CN107966801A - A kind of high speed Fourier lamination imaging device and reconstructing method based on ring illumination - Google Patents

Abstract

The invention discloses a kind of high speed Fourier lamination imaging device and reconstructing method based on ring illumination, device employs annular LED plate as lighting source, the illumination numerical aperture of every LED on annulus is all equal with the numerical aperture of object lens at the same time, and passes sequentially through LED unit brightness calibration, LED location calibration and correction, original image collection, original image pretreatment and high-definition picture initialization and realize iterative reconstruction.Numerical aperture inscribe of the light angle of present invention LED each on annulus all with object lens, so that phase transfer function completely covers all low frequency parts of phase frequency spectrum, so as to only need to shoot a small amount of light field low resolution picture, just can phase distribution that is highly stable and reconstructing object large visual field high resolution exactly, be highly suitable for the unmarked high-speed quantitative phase micro-imaging of living cells.

Description

A high-speed Fourier stack imaging device and reconstruction method based on ring illumination

technical field

The invention belongs to optical microscopic imaging technology, in particular to a ring-shaped lighting-based high-speed Fourier stack imaging device and a reconstruction method.

Background technique

Phase recovery is an important technology in optical measurement and imaging. No matter in the field of biomedicine or industrial detection, phase imaging technology is playing an important role. Throughout the development of optical measurement technology for nearly half a century, the most classic phase measurement method is interferometry. However, the disadvantages of interferometry are also very obvious: (1) Interferometry generally requires a highly coherent light source (such as a laser), which requires a more complex interferometric device; (2) The introduction of an additional reference optical path leads to the measurement environment The requirements become very stringent; (3) The coherent speckle noise introduced by the highly coherent light source limits the spatial resolution and measurement accuracy of the imaging system.

Unlike the interferometry method, another very important phase measurement technique does not require interferometry, and they are collectively called phase recovery. Since it is very difficult to directly measure the phase distribution of the light wave field, it is very easy to measure the amplitude (intensity) of the light wave field. Therefore, the process of using the intensity distribution to recover (estimate) the phase can be regarded as a mathematical "inverse problem", that is, the phase recovery problem. The methods of phase recovery can be subdivided into iterative method and direct method. On the one hand, the direct method is represented by the light intensity transfer equation method and the differential phase contrast method, both of which only need to take at least 2-4 images to directly solve the phase distribution of the object according to their respective phase transfer functions. However, since the phase transfer function values of the two are relatively low in the low-frequency region and approach zero at zero frequency, there are often some low-frequency errors in the restored phase distribution, so additional regularization optimization methods are needed to remove these Low-frequency reconstruction errors make it difficult to guarantee reconstruction accuracy.

On the other hand, the iterative method is represented by the scanning imaging method based on the principle of synthetic aperture imaging, which was first proposed by Hoppe for the study of crystal structure (Hoppe W. Trace structure analysis, ptychography, phasetomography [J]. Ultramicroscopy, 1982, 10 (3 ):187-198.), and verified the effectiveness of this method by studying scanning transmission electron diffraction microscopy imaging of crystals and amorphous. Rodenburg and Faulkner combined the phase recovery algorithm to improve this method several times (Rodenburg J M. Ptychography and related diffractive imaging methods [J]. Advances in Imaging and Electron Physics, 2008, 150: 87-184.). At present, this imaging method has been experimentally confirmed in different wavelength bands such as visible light domain, X-ray, and electron microscope, and several technologies have been developed to improve the resolution and phase recovery accuracy. , Quantitative phase recovery, etc. have great potential. In 2013, Zheng Guoan introduced synthetic aperture imaging from the space domain to the frequency domain, and proposed Fourier stack imaging technology, also called frequency domain stack aperture imaging technology (Zheng G, Horstmeyer R, Yang C. Wide-field, high- resolution Fourier ptychographic microscopy [J]. Naturephotonics, 2013, 7(9):739-745.). Although this method can achieve large-field-of-view and high-resolution microscopic imaging, the spatial bandwidth product has been greatly improved compared with traditional optical microscopy techniques. However, the phase recovery accuracy of this method has not been verified, and there are often a large number of low-frequency phase reconstruction errors. In addition, because it needs to collect a large number of low-resolution images to reconstruct a high-resolution image, the imaging speed is slow, and it is difficult to meet the current needs of various high-speed quantitative phase microscopy imaging (such as label-free dynamic quantitative phase imaging of living cells). Therefore, how to realize high-precision large-field-of-view high-speed quantitative phase microscopy imaging under the premise of only taking a small number of low-resolution images has become a technical problem that Fourier stack imaging technology must overcome.

