CDS&E: Data-enabled closed-loop Koopman control of encapsulated microbubbles

University Of Colorado At Colorado Springs, Colorado Springs CO

Investigators

Abstract

Encapsulated microbubbles show great promise for noninvasive diagnostic and therapeutic medical procedures, such as ultrasound contrast imaging, targeted drug delivery, and gene therapy. The ability to robustly control the oscillations of encapsulated microbubbles can potentially provide enhanced treatment that is highly specific and reduces unwanted side effects. This directly reduces overall medical costs, which is an important societal goal. However, traditional control methods have major limitations when applied to microbubbles due to their complex, nonlinear dynamics. This project will apply state-of-art control techniques based on data science that have several advantages over existing methods, including robustness and versatility. Specifically, this project will develop a novel, data-enabled control framework that uses both observed and simulated data to control the nonspherical oscillations of microbubbles through the applied ultrasound, which can be used for both diagnostic and therapeutic medical applications. The project will also support educational outreach to local secondary schools, where students will perform hands-on experiments using acoustics and data science. Lastly, the project will support the development of a machine learning course focused on engineering applications for undergraduate engineering students at the host institution. The overall research goal of this project is to develop a data-enabled, closed-loop control framework based on Koopman operator theory to acoustically control the nonspherical oscillations of encapsulated microbubbles. The Koopman operator enables the extraction of globally linear observables from a nonlinear dynamical system. The linear observables can be computed using time-series data from the oscillations of the microbubbles. This project will apply a linear quadratic regulator framework to control the Koopman linear observables with two biomedical-related objectives: 1) promote rupture of an encapsulated microbubble through inciting unstable growth of nonspherical shape modes, and 2) drive stable, periodic nonspherical oscillations. This will enable targeted release of medicine from encapsulated microbubbles, improve the uptake of therapeutics across the endothelial layer of blood vessels into the surrounding tissue, and enhance the acoustic echo from ultrasonically driven encapsulated microbubbles. The use of single-frequency, multi-frequency, and broadband ultrasonic forcing will be explored to quantify the degree to which increasing complexity of the acoustic input enhances microbubble control. First, knowledge of the full state of the microbubble will be assumed and used to drive the microbubble to various targets. Second, the acoustic echo reflected from the encapsulated microbubble will be used to estimate the full state through a Koopman Luenberger observer framework. The efficacy of the ultrasound waveform to control the microbubble will be validated using both a computational fluid dynamics model and experiments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.