Bioluminescence tomography (BLT) is a novel optical molecular imaging technique that advanced the conventional planar bioluminescence imaging (BLI) into a quantifiable three-dimensional (3D) approach in preclinical living animal studies in oncology. In order to solve the inverse problem and reconstruct tumor lesions inside animal body accurately, the prior structural information is commonly obtained from X-ray computed tomography (CT). This strategy requires a complicated hybrid imaging system, extensive post imaging analysis and involvement of ionizing radiation. Moreover, the overall robustness highly depends on the fusion accuracy between the optical and structural information. Here, we present a pure optical bioluminescence tomographic (POBT) system and a novel BLT workflow based on multi-view projection acquisition and 3D surface reconstruction. This method can reconstruct the 3D surface of an imaging subject based on a sparse set of planar white-light and bioluminescent images, so that the prior structural information can be offered for 3D tumor lesion reconstruction without the involvement of CT. The performance of this novel technique was evaluated through the comparison with a conventional dual-modality tomographic (DMT) system and a commercialized optical imaging system (IVIS Spectrum) using three breast cancer xenografts. The results revealed that the new technique offered comparable in vivo tomographic accuracy with the DMT system (P > 0:05) in much shorter data analysis time. It also offered significantly better accuracy comparing with the IVIS system (P < 0:04) without sacrificing too much time.

We present a novel perspective on characterizing the spectral correspondence between nodes of the weighted graph with application to image registration. It is based on matrix perturbation analysis on the spectral graph. The contribution may be divided into three parts. Firstly, the perturbation matrix is obtained by perturbing the matrix of graph model. Secondly, an orthogonal matrix is obtained based on an optimal parameter, which can better capture correspondence features. Thirdly, the optimal matching matrix is proposed by adjusting signs of orthogonal matrix for image registration. Experiments on both synthetic images and real-world images demonstrate the effectiveness and accuracy of the proposed method.