Maximum likelihood tomographic reconstruction of extremely sparse solutions in diffuse fluorescence flow cytometry

We apply reparameterization and the maximum likelihood method to a specific fluorescence-mediated tomography problem where the solution is known a priori to be extremely sparse (i.e., all image values are zero except for one). Our algorithm performs significantly better than a standard image reconstruction method, particularly for deep-seated targets, and achieves close to 150 μm accuracy in a 3 mm diameter cross-sectional area with only 12 measurements. Moreover, results do not depend on the selection of a regularization parameter or other ad hoc values, and since reconstructions can be computed very quickly, the algorithm is also suitable for real-time implementation.