Clinical Applicability of Carotid Ultrasound Imaging for Patient Monitoring

Hemodynamic monitoring is essential for assessing a patient’s cardiovascular status and supporting clinical decision-making, especially in perioperative or critical care settings. When a patient enters the hospital, the care team can select from a wide variety of options to monitor the patient’s hemodynamic status. Although many invasive and non-invasive hemodynamic monitoring tools are available, they all provide a trade-off between measurement accuracy and invasiveness. The clinical context dictates the degree of invasiveness that the clinical care team is willing to accept, balancing the risk of complications against the accuracy of the measurements. However, in an ideal situation, non-invasive and accurate hemodynamic measurements would be feasible for all patients requiring hemodynamic assessment. Carotid ultrasound imaging has the potential to play a key role as a suitable modality for this purpose. The common carotid artery is located in close proximity to the heart and may therefore demonstrate central hemodynamic changes. Its superficial location allows for easy visualization with ultrasound, a safe, widely available, portable, and cost-effective imaging modality. Furthermore, ultrasound simultaneously captures both anatomical information, such as vessel orientation and vascular wall thickness, and functional information, including blood flow. The recent emergence of wearable carotid ultrasound patches and software for automated image analysis enhances the potential of carotid ultrasound imaging for continuous patient monitoring. However, the clinical applicability of this technique is yet to be determined. This thesis explores the usage of carotid ultrasound imaging for hemodynamic monitoring, beginning with a narrative review of the current utilization of carotid Doppler ultrasound imaging as a hemodynamic monitoring tool. While carotid Doppler ultrasound is not yet interchangeable with invasive, gold-standard cardiac output monitoring techniques, certain carotid ultrasound-derived parameters show promise for functional hemodynamic monitoring. Specifically, the carotid peak velocity variation appears to be a promising parameter for assessing volume status, i.e., detecting hypovolemia, and preload-induced changes in corrected carotid flow time may serve as an indicator of fluid responsiveness. The remainder of the thesis is structured into three parts, reporting on technological and clinical advancements in the clinical use of carotid ultrasound imaging and comparing it to existing methods. The first part of the thesis focuses on optimizing the measurement setup for carotid ultrasound imaging, intending to enable continuous monitoring. The traditionally used long-axis view is challenging to maintain, and measurements may be influenced by operator experience, limiting its suitability for such applications. Preliminary in-silico and in-vitro results suggest that a new, rotated and tilted view offers improved robustness to motion and reduced operator dependence. Another advantage of this new view is that it allows for patient movement, such as rotation of the head. Unlike the long-axis view, where head rotation can move the artery out of the ultrasound field, the rotated and tilted view maintains the carotid artery within the ultrasound field of view, merely displacing its position. This novel imaging approach is evaluated in a clinical setting by comparing carotid diameter estimates from several probe orientations, i.e., long-axis, rotated and tilted, and short-axis views – the latter serving as the reference. Results demonstrated that the rotated and tilted view provided median carotid diameter estimates comparable to and equally robust as those obtained with the reference view. This finding paves the path for future studies on the clinical usage of the rotated and tilted view measurement approach in new applications, such as wearable ultrasound devices. The second part, which describes clinical advancements of carotid ultrasound imaging, details its predictive value for spinal anesthesia-induced hypotension. Hypotension, commonly occurring after the induction of spinal anesthesia, is associated with post-operative end-organ damage leading to complications. Timely prediction and treatment of spinal anesthesia-induced hypotension may reduce the risk of adverse outcomes. To this end, the predictive value of carotid Doppler ultrasound for spinal anesthesia-induced hypotension is investigated. First, established carotid ultrasound-derived variables are studied, specifically the corrected carotid flow time and carotid peak velocity variation. These variables did not predict spinal anesthesia-induced hypotension. Subsequently, the analysis is extended to investigate various carotid Doppler ultrasound and clinical variables, individually and combined, using linear and nonlinear models. Although individual carotid Doppler ultrasound-derived variables had limited performance, the best-performing model for predicting spinal anesthesia-induced hypotension used a linear combination of carotid Doppler ultrasound and clinical variables in a logistic regression model. Key predictive features included baseline mean arterial pressure, fasting duration, American Society of Anesthesiologists class, and patient weight. In the third part of the thesis, carotid ultrasound is evaluated as an alternative to established methods for assessing fluid responsiveness and left ventricular performance. First, carotid and aortic peak velocity variations are compared to stroke volume and pulse pressure variations, which are established indicators of fluid responsiveness. Unlike aortic peak velocity variation, carotid peak velocity variation exhibited a strong agreement with stroke volume and pulse pressure variations. As such, carotid peak velocity variation qualifies as a potential marker for assessing fluid responsiveness. Second, in cardiac surgery patients, corrected flow times from the carotid artery and left ventricle are compared, the latter being a measure of left ventricular performance and a predictor of heart failure. Parameters are computed both manually and automatically using a dedicated algorithm. While the automated carotid and cardiac measurements demonstrated poor clinical agreement, the manually derived values exhibited strong clinical agreement. This suggests that manually derived corrected carotid flow time may serve as a surrogate measure for the corrected left ventricular ejection time. However, since it is impossible to know which analysis provides the true parameter value, future research is required to determine whether manual or automated analysis is preferred in clinical practice. Overall, this thesis describes the potential of carotid ultrasound imaging in the context of hemodynamic patient monitoring. In particular, it achieves three main objectives: 1) offering insights about a measurement setup using a novel probe orientation that allows continuous measurements in the clinical setting, 2) expanding the understanding of carotid ultrasound imaging for hemodynamic decision-support in perioperative care, such as prediction of hypotension during spinal anesthesia, and 3) evaluating carotid ultrasound as an alternative to established methods for assessing fluid responsiveness and left ventricular performance. This thesis demonstrates that carotid ultrasound is a promising technique for hemodynamic monitoring in various clinical contexts, such as during surgery under spinal anesthesia, and to monitor fluid responsiveness. Moreover, the use of a rotated and tilted probe orientation holds the potential to enable continuous hemodynamic measurements.