Contents of the invention

The purpose of the present invention is to provide a high-speed Fourier stack imaging device and reconstruction method based on ring illumination, which can realize high-precision large-field high-speed quantitative phase microscopic imaging on the premise of only taking a small number of low-resolution images, which greatly Improving the imaging speed and phase reconstruction accuracy of Fourier stack imaging technology.

The technical solution that realizes the object of the present invention is:

A high-speed Fourier stack imaging device based on ring illumination, including a ring LED board, a stage, a condenser, a sample to be tested, a microscopic objective lens, an imaging tube lens, and a camera, wherein the i-th lighted LED on the ring LED board The light emitted by the unit is converged by the condenser lens into parallel light and irradiated on the sample to be tested. The sample to be tested is placed on the stage, and part of the diffracted light passing through the sample to be tested is collected by the microscope objective lens and passed through the imaging tube lens. Convergence illuminates the imaging plane of the camera, and the resulting light intensity map is recorded by the camera.

A high-speed Fourier stack imaging reconstruction method based on ring illumination, the steps are as follows:

Step 1, LED unit brightness calibration: The ring-shaped LED board is used as the illumination source of the microscope, and each LED unit is lit in turn, and after irradiating a blank sample to be tested, use a high-power objective lens to collect corresponding images and calculate the red and green brightness of each LED unit. The normalized brightness correction coefficients corresponding to the three blue bands;

Step 2, LED position calibration and correction: the resolution board is used as the sample to be tested, and the ring-shaped LED board is used as the illumination source of the microscope. Each of the LED units is lit in sequence, and after irradiating the sample to be tested, the under-focus and defocus h distances are collected respectively. Download the corresponding image, and then calculate the illumination angle corresponding to each LED unit in the three bands of red, green and blue through the sub-pixel registration algorithm, and then determine the position of the ring-shaped LED board through nonlinear regression; if there is an LED unit on the ring If the inclined plane wave is not inscribed with the numerical aperture of the objective lens, then the position of the annular LED board is changed by adjusting the fine-tuning translation stage, and the second step is repeated until the inclined plane waves generated by the LED units on all rings are inscribed with the numerical aperture of the objective lens;

Step 3, original image acquisition: the ring-shaped LED board is used as the illumination source of the microscope, and each LED unit is sequentially lit, and the corresponding original low-resolution bright field image is collected after irradiating the sample to be tested;

Step 4, original image preprocessing, including threshold denoising and brightness correction. Firstly, threshold denoising is performed on the original low-resolution images taken according to the average value of dark current noise of the camera, and then each image is divided by The normalized brightness correction coefficient obtained in step 1;

Step 5, high-resolution image initialization: average all captured low-resolution bright-field images, and then initialize the amplitude and phase of the high-resolution images by upsampling;

Step 6, iterative reconstruction: use Fourier stack imaging technology based on pixel merging to perform synthetic aperture calculation on each image collected in the frequency domain one by one, and gradually reduce the update coefficient, and use the cost function value as the criterion. When it is less than a given threshold, the iteration is stopped, and the amplitude and phase of the high-resolution image at this time are the finally obtained large-field high-resolution microscopic image.

