Superpixel Spectral Photoacoustic Imaging of Tissue Chromophores : “En Route to the Clinics”
Photoacoustic imaging (PAI) is an emerging biomedical imaging technology based on the conversion of absorbed nanosecond pulsed laser light into acoustic waves that can be detected just as conventional ultrasound (US) signals. This guarantees the advantages of strong optical contrast as well as high US resolution and imaging depth (up to several cm). Since tissue chromophores have distinct wavelength-dependent spectral absorption signatures, spectral PAI (sPAI) is intrinsically sensitive to visualize and quantify molecular component distribution in tissues. Thus, by illuminating the tissues at different wavelengths, molecular information based on either endogenous chromophores (like oxygenated and deoxygenated hemoglobin, lipid, melanin, collagen, and water) or a variety of exogenous contrast agents or both can potentially be differentiated. In this thesis work, I developed advanced machine learning algorithms that can automatically monitor and quantify the tissue molecular features of vital importance. This aims to identify tissue biomarkers from sPAI without any a priori knowledge and user interactions as it can enhance the diagnosis, treatment, and monitoring of the tissue changes in a large number of diseases. In addition, the development of durable and realistic phantoms would minimize animal experiments in compliance with the 3R principle. This allows to accurately test advanced methods in PAI, which leads to higher chances of success in the in-vivo experiments. Finally, having PA images at multiple scales, resolutions, and depths by using multiple US transducers with different frequency bandwidths, lay the foundation for future advances in clinical translation of the technology. This is fundamental, especially in applications where photoacoustic can contribute by providing novel insights such as to follow-up tissue regeneration in presence of biodegradable implants used to fix bone fractures in human patients.