Abstract
Introduction
Clinical evidence suggests that central (aortic) blood pressure (cBP) is a better cardiovascular (CV) risk indicator than brachial BP [1]. Specifically, an increase in central systolic BP (cSBP) leads to hypertension, which is the single most important cause of disability and mortality worldwide [2]. Current non-invasive approaches to estimate cBP use a generalised mathematical transfer function to estimate the cBP waveform from a peripheral (radial or carotid) pressure waveform [1]. We propose a novel non-invasive approach to estimate cBP from aortic blood flow and geometry. Our algorithms use computational models of arterial haemodynamics which are fully personalised with CV parameters estimated from clinical measurements. The algorithms were developed using an in silico dataset, and validated against invasive cBP from an in vivo dataset of post-aortic coarctation repair patients.
Methods
The in silico dataset contains waveforms representative of 78 healthy adult subjects aged 25-75, with varying cardiovascular properties. It was created using a 1-D model of the largest systemic arteries [3]. The in vivo dataset contains ascending and descending aortic flow and geometry (from MRI), and invasive catheter cBP measurements for 12 post-aortic coarctation repair patients. Each subject’s CV parameters were estimated from their data. These parameters were used with aortic geometry and flow as inputs to two computational models (Figure 1, left): (i) a three-element Windkessel (0-D), and (ii) a 1-D model of the upper-thoracic aorta [4]. Performance was assessed using the mean absolute error (mean error ± standard deviation), MAE, between reference and estimated cSBP for each dataset (Figure 1, right).
Results & Discussion
The cSBP estimation MAEs for the 0-D and 1-D algorithms were, respectively, 5.4 ± 0.7 and 5.6 ± 3.0 mmHg(in silico dataset); and -2.2 ± 7.0 and 0.3 ± 10.2 mmHg (in vivo dataset).ConclusionWe have developed and tested new algorithms for estimating the cBP waveform from phenomena occurring directly in the ascending aorta. Current MAEs are within clinically acceptable ranges (< 6 mmHg). Further development and verification of these methods will be presented at the conference.
References
1. McEniery, C. et al. European Heart Journal. 2014; 35(26), 1719–1725.
2. Williams, B. et al. European Heart Journal. 2018; 39(33), 3021–3104.
3. Charlton, P.H. et al. [submitted to American Journal of Physiology].
4. Alastruey, J. et al. Journal of The Royal Society Interface. 2016; 13(119), 20160073.
Clinical evidence suggests that central (aortic) blood pressure (cBP) is a better cardiovascular (CV) risk indicator than brachial BP [1]. Specifically, an increase in central systolic BP (cSBP) leads to hypertension, which is the single most important cause of disability and mortality worldwide [2]. Current non-invasive approaches to estimate cBP use a generalised mathematical transfer function to estimate the cBP waveform from a peripheral (radial or carotid) pressure waveform [1]. We propose a novel non-invasive approach to estimate cBP from aortic blood flow and geometry. Our algorithms use computational models of arterial haemodynamics which are fully personalised with CV parameters estimated from clinical measurements. The algorithms were developed using an in silico dataset, and validated against invasive cBP from an in vivo dataset of post-aortic coarctation repair patients.
Methods
The in silico dataset contains waveforms representative of 78 healthy adult subjects aged 25-75, with varying cardiovascular properties. It was created using a 1-D model of the largest systemic arteries [3]. The in vivo dataset contains ascending and descending aortic flow and geometry (from MRI), and invasive catheter cBP measurements for 12 post-aortic coarctation repair patients. Each subject’s CV parameters were estimated from their data. These parameters were used with aortic geometry and flow as inputs to two computational models (Figure 1, left): (i) a three-element Windkessel (0-D), and (ii) a 1-D model of the upper-thoracic aorta [4]. Performance was assessed using the mean absolute error (mean error ± standard deviation), MAE, between reference and estimated cSBP for each dataset (Figure 1, right).
Results & Discussion
The cSBP estimation MAEs for the 0-D and 1-D algorithms were, respectively, 5.4 ± 0.7 and 5.6 ± 3.0 mmHg(in silico dataset); and -2.2 ± 7.0 and 0.3 ± 10.2 mmHg (in vivo dataset).ConclusionWe have developed and tested new algorithms for estimating the cBP waveform from phenomena occurring directly in the ascending aorta. Current MAEs are within clinically acceptable ranges (< 6 mmHg). Further development and verification of these methods will be presented at the conference.
References
1. McEniery, C. et al. European Heart Journal. 2014; 35(26), 1719–1725.
2. Williams, B. et al. European Heart Journal. 2018; 39(33), 3021–3104.
3. Charlton, P.H. et al. [submitted to American Journal of Physiology].
4. Alastruey, J. et al. Journal of The Royal Society Interface. 2016; 13(119), 20160073.
| Original language | English |
|---|---|
| Title of host publication | BioMedEng19 Conference Proceedings |
| Place of Publication | London |
| Pages | 108 |
| Number of pages | 1 |
| ISBN (Electronic) | 978-1-9996465-2-3 |
| Publication status | Published - Sept 2019 |
| Event | BioMedEng19 - Imperial College London, London, United Kingdom Duration: 5 Sept 2019 → 6 Sept 2019 https://www.biomedeng19.com/ |
Conference
| Conference | BioMedEng19 |
|---|---|
| Country/Territory | United Kingdom |
| City | London |
| Period | 5/09/2019 → 6/09/2019 |
| Internet address |