Paper Submission & Registration
8th Dutch Bio-Medical Engineering Conference
16:30   Vascular - II
Chair: Richard Lopata
15 mins
Endmember determination of multispectral photoacoustic imaging for carotid plaque vulnerability assessment
Camilo Cano, Min Wu, Marc van Sambeek, Richard Lopata
Abstract: The rupture of vulnerable plaques in the carotid artery is a major cause of stroke. For a better understanding of the pathophysiology of vulnerable plaque a comprehensive characterization of their composition is paramount. Multispectral photoacoustic imaging (MSPAI) can discern the composition and structure of vascular tissue, which provides vital information for vulnerability assessment; however, obtaining unique spectral characteristics of specific chromophores is cumbersome. In this study, we use MSPAI and unmixing techniques to determine plaque constituents. In contrast with conventional linear unmixing, we use piecewise blind endmember detection, which does not require previous knowledge of the sample composition. This way, we can assess different plaque components like necrotic cores and hemorrhages for future clinical applications. The endmember determination method is based on the iterative minimization of a target function. It compares multiple sets of endmembers and abundance over different regions of the image and estimates the better hyperspectral imagery representation. To evaluate the performance of this method, we analyze vessel mimicking phantoms filled with plaque constituents like lipids, blood, and plaque tissue obtained from endarterectomy surgery. Images were acquired for wavelengths from 500 nm to 1300 nm in steps of 5 nm, ensuring that the main features of all chromophores are taken into consideration. It simplifies the assessment of fluence artifacts. Additionally, multi-perspective images are acquired to evaluate the improvement of spatial compounding on MSPAI and the dependency of the method on the acquisition geometry used. Results show that blind unmixing can differentiate constituent regions within a sample and provide the endmembers for further analysis. Some plaque samples can be represented by an independent endmember or as mixture of blood and lipids, illustrating the potential of the technique to assess the presence of lipids and hemorrhages in plaques. Fluence artifacts remain a relevant issue for the qualitative analysis of highly absorbers, with a significant dependence on the acquisition geometry. Spatial compounding increases SNR and CNR improving image quality and structure identification. To conclude, MSPAI can identify plaque composition supporting the diagnosis and monitoring of carotid artery diseases. Next steps include a verification in plaques and the comparison with histology.
15 mins
Physical resilience indicators in patients undergoing major thoracic aortic surgery
Marjolein Klop, Marit Sanders, Richard van Wezel, Jurgen Claassen
Abstract: Background: Open cardiothoracic surgery, a highly complex procedure, can reveal large differences in clinical outcomes between individuals, especially among older patients. Most patients benefit from surgery, whereas some deteriorate cognitively or physically. Recovery after surgery can be described by the concept of physical resilience, which is a patient’s ability to withstand and recover from a perturbation. Therefore, estimating resilience would be helpful to identify patients at risk of complications during recovery after surgery. Even after a small perturbation, for instance induced by a physiological challenge, a less resilient person is thought to recover slower than a more resilient patient, which can be used to assess physical resilience. Aim: To investigate quality and feasibility of physical resilience measurements in clinical practice at the cardiothoracic surgery outpatient clinic. Methods: All patients who had to undergo open thoracic aortic surgery were included. Resilience measurements were performed during an orthostatic challenge (5 minutes sitting followed by 3 minutes standing) and a handgrip challenge (2x maximal grip strength, 1x sustained maximal grip strength). Meanwhile, continuous finger blood pressure and cerebral oxygenation were measured. We determined feasibility of measuring in clinical practice and quality of the acquired signals. Moreover, a preliminary analysis was performed to compare patients with a short (<10 days) and a long (>13 days) hospital stay based on orthostatic recovery values and grip strength values. Preliminary results: We have included 27 patients, who were on average 63.0 (13.3) years, and 70% were men. Until now, 12 patients have had surgery. Hospital stay was 13.1 (9.8) days, of which 1.5 (0.7) days at the ICU. The in-hospital mortality was 8%. Our measurements were successfully embedded in the clinical workflow and well-tolerated by patients. Technically, 96% of all BP measurements were useful, as well as 93% of cerebral oxygenation measurements. Based on duration of hospital stay, results suggest that blood pressure recovery upon standing may be of interest to assess resilience in future clinical research. Conclusion: We showed feasibility of application of resilience measurements at the outpatient clinic. This is a promising first step towards more routine use of physical resilience indicators in clinical practice.