Compared with the prior art, the present invention has significant advantages: (1) the device has both adopted the illumination angle of each LED on the ring in the ring-shaped LED board and is inscribed with the numerical aperture of the objective lens, so that the phase transfer function completely covers the phase All low-frequency parts of the spectrum ensure phase reconstruction accuracy and eliminate low-frequency reconstruction errors. (2) Since the illumination angles of the LEDs on the ring are inscribed with the numerical aperture of the objective lens, the reconstruction resolution can reach twice the numerical aperture of the objective lens, and all images are taken in bright field, with high reconstruction stability and reduced The requirements for high dynamic range of the camera are met. (3) It only needs to take up to 13 low-resolution bright-field pictures within 0.5 seconds, and it can very stably and accurately reconstruct the phase distribution of the object with a large field of view and high resolution, which greatly improves Fourier stack imaging The imaging speed of the technology reaches 2 frames per second, which is very suitable for label-free high-speed quantitative phase microscopy imaging of living cells.

The present invention will be described in further detail below in conjunction with the accompanying drawings.

Description of drawings

Figure 1 is a schematic diagram of the optical path of a high-speed Fourier stack imaging device based on ring illumination: (a) an imaging device that includes a condenser, and (b) an imaging device that does not include a condenser.

Figure 2 is the normalized brightness of each LED unit in the ring LED board: Figure 2(a), Figure 2(b), and Figure 2(c) are the normalized luminance corresponding to the three bands of red, green, and blue respectively. Normalized brightness.

Fig. 3 is a schematic flowchart of the iterative reconstruction method of the present invention.

Figure 4 is the reconstruction result of super-resolution phase recovery using different numbers of original low-resolution images under different maximum illumination tilt angles of simulation: Figure 4(a) is the ideal phase distribution of the simulated object under test; Figure 4(b1 ), Figure 4(b2), and Figure 4(b3) are respectively when the maximum illumination inclination angle of the bright field LED unit is inscribed with the numerical aperture of the objective lens, and are restored from 81, 13, and 5 bright field low-resolution images Figure 4(c1), Figure 4(c2), and Figure 4(c3) are respectively when the maximum illumination tilt angle of the bright field LED unit and the numerical aperture of the objective lens cannot be inscribed, from 81, 13, and Phase results of five brightfield low-resolution images recovered.

Fig. 5 is the reconstruction result of using the present invention to carry out super-resolution phase recovery to the microlens array sample: Fig. 5 (a) is that the LED unit in the center of the ring-shaped LED plate illuminates the microlens array under the 10 times objective lens (numerical aperture 0.4) The low-resolution image taken by the sample; select the image area in the box in Figure 5(a), and the bright field images of this area under 9 different illumination angles are shown in Figure 5(b); Figure 5(c) and Fig. 5(d) are the high-resolution light intensity distribution and phase distribution recovered after iterative reconstruction in this region, respectively.

Fig. 6 is the reconstructed result of dynamic quantitative phase imaging of Hela cells using the present invention: Fig. 6 (a) is a full-field high-resolution phase distribution of unlabeled dynamic quantitative phase of Hela cells; select Fig. 6 (a) The image area in the dotted frame is enlarged as shown in Figure 6(b); Figure 6(c1)-Figure 6(c7) are the selected areas at different times (0s, 250s, 500s, 750s, 1000s, 1250s, 1500s,) Dynamic quantitative phase imaging results of real-time division of Hela cells.

Detailed ways

The actual hardware platform on which the ring-illumination-based high-speed Fourier stack imaging device and the reconstruction method of the present invention depend is a ring-illumination-based microscope.

