Pulse Wave Velocity Measurements by Magnetic Resonance Imaging in Neonates and Adolescents: Methodological Aspects and Their Clinical Implications

Figures

• 

  Fig. 1. Image delineations for aortic length and flow measurements. Neonatal black-blood (A) and adolescent white-blood (G) 3D angiography, and adolescent thoracic coronal overview image with white dotted lines outlining the aortic walls throughout the 2D image stack (F), all with length delineation (Δd) (dashed orange) and perpendicular flow imaging planes in the ascending aorta (blue), descending proximal aorta (green) and descending aorta at the level of the diaphragm (red). Magnitude and quantitative phase-contrast images at ascending and proximal descending aorta (B–C, H–I) and descending aorta at the level of the diaphragm (D–E, J–K), corresponding to the flow imaging planes localizations. Worth noting is that both the ascending and descending flow plane in the aortic arch (B–C, H–I) are collected in a single flow plane as in clinical routine which may result in angular error in the descending aorta, whereas the flow plane at the level of the diaphragm is perpendicular

• 

  Fig. 2. Velocity and flow curves. Curves for neonates (top row) and adolescents (bottom row) based on velocity (left column) and flow (right column), all without baseline correction. The difference in curve shape may be caused by redirected blood volume to the carotids and subclavian arteries. This leads to the decreased magnitude of the aortic descending flow curve (right), while the velocity curve is more constant in shape. The difference in curve shape had least impact on the time-to-foot transit method, which may indicate this method’s intrinsic robustness (c.f. Figure 9). Also, note the challenge to assess PWV in neonates using the TTP method as the descending curve peak time-point may end up prior to the ascending curve peak time-point (top left), resulting in negative PWV values. * = time points used for maximum upslope PWV calculation, + = time points used for TTP PWV calculation (c.f. Figure 3 for details on transit time methods)

• 

  Fig. 3. Flow curve analysis. Example of analysis methods demonstrated on a flow-based pulse wave velocity (PWV) curve. Note that all algorithms also apply to velocity-based curves (not shown). Blue line corresponds to flow in the ascending aorta (AAo) and red line to flow in descending aorta at the level of the diaphragm (DAo). Pulse wave velocity was calculated by aortic length divided by transit time (Δd/Δt). Transit time was estimated using three different algorithms. Algorithms and corresponding points on the flow curves are shown for time-to-foot (TTF), maximum upslope (MS), and time-to-peak (TTP). The algorithms were based on the baseline intersection of an automatically generated tangent superimposed at the point of maximum upslope (TTF); the maximum upslope incline (maximum upslope); and the maximum curve magnitude (TTP)

• 

  Fig. 4. Baseline correction methods. Pulse wave velocity curves with no (left), automatic (middle), and manual baseline correction (right). Automatic correction uses the average from the 80th–95th % segment of the flow curve to approximate the ascending and descending curve, marked with a green box. For manual baseline correction, emphasis was instead put on leveling the curves prior to the upslope. Baseline correction was only applicable for the time-to-foot transit method, as it is the only method that relates to the baseline

• 

  Fig. 5. Minimum timeframes per cardiac cycle for neonates and adolescents. Computer phantom based on neonatal (top row) and adolescent (bottom row) flow data for aortic lengths 25 mm (top left, i.e., neonatal aortic arch), 60 mm (top right i.e., neonatal thoracic aorta, and bottom left, i.e., adolescent aortic arch), and 150 mm (bottom right, i.e., adolescent thoracic aorta) and for different pulse wave velocities. The aortic lengths presented above approximates the shortest lengths in the current population. Vertical red dashed lines denote the cut-off for required timeframes per cardiac cycle for all pulse wave velocities. Horizontal gray dotted lines denote the mean error (min–max) after the cut-off. For both shorter aortic lengths and higher pulse wave velocities, the required timeframes per cardiac cycle for accurate pulse wave velocity measurements and the pulse wave velocity error at inadequate temporal resolutions increased. X-axis offset between neonatal and adolescent curves is likely related to the shorter diastolic period in neonates due to their higher heartrate

• 

  Fig. 6. Aortic length measurements. Aortic arch (top row) and thoracic aorta (bottom row) length measurements in neonates (n = 15, left column), adolescents (n = 71, middle column), and intra-individual variation between adolescent coronal overview images and the reference method 3D angiography (n = 49, right column). Solid circles represent males and open circles females. Red lines with error bars denote median (range). Solid black lines represent bias and dashed lines represent 95% limits of agreement

• 

  Fig. 7. Baseline correction comparison. Bland Altman plots for neonates (A; top panel) and adolescents (B; bottom panel) with comparison of baseline correction for thoracic aorta velocity (top row) and flow curves (bottom row). Agreement between no and automatic baseline correction (left), no and manual baseline correction (middle), and manual and automatic baseline correction (right) is presented in both rows. Manual and automatic baseline correction agreed for both velocity and flow curves, which indicates that these baseline correction methods can be used interchangeably in neonates and adolescents. Solid black lines represent bias and dashed lines represent 95% limits of agreement

• 

  Fig. 8. Transit time measurements. Method comparisons for neonates (top row) and adolescents (bottom row) with thoracic aorta PWV for velocity (left) and flow (right) derived time-to-foot with automatic baseline correction, maximum upslope, and time-to-peak algorithms. Time-to-peak was lower for both velocity and flow curves. Negative PWV values are due to curve shapes with no definitive peak, leading to errors in the TTP algorithm (c.f. Fig 2). Note that the TTP algorithm therefore cannot be recommended in neonates and adolescents. Solid circles represent males and open circles females. Red line with error bars denotes mean ± SD

• 

  Fig. 9. Transit time agreement. Bland–Altman plots for neonates (top row) and adolescents (bottom row) with comparison of thoracic aorta velocity and flow-based maximum upslope (left), time-to-foot with automatic baseline correction (middle), and time-to-peak (right) algorithms. For both neonates and adolescents, limits of agreement were narrower for time-to-foot, which may be an indicator of the transit time estimator’s intrinsic robustness. Note the difference in y-axis ranges for neonates and adolescents. Negative PWV values are due to curve shapes with no definitive peak, which worked poorly with the TTP algorithm (c.f. Fig 2). Solid black lines represent bias and dashed lines represent 95% limits of agreement

• 

  Fig. 10. Aortic pulse wave velocity measurements. Aortic arch (top row) and thoracic aorta (bottom row) flow-based time-to-foot PWV with automatic baseline correction measurements in neonates (n = 15, left column) and adolescents (n = 71, middle column) and intra-individual differences between adolescent coronal overview images and the reference method 3D angiography (n = 49, right column). Solid circles represent males and open circles females. Red lines with error bars denote mean ± SD. Solid black lines represent bias and dashed lines represent 95% limits of agreement