Preza C
Institute for Biomedical Computing, Washington University, St. Louis, Missouri 63110, USA.
J Opt Soc Am A Opt Image Sci Vis. 2000 Mar;17(3):415-24. doi: 10.1364/josaa.17.000415.
An iterative phase-estimation method for the calculation of a specimen's phase function or optical-path-length (OPL) distribution from differential-interference-contrast (DIC) microscopy images is presented. The method minimizes the least-squares discrepancy measure by use of the conjugate-gradient technique to estimate the phase function from multiple DIC images acquired at different specimen rotations. The estimate is regularized with a quadratic smoothness penalty. Results from testing the method with simulations and measured DIC images show improvement in the estimated phase when at least two rotationally diverse DIC images instead of a single DIC image are used for the estimation. The OPL of a cell that is estimated from two DIC images was found to be much more reliable than the OPL computed from single DIC images (which had a coefficient of variation equal to 15.8%).
提出了一种迭代相位估计方法,用于从微分干涉对比(DIC)显微镜图像计算样本的相位函数或光程长度(OPL)分布。该方法通过使用共轭梯度技术最小化最小二乘差异度量,以从在不同样本旋转角度获取的多个DIC图像中估计相位函数。估计值通过二次平滑惩罚进行正则化。用模拟和实测DIC图像对该方法进行测试的结果表明,当使用至少两个旋转角度不同的DIC图像而非单个DIC图像进行估计时,估计相位有所改善。从两个DIC图像估计的细胞OPL比从单个DIC图像计算的OPL可靠得多(单个DIC图像计算的OPL变异系数为15.8%)。