Non-invasive hemodynamic state monitoring using ultrasound

Abstract

Hemodynamic monitoring provides vital information for diagnosing and treating patients in acute clinical settings. A method is investigated to non-invasively monitor changes in the hemodynamic state. The approach utilizes short-axis ultrasound color flow imaging and processing methods to produce simultaneous waveforms for the arterial area and flow. Beat-to-beat measurements of the mean area, peak volumetric flow rate, and heart rate are extracted, and the distribution of these parameters is used to define the hemodynamic state. Changes in the hemodynamic state are detected by calculating a distance between new measurements and the current hemodynamic state, and then comparing this distance to an adaptive threshold. The distribution was modeled as a multivariate normal distribution characterized by a mean vector and a covariance matrix, and the Mahalanobis distance was used as the distance metric. The threshold level was adapted to produce a constant probability of false positives based on the current distribution. The method was evaluated by observing pharmacologically induced changes in the hemodynamic state during an in vivo animal experiment. The ultrasound-based measurements provided sufficient accuracy to discriminate between the hemodynamic states before, during and after infusion of a vasodilator. The ability to detect an acute change in the hemodynamic state was demonstrated in the transient period at the start of the infusion.