Spectral unmixing using component analysis in multispectral optoacoustic tomography

Multispectral optoacoustic (photoacoustic) tomography (MSOT) exploits high resolutions given by ultrasound detection technology combined with deeply penetrating laser illumination in the near infrared. Traces of molecules with different spectral absorption profiles, such as blood (oxy-and de-oxygenated) and biomarkers can be recovered using multiple wavelengths excitation and a set of methods described in this work. Three unmixing methods are examined for their performance in decomposing images into components in order to locate fluorescent contrast agents in deep tissue in mice. Following earlier works we find Independent Component Analysis (ICA), which relies on the strong criterion of statistical independence of components, as the most promising approach, being able to clearly identify concentrations that other approaches fail to see. The results are verified by cryosectioning and fluorescence imaging.