Unsupervised Multi-spectral Photoacoustic Framework for the Detection and Quantification of Tissue Chromophores

Abstract

Spectral Photoacoustic (sPAI) is an innovative imaging technology with great potential to detect and quantify the tissue chromophores. Since tissue chromophores have distinct spectral absorption signatures, sPAI is intrinsically sensitive to molecular components distribution in tissues. Linear unmixing is commonly used to differentiate the underlying components from sPAI. Although this fitting-based approach yields acceptable results, it requires user interaction to provide the source spectral curves as an input. For translational research with patients, this approach can be challenging, as the spectral signatures could differ concerning to the disease conditions. Imaging exogenous contrast can also be challenging as some of the agents are susceptible to spectral changes after the interaction with living tissues. Besides, light fluence attenuation along the imaging depth might induce spectral coloring that compromise the accurate quantification. Here we propose a novel unsupervised sPAI framework that enables fully automatic spectral unmixing and accurate quantification of molecular components. The algorithm has shown improved sensitivity and specificity in tissue-mimicking phantom and in vivo experiments.