Zhaolu Zuo, Shaobin Dou, and Deyi Kong
Fluorescence molecular, compressed sensing, image reconstruction, sparse view
With the development of fluorescence molecular tomography technology, densely sampled measurements can be easily obtained. However, challenges still remain in fast image reconstruction, phototoxicity, photobleaching of the fluorescent proteins, long time under anesthetic and specimen movement during the acquisition, etc. In this work, a novel compressed sensing reconstruction method was presented to reduce the acquisition time or number of projections, while yielding better image quality. We defined a differentiable convex total variation function as the optimization objective, which improved the calculation speed and stability of the finite difference transform. In addition, maximum likelihood expectation maximization algorithm was proposed as the projection onto convex sets process to improve the convergence speed of the algorithm. For testing and evaluating the proposed algorithm, we also designed a head model of emission mode in MATLAB. Phantom simulations demonstrated that the proposed method could achieve fine reconstruction image by sparse view.
Important Links:
Go Back