As shown in Figure 1, the high-speed Fourier stack imaging device based on ring illumination of the present invention includes a ring LED plate 1, an object stage 2, a condenser lens 3, a sample to be tested 4, a microscopic objective lens 5, an imaging tube lens 6, and a camera 7, The key difference from the existing Fourier stack imaging system is that this system only retains the LED unit and the central LED unit whose illumination tilt angle is inscribed with the numerical aperture of the objective lens, which greatly reduces the number of bright field images required. Not only to reduce the number of raw image acquisitions, but also to ensure the reconstruction accuracy of phase recovery. The annular LED plate 1 is placed on the front focal plane position of the condenser lens 3, and the central LED unit of the annular LED plate must be positioned on the optical axis; The plane is placed on the rear focal plane of the imaging tube lens 6, and the sample 4 to be measured on the stage 2 is adjusted to the front focal plane of the microscope objective lens 5 during imaging, forming an infinity-corrected imaging system. The light emitted by the i-th lighted LED unit 8 on the annular LED board 1 is converged by the condenser lens 3 to become parallel light and irradiates on the sample 4 to be tested. The sample 4 to be tested is placed on the stage 2 and transmitted through the A part of the diffracted light from the test sample 4 is collected by the microscope objective lens 5, and converged by the imaging tube lens 6 to illuminate the imaging plane of the camera 7, and the formed light intensity map is recorded by the camera 7.

There is also a high-speed Fourier stack imaging device based on ring illumination, which includes a ring LED board 1, an object stage 2, a sample to be tested 4, a microscope objective lens 5, an imaging tube lens 6, and a camera 7, wherein the ring LED board 1 has the first The light emitted by i lighted LED units 8 is directly irradiated on the sample 4 to be measured, which can be approximated as parallel light irradiation for the local field of view. The sample 4 to be measured is placed on the stage 2, and through Part of the diffracted light from the sample 4 is collected by the microscope objective lens 5 , and converges through the imaging tube lens 6 to illuminate the imaging plane of the camera 7 , and the formed light intensity map is recorded by the camera 7 .

As shown in Figure 1, the annular LED board 1 is placed on the front focal plane of the condenser 3, where f is the focal length of the condenser 3, generally between 10-50mm. The annular LED board 1 includes several (up to 13) LED units 8, the central LED unit of the annular LED board must be located on the optical axis, and the rest of the LED units are evenly arranged in a circle with the central LED unit as the center and a radius of d superior. Each LED unit is a red, green and blue three-color LED unit, and its typical wavelengths are red light 633nm, green light 525nm and blue light 465nm. The distance d from each LED unit to the center is typically 10-20mm. Ring LED board 1 can choose to process customized or other commercial ring LED boards, or use a commercial square LED array to light up only the LED units inscribed with the numerical aperture of the objective lens and the center LED unit located on the optical axis to achieve ring lighting .

In the present invention, in order to meet the illumination inclination angles of all LED units on the ring, they must be inscribed with the numerical aperture of the objective lens, the numerical aperture of the microscopic objective lens 5 is NA _obj , the distance from the annular LED board 1 to the sample is f, and the annular LED The radius of the LED ring in the board 1 is d, and the light emitted by the LED unit 8 is converged by the condenser lens 3 to become parallel light, and the included angle with respect to the optical axis is _θi , which must satisfy At the same time, in order to meet the minimum spatial sampling rate required by the reconstruction algorithm, the illumination wavelength of the annular LED panel 1 is λ, the magnification of the microscope objective lens 5 is Mag, and the pixel size is Δx _cam , and must satisfy

Camera 7 of the present invention can be selected to use color or monochromatic camera, if it is monochromatic camera, only need to allow each LED unit 8 in the annular LED board 1 to send red or green or blue monochromatic light respectively when imaging, with monochromatic The camera 7 can record all the monochrome images sequentially; if it is a color camera, each LED unit 8 in the annular LED board 1 can emit red, green and blue light at the same time when imaging, and record all the color images with the color camera 7.

