Abstract

Photoacoustic microscopy (PAM), combining optical absorption contrast with ultrasound detection, enables label-free, multi-parametric imaging of microvascular function in vivo, including hemoglobin concentration, oxygen saturation, and blood flow speed. Over the past decades, this technique has emerged as a powerful platform for investigating microvascular structure and function in both preclinical and clinical studies. However, several key limitations still hinder its broader application in biomedical research. First, the contrast of PAM primarily arises from hemoglobin absorption within blood vessels, restricting its ability to probe complementary cellular/molecular processes involved in neurovascular interactions, which are critical for understanding the mechanisms underlying brain function and metabolism. Second, conventional PAM exhibits pronounced anisotropy in spatial resolution, characterized by optical-diffraction-limited microscopic lateral resolution but acoustically defined mesoscopic axial resolution. This substantial disparity significantly constrains the capability of PAM for high-fidelity three-dimensional (3D) imaging of microvascular function. Third, recently developed optical ultrasound detectors, although capable of providing ultrawide bandwidth and high sensitivity, are typically implemented as single-element sensors, resulting in limited field of view and reduced imaging throughput. In this dissertation, I present several technological advances aimed at overcoming these limitations by enriching imaging contrast, improving spatiotemporal resolution, and expanding imaging coverage, thereby advancing the capability of PAM for neurovascular imaging. First, to enrich imaging contrast, I developed an integrated two-photon and photoacoustic microscopy (TPM-PAM), enabling real-time imaging of single-neuron calcium activity alongside oxygen release from individual red-blood-cells (RBCs) in awake mice. In TPM-PAM, a transparent micro-ring resonator (MRR)-based ultrasound sensor breaks the long-standing tradeoff between optical access and acoustic sensitivity, while dual-wavelength kymography simultaneously quantifies single-RBC oxygenation and flow to derive its oxygen release rate. Incorporating nonlinear optical manipulation of the neurovascular unit with cellular precision, TPM-PAM reveals distinct neurometabolic responses to whisker stimulation, single-capillary occlusion, and single-neuron stimulation. This work establishes a powerful platform for dissecting neurometabolic coupling at the cellular scale and understanding oxygen-metabolic regulation in brain health and disease. Secondly, to improve spatiotemporal resolution, I proposed a super-resolution PAM technique based on physics-informed deep neural networks (DNNs), allowing high-speed 3D imaging of oxygen dynamics with isotropic, single-RBC resolution. The network is trained using co-registered 3D PAM and TPM vascular datasets acquired with the integrated TPM-PAM system. Physics-informed constraints, including the isotropy of vascular diameter and the consistency of projected oxygenation information, are incorporated to guide the training process. This approach refines the axial resolution of PAM from tens of micrometers to 4.9 μm, enabling high-fidelity 3D functional imaging of microvasculature in the live mouse brain. Moreover, the trained network is generally applicable across imaging platforms, achieving a twofold axial-resolution improvement on datasets from a piezoelectric transducer-based PAM system with lower acoustic detection bandwidth. Together, this work overcomes a long-standing bottleneck of PAM and bridges a critical gap in oxygen-metabolism imaging. Lastly, to expand imaging coverage, I introduced a polymer MRR array-based ultrasound detector and demonstrated its capability for wide-field functional PAM imaging in vivo. Specifically, a 2 × 2 MRR array with distinct resonance modes within the free spectral range (FSR) is fabricated using soft nanoimprinting lithography, enabling stable operation with a single driving laser source. When integrated into the PAM system, the MRR array provides uniform, high-sensitivity imaging of vasculature and oxygenation across a 3 × 3 mm2 field of view in the live mouse brain, thereby expanding imaging coverage without compromising sensitivity or bandwidth compared with single-element detection.

Committee Chair

Song Hu

Committee Members

Adam Bauer; Chao Zhou; Manu Goyal; Quing Zhu

Degree

Doctor of Philosophy (PhD)

Author's Department

Biomedical Engineering

Author's School

McKelvey School of Engineering

Document Type

Dissertation

Date of Award

4-29-2026

Language

English (en)

Download

Available for download on Tuesday, June 15, 2027

Share

COinS