Fluorescence Molecular Tomography With Optimal Radon Transform Based Surface Reconstruction

Xin Liu,Daifa Wang,Jing Bai

DOI: https://doi.org/10.1109/IEMBS.2009.5334178

2009-01-01

Abstract:Full angle non-contact fluorescence molecular tomography allows acquiring large data sets from complete angles, and simplifies the experimental setups. Accurately extracting animal surface is important for this kind of imaging systems. However, in in-vivo experiments, mouse breath movements and mechanical errors will influence the surface reconstruction. An optimal radon transform based surface reconstruction method is proposed to handle these two factors. The proposed method uses a line searching method to minimize the mismatch between the reconstructed 3D surface and the projected silhouettes at different angles. Therefore, the proposed method generates the optimal 3D surface compared to other methods based on radon transform. Results show that the mean mismatch of 3D surface generated is less than two CCD pixels (0.154 mm) in in-vivo experiments. In-vivo fluorescence molecular tomography is also performed to demonstrate the efficiency of the proposed method.