The steps of the high-speed Fourier stack imaging reconstruction method based on ring illumination in the present invention are as follows:

Step 1, LED unit brightness calibration. The ring-shaped LED board is used as the illumination source of the microscope, and each of the LED units is sequentially lit. After irradiating the blank sample to be tested, a high-power objective lens is used to collect the corresponding image and calculate the normalization of each LED unit in the three bands of red, green and blue. luminance correction factor. The implementation process is as follows: the ring-shaped LED board 1 is used as the illumination source of the microscope, and each LED unit in it is sequentially lit, and after irradiating a blank sample to be tested, a high-power objective lens (a typical value of a microscope objective lens with a numerical aperture of 40 times 0.95) is used to collect the corresponding Image. Since the entire ring-shaped LED board contains N LED units (N≤13), each LED unit emits monochromatic light in three colors of red, green and blue, and a total of 3N low-resolution images are taken, then the i-th LED unit is in The image captured by the blank sample to be tested under the color c is recorded as Wherein i=1, 2,..., N, c=r, g, b, r is the two-dimensional coordinate r=(x, y) of the real space. Then calculate the average light intensity for each image Become the average brightness of each LED unit in the three bands of red, green and blue, where N _pixel is an image The total number of pixels in . The average brightness corresponding to the LED unit in the center of the annular LED board is Then the normalized brightness correction coefficient corresponding to each LED unit in the three bands of red, green and blue for

Step 2, position calibration of the ring LED board. The resolution board is used as the sample to be tested, and the ring-shaped LED board is used as the illumination source of the microscope. Each of the LED units is lit up in sequence. The pixel registration algorithm calculates the illumination angle corresponding to each LED unit in the three bands of red, green and blue, and then determines the position of the annular LED board through nonlinear regression.

The specific implementation process is as follows: the resolution board is used as the sample to be tested, and the ring-shaped LED board is used as the illumination source of the microscope, each of the LED units is sequentially lit, and after irradiating the sample to be tested, the under-focus and defocus h distances (h typical Values from 10 to 30 microns) correspond to in-focus images and defocused image According to the angular spectrum diffraction theory, the focused image corresponding to the central LED unit Propagate the h distance numerically along the optical axis to obtain a numerical defocused image Each out-of-focus image is then computed by a sub-pixel registration algorithm Relative to numerically defocused images offset (PY _x , PY _y ), then the spatial frequency vector of the illumination light corresponding to the i-th LED unit is

Wherein, (u _x , u _y ) is the spatial frequency along the xy direction, and λ is the wavelength of the illuminating light.

Finally, determine the position of the annular LED board through nonlinear regression, the formula is:

x _i ＝d[cos(θ)m _i +sin(θ)n _i ]+Δx

y _i ＝d[-sin(θ)m _i +cos(θ)n _i ]+Δy

Among them, Q(...) is the objective function of the nonlinear regression method, (θ,Δx,Δy,f) are the four position parameters of the updated circular LED board, which are the rotation error, the translation error in the x direction, The translation error and focus error in the y direction, (θ ⁰ ,Δx ⁰ ,Δy ⁰ ,f ⁰ ) are the initial position parameters of the ring LED board, Indicates the nonlinear regression operation, d is the distance between two adjacent LED units of the ring-shaped LED board, ( _xi , y _i ) indicates the spatial position coordinates of the i-th LED unit, λ is the wavelength of the illumination light, (m _i ,n _i ) is the row number and column number corresponding to the i-th LED unit.

It is also required in step two: if the four position parameters of the updated circular LED board are respectively θ=0, Δx=0, Δy=0, There is no need to perform position correction, otherwise it is necessary to change the position of the ring-shaped LED board by adjusting the fine-tuning translation stage, and repeat step 2 until the inclined plane waves generated by all LED units on the ring are inscribed by the numerical aperture of the objective lens, that is, to achieve If the fine-tuning translation stage is not adjusted to correct the position of the ring-shaped LED board, it will lead to a serious decline in the accuracy of phase recovery.

