Experimental three-dimensional bioluminescence tomography reconstruction using the l _p regularization

The three-dimensional reconstruction problem of bioluminescence tomography (BLT) is an ill-posed inverse problem. In the finite element framework, the reconstruction problem can be considered as an optimization problem. To address this problem, we present a novel l _p regularization method based on the compressed sensing in this letter, which does not require a permissible source area or multispectral data. The simulation results of a 3D mouse atlas and in vivo mouse experiment show that the proposed method obtained more sparse and accurate solutions than the l ₁ and l ₂ regularization methods in terms of both localization and quantification.