Nonlinear reconstruction schemes and experimental design for fluorescence diffuse optical tomography

Fluorescence diffuse optical tomography is a rather new optical imaging modality. The object, which is usually a small animal, is illuminated through a set of light sources placed on its surface. The light spreads inside the object and excites fluorophores which themselves emit photons at a longer wavelength. This secondary light is recorded by detectors on the boundary. From the knowledge of the light sources and the boundary measurements, it is sought to reconstruct the distribution of the fluorophore inside the object.



The reconstruction problem is nonlinear and ill-posed which makes it vulnerable to noise. In the first part of this thesis the use of nonlinear inversion schemes together with more advanced regularisation terms as total-variation regularisation and a method of levelset type will be investigated.



The second part deals with the optimal source and detector placement for a given target volume which has not been addressed so far. By minimising the redundancy in the measurements, the best optode locations can be selected and signals originating outside the target volume are suppressed.