Abstract
Inverse source reconstruction is the most challenging aspect of bioluminescence tomography (BLT) because of its ill-posedness. Although many efforts have been devoted to this problem, so far, there is no generally accepted method. Due to the ill-posedness property of the BLT inverse problem, the regularization method plays an important role in the inverse reconstruction. In this paper, six reconstruction algorithms based on lp regularization are surveyed. The effects of the permissible source region, measurement noise, optical properties, tissue specificity and source locations on the performance of the reconstruction algorithms are investigated using a series of single source experiments. In order to further inspect the performance of the reconstruction algorithms, we present the double sources and the in vivo mouse experiments to study their resolution ability and potential for a practical heterogeneous mouse experiment. It is hoped to provide useful guidance on algorithm development and application in the related fields.
Highlights
In the past decade, bioluminescence imaging (BLI) has been one of the hot topics in optical imaging and has been successfully applied in tumorigenesis studies, cancer diagnosis, metastasis detection, drug development, gene therapy, etc. [1,2,3,4,5]
We evaluated six reconstruction algorithms, including the Truncated Total Least Square method (TTLS) [27], the Incomplete Variables Truncated Conjugate Gradient method (IVTCG) [25], the Truncated Newton Interior-Point method (TNIPM) [24], the Primal-Dual Interior-Point method (PDIP) [26], and the Weighted Iterative Shrinkage/Thresholding algorithm (WISTA) [28] that were developed by our group, and the Tikhonov regularization that we used for the bioluminescence tomography (BLT) reconstruction in its standard way [31]
In order to investigate the performance of these algorithms in the BLT inverse reconstruction, we carried out five single source experiments, a double source experiment and an in vivo experiment
Summary
Bioluminescence imaging (BLI) has been one of the hot topics in optical imaging and has been successfully applied in tumorigenesis studies, cancer diagnosis, metastasis detection, drug development, gene therapy, etc. [1,2,3,4,5]. Along with the development of the compressive sensing (CS) theory, the lp regularization method has become the mainstream strategy to obtain the optimal solution of the BLT inverse problem [17,22,24,25,26]. The most commonly used form of regularization is the Tikhonov-type regularization This method achieves a compromise between the minimization of the residual norm and the penalty term [32]. Considering the source sparsity characteristics and the insufficiency of the measured data, the l1-norm penalty term has been adopted in the regularization method and was successfully applied to BLT [24,25,26,33].
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
