Key Laboratory of Optoelectronic Devices and Systems of Ministry of Education and Guangdong Province, College of Physics and Optoelectronic Engineering, Shenzhen University, Shenzhen 518060, China
Fund Project:Project supported by the National Natural Science Foundation of China (Grant Nos. 61975131, 61775144, 61835009) and the Basic Research Project of Shenzhen, China (Grant Nos. JCYJ20200109105411133, JCYJ20170412105003520, JCYJ20180305125649693)
Received Date:09 November 2020
Accepted Date:22 March 2021
Available Online:07 June 2021
Published Online:05 August 2021
Abstract:Quantitative phase imaging (QPI), which combines phase imaging with optical microscopy technology, provides a marker-free, fast, non-destructive, and high-resolution imaging method for observing transparent biological samples. It is widely used in life science, biomedicine, etc. As an emerging QPI technology, spiral phase contrast microscopy (SPCM) uses a spiral phase filter to achieve edge enhancement of amplitude or phase objects. Using the multi-step phase-shifting technology, a complex sample can be measured quantitatively, which has the advantages of high stability, high sensitivity and high precision. However, the SPCM requires at least three-step phase-shifted spiral phase filtered images to achieve the quantitative reconstruction of the amplitude and phase of a sample, and the image acquisition process and the reconstruction process are relatively complicated, which require high stability of system, and the SPCM has low temporal resolution. In order to further improve the performance of SPCM and increase the system stability, sensitivity and temporal resolution, in this paper a quantitative phase imaging method and system based on a fractional spiral phase plate is proposed. Through a sample intensity image filtered by a fractional spiral phase plate, the modified Gerchberg-Saxton iterative phase retrieval algorithm is used to quantitatively reconstruct the phase of a pure phase sample, which simplifies the experimental process and phase reconstruction steps of spiral phase contrast microsocopy. In the computer simulation experiments, the phase imaging process and the reconstruction process of spiral phase plates based on different topological charges are studied, the feasibility of which is analyzed. Finally, through imaging and phase reconstruction of the phase grating and biological cell sample, it is verified that the phase contrast microscopy method based on the fractional spiral phase plate can effectively improve the contrast of spiral phase contrast microscopy and can obtain a quantitative reconstruciton of a weak phase object. The phase information of a sample has significance in research and application for developing the spiral phase contrast microscopy. Keywords:quantitative phase imaging/ spiral phase contrast microscopy/ spiral phase plate/ phase retrieval
其中$\mathop \varphi \nolimits_{\rm{0}}'$表示迭代恢复出的相位值, $\mathop \varphi \nolimits_{\rm{0}} $ 是输入平面样品的相位值. 综合考虑MSE和SSE, 考虑到计算时间, 选择了迭代次数设置为20次. 样品原图像, 整数阶螺旋相位片滤波与分数阶螺旋相位片滤波后的成像强度图对比及重构出的样品相位图对比结果如图5所示. 图 5 螺旋相位滤波成像及恢复结果对比 (a) 相位型样品原图; (b) 传统整数阶螺旋相位片滤波图像; (c) 对图(b)用SGSA恢复的样品相位图; (d) 分数阶螺旋相位片滤波图像; (e) 对图(d)用SGSA恢复的样品相位图 Figure5. Comparisons of the recorded images and the recovered results: (a) The ground truth phase sample image; (b) the recorded image via traditional integer order spiral phase plate filtering; (c) the recovered phase sample image using SGSA for panel (b); (d) the recorded image via fractional spiral phase plate filtering; (e) the recovered phase sample image using SGSA for panel (d).
根据图6的模拟成像结果, 选择拓扑荷数较小的分数阶螺旋相位滤波器获得的相位滤波图像, 有利于研究人员直接观察滤波图像, 获取样品的状态, 选择样品感兴趣区域进行成像, 然后再利用相位恢复算法做进一步重构处理. 为了说明分数阶螺旋相位滤波器的特性, 首先, 比较了不同分数阶拓扑荷数下的相位滤波图像; 其次, 在相位滤波强度图对比度接近的情况下, 再继续比较分析了拓扑荷数取0.1与拓扑荷小于0.1的恢复相位, 如图7所示, 并以测量图像质量的结构相似度(structural similarity, SSIM)指数来定量对比: 图 7 不同拓扑荷值恢复结果对比 (a) 拓扑荷取0.1时经SGSA恢复出的相位图; (b) 拓扑荷取0.08时经SGSA恢复出的相位图 Figure7. Comparisons of the recovered results for different topologies: (a) The reovered phase image using SGSA when the topology is 0.1; (b) the revoverd phase image using SGSA when the topology is 0.08.
基于SLM的螺旋相衬定量相位成像系统的实验光路如图9所示. 图 9 (a) 基于螺旋相位片滤波的定量相位成像系统光路图; (b) SLM上加载的整数阶叉形光栅; (c) SLM上加载的分数阶叉形光栅 Figure9. (a) Optical setup of quantitative phase imaging system based on a spiral phase filter; (b) the integer order fork grating loaded on SLM; (c) the fractional fork grating loaded on SLM.
首先, 利用定做的相位光栅($ {\rm{Si}}{{\rm{O}}_2} $玻璃, 折射率1.456, 光栅刻线理论设计深度200 nm, 光栅刻线实际测试深度为150 nm, 光栅周期为6 μm)进行拓扑荷数$l = 1$的螺旋相位片滤波的成像实验, 目的是与分数阶螺旋相位滤波成像结果进行对比, 实验结果如图10所示. 图 10 定制相位型光栅的成像 (a) 相位型光栅未滤波明场强度图; (b) 整数阶螺旋相位片滤波成像边缘增强图; (c) 恢复相位图; (d) 恢复深度图, 横纵坐标数值为像素值, 每个像素代表0.325 μm, 总长度为332.8 μm Figure10. Imaging of a custom phase gratinig: (a) The unfiltered bright field image of the phase grating; (b) the recorded integer-order spiral phase filtered edge enhancement image; (c) the recovered phase image; (d) the recovered depth image, the abscissa and ordinate values are pixel values, each pixel represents 0.325 μm, and the total length is 332.8 μm.
