D-flow) has been developed to achieve a comprehensive acquisition of blood flow through the heart [32, 33]. Phase-contrast flow-encoding acquisitions are performed in all three spatial dimensions of space and time along the cardiac cycle (3D + time = 4D). This technique has existed since earlier 1990s; however, computational and hardware limitations limited its clinical applicability [34]. During the last decade, both computational power and hardware development have allowed for a realistic integration into clinical settings. In recent years, a scientific consensus and acquisition recommendations have been published with the purpose of standardizing the acquisition and analysis of 4D-flow data [35–37]. Acquisition parameters are optimized to provide the best possible imaging accuracy in each protocol. For AF imaging, the slab acquisition can be sagittal or axial with whole-heart coverage using isotropic spatial resolution (2–2.5 mm). Retrospectively ECG-gated acquisition with 30 phases is advised for adequate coverage of the cardiac cycle. A respiratory navigator can be used to reduce respiratory motion. Acceleration methods, such as parallel imaging or compressed sensing, can be used to achieve an acquisition time between 5 and 10 min. A sample of acquisition is illustrated in **Figure 2**.

There have been efforts to quantify 4D-flow measurements to improve diagnosis and evaluation of disease and risk assessment of AF (**Table 1**).

An initial application of 4D-flow is the generation of phase-contrast angiogram (PC MRA), which can be obtained by multiplying the velocity magnitude and the cine magnitude volumes. One of the advantages of PC MRA versus the CE MRA is that it can be obtained without contrast and allow us to obtain comparable anatomic characterization for the pulmonary veins (**Figure 3**). One limitation is the limited characterization of the LAA. The primary method of characterizing hemodynamics in LA, from 3D blood flow velocities, is to take an average from all voxels through a whole

#### **Figure 2.**

*Acquisition planning of 4D-flow. Region of interest covers the whole heart, as illustrated by the blue rectangle. The acquisition requires electrocardiogram gating and respiratory control, as it is shown by the small orange rectangle. Velocity encoding in each direction of the volume of interest is used to obtain velocity phases, which are subtracted from encoding reference to calculate blood flow velocities within the volume (X, Y, Z). The cardiac cycle average magnitude facilitates the anatomic visualization of the heart.*


*LA: left atrium; LAA: left atrial appendage; AF: atrial fibrillation; AF-sinus: previous history of AF, but in sinus rhythm at time of imaging; AF-afib: in AF at the time of imaging; KE: kinetic energy; CHA2DS2-VASc: stroke risk stratification system that accounts for the patient history of congestive heart failure, hypertension, age > 75 years, diabetes mellitus, stroke, vascular disease, age between 64 and 75 years, and sex.*

#### **Table 1.**

*Summary of 4D-flow studies on AF. Flow parameter comparisons between healthy controls and AF patients have revealed consistent differences, such as reduced flow velocity and increased stasis. In recent years, the interest in novel flow parameters and association with risk factors are grown.*

cardiac cycle or peak velocity. Although there is some contradiction between studies, most of the recent studies characterizing AF blood flow with relatively large cohorts agree that there is a significant decrease in mean and peak flow velocity in LA, even in paroxysmal AF patients with sinus rhythm [39–42, 44]. Most notably, the increase *Cardiovascular Magnetic Resonance Imaging of Atrial Fibrillation: An Advanced… DOI: http://dx.doi.org/10.5772/intechopen.105498*

#### **Figure 3.**

*Comparison of contrast-enhanced angiogram and phase-contrast angiogram in a patient with atrial fibrillation. Contrast-enhanced (CE) acquisition requires the use of an agent to depict the left atrial anatomy. Phase-contrast (PC) angiograms derived from 4D-flow do not require the use of a contrast agent. CE angiogram benefits from a higher spatial resolution (1 mm isotropic) than PC angiogram (~2.5 mm isotropic). Both the acquisitions are useful to characterize the pulmonary vein structure.*

in CHA2DS2-VASc score has been associated with reduced mean LA velocity [40, 41], which suggests that 4D-flow measurement may be able to improve risk assessment.

