提出了一种基于截断完全最小二乘法(TTLS)的生物发光断层成像(BLT)重建算法,并在扩展广义交叉验证(GCV)的基础上,设计了一种用于确定最佳截断水平的混合广义交叉验证方案(HGCV).与现有的只考虑测量噪声的重建算法不同,这种TTLS结合HGCV的重建算法可将模型离散、解剖结构获取以及光学参数测定中的误差与表面测量误差同时处理,仿真及物理仿体实验验证了该算法的有效性和鲁棒性.
Reconstruction algorithm for bioluminescence tomography(BLT) was proposed by truncated total least squares(TTLS) method.Hybrid generalized cross validation(HGCV) scheme was designed to determine the optimal truncation level by extending the GCV criterion in that algorithm,which deals with various errors caused by model discretization,acquisition of anatomical structures,determination of optical parameters and external measurement together,which was different from the present algorithms in BLT that only consider measurement noise.Numerical simulations and physical phantom experiment demonstrate the effectiveness and robustness of the algorithm for BLT problem.