Estimating central blood pressure from aortic flow: development and assessment of algorithms

Figure 1. Study methodology. 1) Central blood pressure (cBP) estimation algorithms consisted of three steps. A: clinical data acquisition and preprocessing: blood flow measured at the ascending and descending [one-dimensional (1-D) algorithm only] aorta; peripheral blood pressure (BP) measurement; and aortic anatomy (1-D algorithm only). B: cardiovascular (CV) parameters estimated from clinical data. C: these parameters, along with the noninvasive clinical data, were used as inputs to one of three cBP models. 2) Algorithm performance was assessed by comparing cBP estimates provided by each model to reference values.

Figure 2. Clinical central blood pressure (cBP) wave morphologies: (left) aortic coarctation dataset (obtained invasively), (middle) normotensive (noninvasive) dataset, and (right) hypertensive (noninvasive) dataset. Black lines illustrate a random patient’s cBP waveform. Shaded regions represent the range of cBP waves within each dataset.

Figure 3. Generating datasets of virtual subjects. A, top: A range of values for each cardiovascular (CV) parameter was obtained from the clinical literature for healthy individuals (see Table A1). A, bottom: the thick line illustrates the flow wave corresponding to the baseline values of stroke volume (SV) and heart rate (HR), and the shaded region represents the range of flow waves corresponding to all SV and HR variations. B: two reduced-order models were used to generate central blood pressure (cBP) waves. C: cBP waves generated by each model: black lines illustrate the cBP wave corresponding to the baseline set of parameter variations, and shaded regions represent the range of cBP waves within each dataset.

Figure 4. Bland–Altman plots for the optimal cardiovascular (CV) parameter estimation methods. They were obtained from all one-dimensional (1-D) dataset waves using the clinical scenarios carotid + (top) and carotid– (bottom).