Displacement encoding with stimulated echoes (DENSE) magnetic resonance imaging (MRI) is a technique that encodes tissue displacement into the magnetic resonance signal phase [1]. DENSE MRI has primarily been utilized for the assessment of myocardial motion and applied to the identification of markers for early cardiac dysfunction [2,3], the response to cardiac resynchronization therapy [4,5], and the detection of transmural infarction [6,7]. Notably, recent advances in spiral k-space sampling DENSE MRI with custom noise-reducing post-processing allow the imaging and kinematic analysis of thinner structures, such as the aortic wall [[8], [9], [10]]. These improvements have even shown the potential to assess radial, circumferential, and shear strains on the thin walls of large arteries [11]. However, the aortic wall is significantly thinner than the myocardium and associated with smaller displacements, making these measurements more sensitive to the effect of MRI signal noise and error.

Effects from concomitant fields, non-linearity of magnetic gradients, eddy currents, and mechanical field oscillations are potential sources of noise and error on MRI-derived measurements [12]. Signal noise introduces uncertainty to DENSE MRI-derived measures of displacement, strain, and change in volume [13]. The signal-to-noise ratio (SNR) is a commonly employed measure to assess how image-based metrics could be affected by signal noise, typically defined as the ratio of the averaged signal within the region of interest (ROI) to the signal noise. A threshold of SNR < 2 has been proposed to determine if an MRI signal is biased due to noise [14]. Noise can be quantified in vivo by measuring the temporal variation of the signal from static structures (or background). This approach does not require additional scan time, dedicated hardware, or the use of phantoms [15].

The accuracy of DENSE MRI has been validated in both cardiac and aortic phantoms by comparison to tracking of physical markers on in vitro experimental setups [[16], [17], [18], [19]]. However, fewer studies have specifically evaluated the signal noise and potential errors of DENSE MRI. Previous studies of cardiac and ventricular kinematics using DENSE MRI have reported SNR estimations ranging from 10 to 35, a magnitude that depends on the subject, the specific DENSE MRI scanning sequence, and scanning parameters such as magnetization energy, encoding frequency, and flip angle [18,[20], [21], [22]].

To the authors' knowledge, no prior study has addressed the estimation of signal noise and error for aortic DENSE MRI, which is critical, given the relatively lower thickness and displacement of the aortic wall in comparison with the myocardium. Therefore, the goal of this study was to perform a detailed analysis of background noise, SNR, and uncertainty of the encoded displacement field from a 2D cine DENSE sequence for the study of human aortic kinematics, and to evaluate the effectiveness of correction methods for offset error and noise. The analysis is performed on datasets collected from healthy-adult volunteers at three locations of the descending aorta – distal aortic arch (DAA), descending thoracic aorta (DTA), and infrarenal abdominal aorta (IAA) - and on in vitro setups with polyvinyl alcohol (PVA) phantoms.