A MULTI-SCALE METHOD FOR BIOLUMINESCENCE TOMOGRAPHY USING MULTIPLE TYPES OF A PRIORI INFORMATION

Bioluminescence tomography (BLT) is a novel in vivo technique which may reveal the molecular and cellular activity at the whole-body level in small animal studies. In order to reconstruct bioluminescence sources accurately, it is necessary to incorporate all types of a priori information available and develop novel reconstruction algorithms. In this paper, a multi-scale method is proposed for BLT based on adaptive finite element analysis. Specifically, multiple constraints, including multi-spectral measurements, anatomical and optical knowledge, are imposed to deal with the solution non-uniqueness of BLT. Also, adaptive finite element meshing and permissible source region refinement are utilized to help improve the reconstruction quality and speed. Finally, with the micro-MRI based virtual mouse phantom (MOBY) and the Monte Carlo based synthetic data, the feasibility and effectiveness of this BLT approach is demonstrated