Reconstruction of in vivo fluorophore concentration variation with structural priors and smooth penalty

Reconstruction of fluorophore concentration variation in fluorescence molecular tomography is expected to reveal the metabolic processes of fluorescent biomarkers in vivo. However, the complicated and strong noise within in vivo data inhibits its applications for in vivo cases. A smooth penalty method is presented in this work to suppress the noise. The method is based on a recursive reconstruction scheme which reconstructs the fluorophore concentration variation rates (FCVRs) of two neighboring frames at the same time within an inner iteration. In addition, the performance of the Laplacian-type regularization incorporating structural priors is investigated. Results of simulations suggest that the smooth penalty method almost has no influence on the reconstructed FCVRs when the target curve is smooth, and results of in vivo experiments on mice indicate that the method is capable of suppressing the noise and achieving smooth time courses of fluorescent yield. Results of both the simulations and in vivo experiments demonstrate that the Laplacian-type regularization can improve the image quality.