从图10可以看出, 整数阶螺旋相位滤波的图像边缘增强效果显著, 再对采集到的强度图利用SGSA进行相位恢复, 可进一步提高图像的对比度. 由SSE随迭代次数变化曲线图11可知, 只需两次迭代即可使得SGSA趋于收敛, 得到样品相位图, 耗时1.412 s (Windows10, 内存16 GB, CPU: Intel (R) Core(TM) i5-9400F CPU @2.90 GHz; Matlab R2018b). 图 11 经整数阶螺旋相位片滤波后再由SGSA重构的相位型光栅SSE随迭代次数的变化 Figure11. SSE error vs. the number of iterations for the grating phase reconstruction problem of phase retrieval from a integer-order spiral phase filtering intensity measurement using the SGSA.
再用MATLAB对恢复相位图进一步分析可知, 得到恢复相位光栅顶部相位值为0.24π, 光栅底部相位值为–0.06π, 平均相位差为0.30π, 结合公式$\varphi=\dfrac{{{\rm{2\pi }}}}{\lambda }{{(}}n{{ - }}1{{)}}h$可得, 恢复相位图的定量深度h平均为216 nm, 与已知光栅刻线深度150 nm的高度值误差为66 nm 左右. 分析其误差产生的原因, 主要来自于用SLM的像素化所产生的螺旋相位片精度不高; 另外, 加工的光栅刻线深度也有一定的误差. 为了进一步提高测量精度, 可以加工相关高精度相位片. 其次, 对拓扑荷数$l = 0.1$分数阶螺旋相位滤波进行实验, 结果如图12所示. 图 12 定制相位型光栅的成像 (a) 相位型光栅未滤波明场强度图; (b) 拓扑荷l取0.1时分数阶螺旋相位片滤波成像强度图; (c) 恢复相位图; (d) 恢复深度图(坐标同图10(d)) Figure12. Imaging of a custom phase grating: (a) The unfiltered phase grating bright field image; (b) the fractional spiral phase plate filtered image when the topological charge l is 0.1; (c) the recovered phase image; (d) the recovered depth image (the coordinates are the same as Fig. 10. (d)).
从图13可以看出, 当使用l = 0.1的分数阶螺旋相位片直接对相位型样品进行螺旋相位滤波成像时, 成像图的对比度相对于未经滤波得到的成像图对比度明显得到提升, 效果与模拟结果得出的结论一致. 因此, 为了方便观察样品, 可以直接利用分数阶螺旋相位片进行螺旋相位滤波成像. 然后对单次拍摄采集到的单幅螺旋相位滤波强度图, 利用SGSA进行相位恢复进一步重构样品的相位信息. 实验结果证明, 分数阶螺旋相位滤波可以得到保留更多低频信息的高对比度强度分布图, 有利于对相位物体进行实时显微观测. 从图13的误差曲线可知, 需迭代10次, 耗时2.839 s, 误差几乎减为0. 进一步分析可知, 恢复相位光栅顶部平均相位值为0.25π, 光栅底部平均相位值为–0.03π, 均值相位差为0.28π, 则恢复相位图的定量深度h为198 nm, 与已知光栅实际刻蚀深度150 nm的高度值误差为48 nm. 与整数阶SPC相比, 测量误差减小27%. 图 13 经分数阶螺旋相位片滤波后再由SGSA重构的相位光栅SSE随迭代次数的变化 Figure13. SSE error vs. the number of iterations for the grating phase reconstruction problem of phase retrieval from a fractional spiral phase filtering intensity measurement using the SGSA.
为了进一步验证分数阶SPC成像效果, 对SH-SY5Y人神经母细胞瘤细胞进行成像, 如图14所示. 图 14 SH-SY5Y细胞成像 (a) SH-SY5Y细胞未滤波明场强度图; (b) 拓扑荷l取0.1时分数阶螺旋相位片滤波神经元细胞成像强度图; (c) 恢复相位图; (d) 定量相移图(坐标同图10(d)) Figure14. SH-SY5Y cell imaging: (a) The unfiltered SH-SY5Y cell bright field image; (b) the intensity image of the neuron cell using the fractional spiral phase filter when the topological charge l is 0.1; (c) the recoverd phase image; (d) the quantitative phase image (the coordinates are the same as Fig. 10. (d)).
从图15误差随迭代次数变化曲线可知, 只需迭代5次即可, 耗时1.782 s. 由于并不清楚细胞样品内部的具体折射率分布, 因此, 只给出细胞的定量相移图. 恢复相位图的最大相位值为${\rm{0}}.{\rm{6141\pi }}$, 最小相位值为${{ - 0}}.{\rm{3234\pi }}$. 从图15可以看出, 当使用分数阶螺旋相位片对生物细胞进行成像时, 成像效果明显, 对比度高, 再经过SGSA恢复后得到的样品的相位信息, 与模拟结果相一致. 图 15 SH-SY5Y细胞经分数阶螺旋相位滤波后再由SGSA重建的SSE随迭代次数的变化 Figure15. SSE error vs. the number of iterations for the SH-SY5Y cells reconstruction problem of phase retrieval from a fractional spiral phase filtering intensity measurement using the SGSA.