A Novel Ultrasound Edge-Detection Algorithm for the Assessment of Local Arterial Stiffness

Introduction: Arterial stiffening is highly associated with the development of cardiovascular diseases. Currently, pulse wave velocity is the gold-standard non-invasive assessment of arterial stiffness. However, pulse wave velocity is limited since the current techniques require an assessment across large arterial segments. Purpose: Therefore, the purpose of this study was to develop an ultrasound edge-detection algorithm that can assess local arterial stiffness. Hypothesis: We hypothesized that our novel edge-detection algorithm would accurately quantify artery lumen diameters in-vivo and would accurately characterize the stiffnesses of synthetic silicone aorta models. Methods: To validate the accuracy of the ultrasound edge-detection algorithm in-vivo, 30 ultrasound images that were a combination of the brachial, popliteal, and carotid arteries were used to perform edge-detection with both the novel algorithm and the standard method for edge-detection assessment, Brachial Analyzer (Vascular Research Tools 6). Accuracy was assessed with correlation and Bland-Altman analyses. To validate if the novel edge-detection algorithm could be used to quantify stiffness, two silicone aorta models (shore hardnesses of 10A and 20A) were constructed and used in a flow circuit experiment. Intraluminal pressure was increased linearly for each aorta model and real-time lumen diameters were assessed at 30Hz with the novel edge-detection algorithm. The mean lumen diameters were plotted against mean intraluminal pressures to estimate stiffness. Results: The linear regression equation for lumen diameters assessed with the novel edge-detection algorithm vs. Brachial Analyzer was f(x)=0.956x+0.409 (r 2 =0.989, p<0.01). Bland-Altman analysis of the novel edge-detection algorithm compared to Brachial Analyzer demonstrated a bias of -0.16 mm and a 95% confidence interval (CI) between -0.46 – 0.12 mm. The slope of lumen diameter vs. intraluminal pressure for the shore 20A aorta model was 0.14 mm·mmHg -1 (95% CI: 0.04-0.24 mm·mmHg -1 ) and the slope for the shore 10A aorta model was 0.07 mm·mmHg -1 (95% CI: 0.06-0.08 mm·mmHg -1 ). The expected fold difference between the shore 20A and 10A aorta models was 2, and the edge-detection algorithm estimated a fold difference of 2.00 (95% CI: 1.71 – 3.15). Conclusions: Our novel edge-detection algorithm demonstrated excellent agreement with the current gold standard method for in-vivo analysis. Furthermore, our edge-detection algorithm accurately quantified real-time intraluminal diameters in a flow circuit model and accurately estimated the difference in stiffness between shore 20A and 10A silicone aorta models. Therefore, our novel ultrasound algorithm may be useful for assessing local arterial stiffnesses in-vivo. This research was supported by the NASA Nebraska Space Grant. This abstract was presented at the American Physiology Summit 2025 and is only available in HTML format. There is no downloadable file or PDF version. The Physiology editorial board was not involved in the peer review process.