RBF level-set based fully-nonlinear fluorescence photoacoustic pharmacokinetic tomography

Pharmacokinetic fluorescence optical tomography (PK-FOT) and dynamic contrast enhancement (DCE) based multispectral optoacoustic tomography (DCE-MSOT) are non-ionizing alternatives to nuclear medicine and radiological modalities such as DCE-PET/CT/MRI for spatially-resolved quantitative imaging of PK parameters and fluorophore-concentrations. The present work introduces for the first time in literature, a fluorescence photoacoustic tomography (FPAT) based fully-nonlinear PK-FPAT reconstruction framework; in a 2-compartment PK-model and optical-fluorescence modelled frequency domain photoacoustic equation setting. From boundary pressure measurements, we solve the dynamic FPAT (compartment-concentration) state and (PK) parameter estimation problem with two shape-based RBF level-set reconstruction schemes in regularized trust region settings; a Jacobian-based Gauss–Newton filter and our newly proposed gradient-based gradient filter. The reconstruction algorithms are validated in two dimensional settings with synthetic cancer mimicking phantoms. Our PK-FPAT algorithms lead to more stable and superior reconstructions (observed in reconstructed normalized mean square errors having lesser-variation-between and reduced-values-across data-noise levels, respectively) than those obtained by PK-FOT for similar test cases, while, with respect to current DCE-MSOT schemes, incorporating more complete forward models including optical fluorescence and coupled-ODE compartment models with no simplifying assumptions (within the accuracy of the models considered) on the fluence, and reconstructing actual (rather than scaled) PK-parameters, in a fully-nonlinear framework.