Mesh optimization for fluorescence molecular tomography

Fluorescence Molecular Tomography is an optical imaging technique which aims at reconstructing the 3D distribution of fluorescent markers in bio-tissues based on surface measurements of emitted photons and a model of light propagation. The gold standard of accuracy in creating this light propagation model is the Monte Carlo method (MC), which simulates the path of photon packets through a discretized model of the tissue. One drawback of MC is the computational burden associated with its stochastic nature. Mesh based MC are computational implementations of MC techniques with favorable computational costs. Herein, we investigate the effects of locally refining a mesh discretization on reconstruction accuracy in mesh based Fluorescence Molecular Tomography. Using a mouse model created from mu CT data and average murine optical properties, we are investigating the performances of mesh refinement strategies in reconstructing an 48.9 mm(3) fluorescence inclusion in the center of the model. Iterative mesh optimization is applied to the inverse problem in which after each reconstruction, the mesh is refined in the area of interest. Performance of the method is evaluated in terms of in volume and center of mass position of the inclusion compared to the ground truth. Our preliminary results indicate that accuracy improves with each refinement until convergence. Moreover, a method for rescaling analytically the forward model to fit each new mesh is also proposed in order to reduce the computational expense of the procedure while maintaining the improvements in accuracy