Kinetic energy, which is proportional to the mean square of velocity, was also markedly lower in AF patients than in controls [44]. Similarly, energy loss is also reduced (**Figure 4**). Left atrial flow stasis map proposed by Markl et al. [39] focuses on the flow

#### **Figure 4.**

*Kinetic energy and energy loss. On the left, a control quantification sample of energy loss (EL) and kinetic energy (KE) in the left atrium (LA). On the right, a preablation atrial fibrillation (AF) patient. Local energy differences are mostly generated from the pulmonary veins during inflow, as it is showed by the red regions. RA: right atrium; RV: right ventricle; LV: left ventricle.*

#### **Figure 5.**

*Left atrial stasis maps. Sample of a patient 3D stasis map pre- and postablation showing regions with elevated stasis (red arrows). Larger stasis regions are indicators of possible thrombus formation. The region near the left atrial appendage typically shows elevated stasis after ablation.*

stagnation at individual voxel. This method counts the number of time frames under threshold velocity (0.1 m/s) at each voxel, which is supposed to increase the chance of thrombosis. The result can be shown as a map projected on the MRA image as well as the average ratio relative to the one whole heart cycle. Several studies have consistently reported flow stasis to be elevated in AF patients both in sinus rhythm and in fibrillation [39, 41, 42, 45]. An example of flow stasis is displayed in **Figure 5**.

In addition, flow patterns through the pulmonary vein into the LA have been studied [42]. The 3D asymmetrical configuration of the systemic pulmonary veins allows the development of vortical flow patterns during early diastolic LV filling while avoiding/reducing blood stasis [42]. AF patients often show LA inflow fragmentation and vortex formation in the LA (see **Figure 6**). Increment of vortex size can be observed in paroxysmal AF, and it is associated with higher risk score. Similarly, decreased LA velocity and increased LA blood flow stasis have also been reported in the LAA [41, 44]. However, 4D-flow MRI special resolution may not facilitate the accurate segmentation the LAA [35]. Despite the latter, 4D-flow parameters have shown excellent reliability and reproducibility in AF patients [45]. LA peak velocity and vorticity were found to be more reproducible and independent of physiological biomarkers than LA mean velocity, LA vortex volume, and blood flow stasis.

#### **Figure 6.**

*Evolution of atrial vortex formation in atrial fibrillation. A vortex typically forms during left atrial inflow and tends to disappear during ejection. However, in atrial fibrillation patients, small vortices remain in the atrium during the cardiac cycle.*

*Cardiovascular Magnetic Resonance Imaging of Atrial Fibrillation: An Advanced… DOI: http://dx.doi.org/10.5772/intechopen.105498*

There was an approach to associate risk factors with LA flow characteristics [46]. This study presented patients with moderate to high CHA2DS2-VASc scores have impaired LA flow parameters even though they have restored from arrhythmia or have no AF history. Recently, a novel sequence to evaluate 3D hemodynamics with a fully self-gated and free-running sequence, called 5D flow, has been proposed [47]. This method extracts cardiac and respiratory signals from SI projection signals, eliminating the need for ECG gating, and adds the respiratory phase as the fifth dimension by sorting acquired scan lines according to respiratory signal. This method can be extended to be used in AF patients by replacing the respiratory dimension with the RR length dimension to tackle the variability of arrhythmic heartbeats [48]. The study successfully found a correlation between flow parameters and AF burden with reasonable scan time (<10 min).

#### **4. Conclusion**

In conclusion, advances in cardiac magnetic resonance imaging can facilitate the assessment of cardiac function and left atrial structure. This chapter aimed to introduce a standard cardiac MRI protocol for atrial fibrillation. Advanced hemodynamics using 4D-flow can improve the assessment of the left atrium flow patterns and efficiency throughout the cardiac cycle. Novel flow biomarkers such as 3D stasis, kinetic energy, or vortex formation may unmask the presence of LA/LV disease in atrial fibrillation.

#### **Acknowledgements**

The authors were supported by The University of Calgary, URGC SEM #1054341 and JG start-up funding. Research unrestricted funding was also provided by The Libin Cardiovascular Institute and Siemens Healthineers. JG acknowledges the Natural Science and Engineering Research Council of Canada, grants RGPIN-2020- 04549 and DGECR-2020-00204.

#### **Conflict of interest**

The authors have no conflict of interest.

*Atrial Fibrillation - Diagnosis and Management in the 21st Century*
