Tomographic bioluminescence imaging by an iteratively re-weighted minimization

Ping Wu,Kai Liu,Zhenwen Xue,Wei Guo,Chenghu Qin,Jie Tian

DOI: https://doi.org/10.1117/12.911725

2012-01-01

Abstract:Tomographic bioluminescence imaging (TBI), with visible light emission in living organisms, is an effective way of molecular imaging, which allows for the study of ongoing tumor biological processes in vivo and non-invasively. This newly developed technology enables three-dimensional accuracy localization and quantitative analysis of the target tumor cells in small animal via reconstructing the images acquired by the high-resolution imaging system. Due to the difficulty of reconstruction, which is often referred to an ill-posed inverse problem, continuous efforts are still made to find more practical and efficient approaches. In this paper, an iteratively re-weighted minimization (IRM) has been applied to reconstruct the entire source distribution, which is known as sparse signals, inside the target tissue with the limited outgoing photon density on its boundary. By introducing a weight function into the objective function, we convert the lp norm problem into a more simple form of l2 norm to reduce the computational complexity. The weight function is updated in each iterative step to compute the final optimal solution more efficiently. This method is proved to be robust to different parameters, and mouse experiments are conducted to validate the feasibility of IRM approach, which is also reliable at whole-body imaging.