Research Of BLT Reconstruction Problem Based On Regularization Method And Level Set Method

As a modality of optical Molecular Imaging, Bioluminescence Tomography (BLT) can reconstruct the location and intensity of interior bioluminescent source of small animal, which may help in in-vivo monitoring and detecting physiological reactions or pathological changes on molecular and cellular level. Because of its high sensitivity and low background noise, BLT has been attracting more and more attention and become a new hotspot of the investigations in medical imaging research field. The research of BLT technology involves the reconstruction algorithm, development of simulation environment and design of imaging system.Reconstruction of BLT aims to solve the distribution of interior bioluminescent source based on the detected optical signals on animal surface. However, the reconstruction is an ill-posed problem in theory. Particularly, the interior bioluminescent source in a BLT system is hard to be modulated, which makes the imaging technique more ill-posed and the reconstruction process more complicated. Nowadays, the study of ill-posed reconstruction problem is still lying in the beginning phase, and there's no unified standard of the solution yet. In this paper, based on the imaging principle of BLT and its corresponding mathematical model, we mainly focused on the investigation of the reconstruction algorithm.In this paper, regularization methods including Tikhonov and truncated singular value decomposition (TSVD) regularization method were proposed to solve the ill-posed problem and realize the reconstruction of bioluminescent source. The regularization methods can reduce the illness of the ill-posed problem and improve the stability of the solution by adding constraints to the solution domain. In addition, L-curve method and generalized cross validation (GCV) method were studied to determine the regularization parameter of BLT reconstruction problem, which ensure that the calculated solutions is approximate to the real solutions. A level set based method was also proposed for the reconstruction of bioluminescent source. In this method, a level set function was introduced for the representation of source distribution. By evolving the level set function, the solve distribution can be updated in an indirect mode, which improved the stability of reconstruction process. Finally, we provided the results of reconstruction based on a series of simulation experiments and physical experiments to verify the reliability and practicability of the two reconstruction approaches.