基于B样条参数化描述的形状扩散光学层析成像方法

B-Spline Description for the Shape-Based Image Reconstruction Algorithm of Diffuse Optical Tomography

为了改善扩散光学层析成像(DOT)逆问题的病态性,本文发展了一种基于B样条参数化描述的形状DOT重建算法.在合理地假设组织体各区域光学参数均匀分布的前提下,通过采用B样条参数化描述复杂区域的边界,将DOT的逆问题从大规模的体元光学参数重建转化为只重建各子区域的一组光学参数和复杂形状参数.在信噪比为35,d B条件下对不同形状模拟异质体的重建结果表明,该算法在不同的光学参数范围、异质体初始形状估计、光学参数初始估计下,均可有效地重建出异质体的形状和光学参数.稳态DOT系统的仿体实验验证表明,背景光学参数和异质体光学参数的重构误差分别小于1.18%和12.00%.该算法将对肿瘤诊断漫射光层析成像的发展具有推动作用.

The shape-based image reconstruction algorithm based on the B-spline description is proposed for reducing the ill-poseness of the inverse problem of diffuse optical tomography（DOT）. Under the assumption of the uniform distribution of the optical properties in each region,by describing the boundary of the complex inhomogeneity with the B-spline,instead of reconstructing the optical parameters of every voxel,the proposed algorithm only reconstructs the optical parameters of every region and the shape parameters of the complex region. The simulation results of different inhomogeneity shapes at the signal-to-noise-ratio of 35,d B verify that the shape of the inhomogeneity and the optical parameters of both inhomogeneity and background could be realized effectively within different optical parameter ranges and under different initial guesses of both inhomogeneity shape and optical parameters. The results of the experiments on a solid phantom with a continuous-wave DOT measurement system show that the relative reconstruction errors of the optical parameters of both the background and the inhomogeneity are less than 1.18% and 12.00%,respectively. The algorithm proposed in this paper will promote the development of DOT aiming at tumor diagnosis.