生物折射率三维无标记定量成像研究进展

Progress of Three-Dimensional,Label-Free Quantitative Imaging of Refractive Index in Biological Samples

折射率是生物样本最重要的光学属性,经常作为内源性“标记物”进行无标记定量成像。虽然通过测量光程差获取相位信息的传统定量相位成像方法已被广泛研究,然而其获取的相位结果是样本折射率与厚度的耦合产物,无法重建三维形态学信息。近年来,以光学投影层析方法为开端,研究人员率先开启了以三维折射率定量成像为目标的形态学特征重建方法研究。然而光学投影层析方法未考虑衍射效应,导致其精度不足。为解决该问题,基于散射反演求解的光学衍射层析技术应运而生,并在无标记生物三维成像方面展现出巨大的潜力。本文锁定生物折射率三维无标记定量成像研究,聚焦光学投影层析和光学衍射层析两种方法的发展历程,从正向测量模型、反演算法以及实现方法三方面进行综述,并对该研究未来的工作进行展望。

Significance Optical microscopic imaging is vital for observing and studying biological cells and tissues. Fluorescent probes and other chemical reagents are commonly used for exogenous labeling of cells in clinical diagnosis and biomedical research because biological samples absorb little visible light, making them colorless and translucent. Although these labeling methods can improve imaging contrast and specificity, phototoxicity destroys samples to some extent and photobleaching affects long-term observation. Furthermore, many subcellular structures, such as lipids, are extremely difficult to label, which limits labeling methods’ applications. Hence, optical imaging methods for label-free samples are considered to be an important research field for simplifying the time-consuming sample preparation process and for reducing biological interference from fluorescent reagents on samples to meet clinical requirements. The interaction of light and samples can become a promising solution using an optical microscopic imaging method to obtain morphological and biological features of label-free samples. Considering light as electromagnetic waves, biological samples can be characterized by their complex refractive index. The quantitative measurement of the refractive index has experienced a long period of development as an endogenous “label” of biological samples, which includes: a) the average refractive index of a cell population suspended in a medium;b) the effective refractive index of a single cell;c) the two-dimensional(2 D) refractive index mapping imaging of a single cell;and d) the three-dimensional(3 D) refractive index distribution imaging of a single cell. Among them, a) and b) are not involved in the imaging process, and only one refractive index value is measured to represent a cell population or a single cell, which can provide very limited information. As a result, researchers are interested in label-free 2 D imaging and 3 D imaging based on a sample’s refractive index. For label-free “thin” biological samples, quantitative phase imaging(QPI) is a powerful quantitative imaging method. Currently, QPI’s ability to analyze the interior information of samples can reach the diffraction limit at lateral resolution. However, because the phase results are the integral of the samples’ refractive index along the light propagation direction, the 3 D refractive index distribution of the samples cannot be obtained.Researchers have set label-free 3 D imaging of biological samples as their next goal to obtain more accurate morphological information, such as nuclear shape, dry mass, and nucleo-cytoplasmic ratio. In recent years, optical projection tomography(OPT) has pioneered the study of morphological feature reconstruction aiming at the 3 D quantitative imaging of refractive index. However, OPT does not consider the diffraction effect, resulting in its inaccuracy. To solve this problem, based on the solution of inverse scattering, optical diffraction tomography(ODT) was developed that has shown great potential in 3 D imaging of label-free biological samples. Although many corresponding advances have been made in the 3 D quantitative imaging of refractive index, there are still a series of challenges in the area of improving the feasibility and performance. To rationally guide the successive development in this field, it is necessary to give a summary of existing research.Progress This study introduces research on the 3D quantitative imaging of refractive index in label-free biological samples. In Section 2, the forward measurement model from OPT to ODT is detailed. The Fourier-slice theorem in OPT(Fig. 1) is derived using the approximation of straight-line propagation of light, which relates the phase results of QPI to the refractive index of samples. After considering the diffraction effect, based on the Helmholtz equation, the Born approximation and Rytov approximation are used to obtain the Fourier diffraction theorem in ODT(Fig. 3), which relates the scattering field to the refractive index of samples. The following inversion algorithms are guided by the two approximations. In Section 3, the inversion algorithms of 3D refractive index quantitative imaging are divided into four categories according to the progression of development. The first is a direct inversion(Fig. 4), which obtains the 3D refractive index distribution of samples from phase or scattering field data at different angles using the Fourier-slice theorem or the Fourier diffraction theorem directly. The filtering methods(filtering backprojection method in OPT and filtering backpropagation method in ODT) then transfer the filling of measurement information from the Fourier domain to the spatial domain to reduce the influence of interpolation artifacts. Subsequently, regularization and machine learning approaches from the optimization area are brought into ODT to overcome the approximation of “weak” scattered and “thin” samples, and obtain the refractive index of multiplescattering samples from sparse data. Section 4 summarizes the implementation methods of 3D refractive index quantitative imaging. The measurement based on the digital holographic is first explained and then the implementation methods of the sample rotation mode and incident wave rotation mode are introduced(Fig. 6). The ODT based on partially coherent light and non-interference is then introduced. Section 5 summarizes the previous sections and puts forward the prospect of future research work.Conclusion and Prospect The use of the refractive index as an endogenous label in 3D quantitative imaging for label-free biological samples is gaining popularity. However, some of the following improvements still need to be made: 1) a more accurate scattering measurement model should be established to improve the imaging resolution at the theoretical level;2) there is still a necessity to overcome the limitation of sparse data, develop a more effective refractive index 3D reconstruction inversion algorithm, and improve the reconstruction accuracy and calculation speed;3) as the refractive index is nonspecific for biological features, in-depth analysis of the refractive index in samples is required. At the application level, this study requires mature equipment to advance clinical procedures. Furthermore, its applications can be expanded to other domains of study, including water environment monitoring, food safety, material characterization, virology, and single-molecule analysis.