中文摘要：

针对现有的以修正的朗伯-比尔定律（modified Lambert-Beer law，MLBL）为理论模型的光学拓扑方法存在量化度不准确的问题，提出了一种基于最小二乘拟合算法的时域光学拓扑成像方法。该方法利用时域扩散方程在半无限空间条件下的解析解和测量曲线实现最佳匹配以计算相邻的源点与探测点之间的待测组织吸收系数变化值。对该时域光学拓扑成像方法进行了一系列的模拟验证，并以基于时间相关单光子计数（time-correlated single photon count-ing，TCSPC）技术的多通道时间分辨测量系统为平台进行了实验验证。结果表明：所提出的最小二乘拟合算法总体上优于传统的基于经验值光程的MLBL方法。

英文摘要：

To cope with the low quantification in the established optical topography that originates from the exces-sively simplified computation model based on the modified Lambert-Beer’s Law（MLBL）,we proposed a least-squares fitting scheme for time domain optical topography that seeks data matching between the time-resolved meas-urement and model prediction calculated by analytically solving the time-domain diffusion equation in semi-infinite geometry. A series of experiments were also conducted,including simulation and phantom verification.The phan-tom verification was performed with multi-channel time-resolved measurement system based on TCSPC.Our simula-tive and phantom experiments demonstrate that the proposed curve-fitting method is on the whole,superior to the conventional MLBL-based one in quantitative performance.