Improved Reconstruction for Bioluminescence Tomography by Using the Simplified Spherical Harmonics Approximation Model and Homotopy Method

Bioluminescence tomography (BLT) is a promising molecular imaging tool that may play an important role in preclinical research. However, the accurate and stable reconstruction of interior targets remains a challenge for wide application of this imaging modality. We present a novel reconstruction scheme to improve the BLT result by using the simplified spherical harmonics approximation model and homotopy method. In order to accurately model light propagation in biological tissue, the best approximation order is determined by comparing results of the SPN model (N = 1, 3, 5) with the Monte Carlo (MC) simulation data. Based on this forward model, a two-stage reconstruction using l(1) Homotopy algorithm is then performed. Experimental results on a digital mouse model demonstrate that proposed reconstruction method yield improved results in terms of location accuracy, reconstructed density and time cost.