15 mins
Automatic 3D+t ultrasound-based geometry and elasticity determination of abdominal aortic aneurysms
Esther Maas, Hans-Martin Schwab, Joerik de Ruijter, Emiel van Disseldorp, Frans van de Vosse, Marc van Sambeek, Richard Lopata
Abstract: The rupture risk of an abdominal aortic aneurysm (AAA), which determines whether elective surgery should be performed, is currently assessed from its maximal diameter. However, this criterion is not sufficient for all patients: some large aneurysms remain stable, while other aneurysms rupture before the threshold diameter for surgery is met. Previous studies identified other parameters related to AAA rupture risk and growth: aortic wall elasticity, AAA volume and peak wall stress [1-3]. These first two parameters could be determined directly from time-resolved 3D ultrasound (3D+t US), while the latter could be determined from 3D+t US combined with finite element modeling. The objective of this study was to determine both the geometry (for calculating the volume and wall stress) and the elasticity of a AAA from a single 3D+t US acquisition in a fully automatic manner. 3D+t US acquisitions were acquired from 34 AAA patients, who underwent a CT scan within one month of the ultrasound acquisition. The AAA was automatically segmented on all volumes of the 3D+t data using an adaptation of a Star-Kalman method [4], and the phase in the heart cycle was determined from this by calculating the diameter change over time. From these segmentations, firstly, the US geometry at the frame around mean arterial pressure was compared to the corresponding CT geometry. The similarity index of the geometries was 0.89 [0.86-0.92] (median [IQR]) and the Hausdorff distance 5 [4.4-5.6] mm, showing good correspondence. Secondly, using the segmentation and patient-specific blood pressure, the pressure-strain elastic modulus Ep and relative volume change ΔV/V0 were determined around a region of maximal AAA diameter. The Ep of 3.83 [2.50-5.17] ·105 Pa (median [IQR]) and ΔV/V0 of 0.021 [0.0076-0.029] are in range of previous research on 2D ultrasound and MR, respectively [1,5]. Hence, it is now possible to automatically and accurately determine both the 3D geometry and the elasticity of the AAA wall, using a single ultrasound acquisition. As this method is non-invasive and does not require user input, it allows for longitudinal studies, enabling concurrent investigation of influence of aortic wall elasticity, AAA volume, and peak wall stress on rupture risk.
15 mins
Analysis of carotid haemodynamics in an aging virtual population
Irene Suriani, Massimo Mischi, Kevin D. Lau
Abstract: The analysis of arterial haemodynamic waveforms enables the derivation of clinically relevant parameters (e.g., cardiac output, fluid responsiveness, arterial stiffness) that can be used to support clinical decision making (e.g., administration of drugs/fluids, hypertensive treatment). In particular, the use of ultrasound-derived carotid waveforms as a tool for haemodynamic monitoring shows potential in both peri-operative and critical care situations, due to its non-invasiveness, ease of placement and unobtrusiveness [1], [2]. However, in order to derive clinical insight from the obtained carotid waveforms, it is essential to understand the natural variation that can be expected from physiological differences, e.g. subject’s age, gender, arterial properties. The use of one-dimensional (1D) modelling of arterial haemodynamics is a computationally inexpensive approach that provides a tool to understand how the morphology of arterial waveforms relates to the arterial properties and haemodynamic state of the circulatory system [3]. 1D models are parameterised using in-vivo measurements of arterial diameter, length and stiffness, and have been shown to reproduce realistic pressure and flow waveforms along the arterial network [4]. Moreover, such model can be used to create virtual populations by including population-based variations in the prescribed parameters [5]. In this work we extend the approach adopted by Willemet et al. [5] and focus on the study of carotid haemodynamics. We generated a virtual population of healthy subjects parametrised as a function of age, by varying both the arterial structural parameters (elastic and muscular arteries’ stiffness and diameter, peripheral resistance), as well as hemodynamic parameters (heart rate, stroke volume). We improved the definition of carotid properties by introducing recently published age-related reference values of carotid stiffness (up to two-times higher when compared to the original virtual population [5]) and also diameter [6], [7]. Using the generated population we observed and quantified the changes induced in the pressure, diameter and flow waveforms of the carotid due to the variation of each parameter. The analysed results indicate the physiological variations in carotid waveforms that can be expected from an aging population, as well as what information can be inferred from these.   [1] M. Gassner, K. Killu, Z. Bauman, V. Coba, K. Rosso, and D. Blyden, “Feasibility of common carotid artery point of care ultrasound in cardiac output measurements compared to invasive methods,” J. Ultrasound, vol. 18, no. 2, pp. 127–133, 2015, doi: 10.1007/s40477-014-0139-9. [2] L. Beier, J. Davis, D. Esener, C. Grant, and J. M. Fields, “Carotid Ultrasound to Predict Fluid Responsiveness,” J. Ultrasound Med., vol. 39, no. 10, pp. 1965–1976, 2020, doi: 10.1002/jum.15301. [3] J. Alastruey, K. H. Parker, and S. J. Sherwin, “Arterial pulse wave haemodynamics,” 2012. [4] P. Reymond, Y. Bohraus, F. Perren, F. Lazeyras, and N. Stergiopulos, “Validation of a patient-specific one-dimensional model of the systemic arterial tree,” pp. 1173–1182, 2011, doi: 10.1152/ajpheart.00821.2010. [5] M. Willemet, P. Chowienczyk, and J. Alastruey, “A database of virtual healthy subjects to assess the accuracy of foot-to-foot pulse wave velocities for estimation of aortic stiffness,” pp. 663–675, 2019, doi: 10.1152/ajpheart.00175.2015. [6] T. Uejima et al., “Age-speci fi c reference values for carotid arterial stiffness estimated by ultrasonic wall tracking,” pp. 214–222, 2020, doi: 10.1038/s41371-019-0228-5. [7] K. Hirata, T. Yaginuma, M. F. O’Rourke, and M. Kawakami, “Age-related changes in carotid artery flow and pressure pulses: Possible implications for cerebral microvascular disease,” Stroke, vol. 37, no. 10, pp. 2552–2556, 2006, doi: 10.1161/01.STR.0000242289.20381.f4.

end %-->