Step three, original image acquisition. The ring-shaped LED board is used as the illumination source of the microscope, and each of the LED units is illuminated in sequence, and the corresponding low-resolution original image is collected after irradiating the sample to be tested.

Step 4, original image preprocessing. Raw image preprocessing includes threshold denoising and brightness correction. Firstly, threshold denoising is performed on the original low-resolution images taken according to the average value of dark current noise of the camera as the threshold value, and then each image is divided by the normalized brightness correction coefficient obtained in step 1. The implementation process is as follows: turn on each of the LED units in sequence and use red, green and blue monochromatic light to illuminate the original low-resolution image captured by the sample 4, which is denoted as I _i,c , and turn off all the LED units to capture a dark image. Current noise image I _dark . Then, according to the average value of the dark current noise of the camera as the threshold, the threshold denoising is performed on the original low-resolution image taken, and the formula is

in Indicates the image after threshold denoising, and mean(...) indicates the average gray value of the image. After that, divide each image by the normalized brightness correction coefficient obtained in step 1 to complete the image brightness correction. The formula is:

in, is the brightness-corrected image, is the image after threshold denoising, is the normalized brightness correction coefficient obtained in step 1.

Step five, high-resolution image initialization. All intensity-corrected low-resolution brightfield images were averaged and then upsampled to initialize the amplitude and phase of the high-resolution image. The formula for high-resolution image initialization is:

in, is the initialized high-resolution complex amplitude image, UP[...] means to perform upsampling nearest neighbor interpolation, and N _b is the number of bright field images.

Step six, iterative reconstruction. The Fourier stack imaging technology based on the iterative algorithm of pixel merging is used to perform synthetic aperture operation on each collected image one by one in the frequency domain, and gradually reduce the update coefficient. When the cost function value is less than a given threshold, the iteration is stopped. The amplitude and phase of the high-resolution image at this time are the finally obtained large-field high-resolution microscopic image. The formula based on the pixel binning iterative algorithm is:

Among them, F{...} means to perform Fourier transform, F- ¹ {...} means to perform inverse Fourier transform, UP[...] means to perform upsampling nearest neighbor interpolation, DOWN[...] means to perform downsampling Sample binning. is the high-resolution spectrum of the sample to be tested, k represents the kth iteration, P _i ^k is the spectral aperture function of the microscope objective lens, is the updated local frequency spectrum of the sample to be tested, and γ is a constant used to ensure that the denominator is not zero, with a typical value of 0.001. |...| means to find the modulus of a two-dimensional complex matrix, and |...| _max means to find the maximum value in the modulus of a two-dimensional complex matrix. α ^k is the update coefficient of the spectrum of the sample to be tested in the k-th iteration, β ^k is the update coefficient of the aperture function of the microscopic objective lens in the k-th iteration, and COST ^k is the cost function. When the k-th iteration ends, the cost function COST ^k is less than a certain A fixed threshold ε (the typical value of ε is 0.001), it is considered that the iteration has converged, and the iteration is stopped. The amplitude and phase of the high-resolution image at this time are the finally obtained large-field high-resolution microscopic image.

The above reconstruction process is only suitable for reconstructing monochrome images. If a true color image needs to be reconstructed, each LED unit uses red light, green light, and blue light to illuminate the object respectively. Then image reconstruction is performed for each illumination wavelength, and the reconstructed three groups of images can be combined as the red, green, and blue components of the final true-color image.

In order to test the high-speed Fourier stack imaging reconstruction method based on ring illumination, a simulation test was carried out first. The ideal phase distribution of the object to be measured is shown in Figure 4(a), and then the phase recovery results of the two Fourier stack imaging systems were simulated. First, when the maximum illumination tilt angle of the bright-field LED unit is inscribed with the numerical aperture of the objective lens, the phase results recovered from 81, 13, and 5 bright-field low-resolution images are shown in Figure 4(b1) and Figure 4(b2) ), as shown in Figure 4(b3). No obvious reconstruction error can be seen in the reconstruction results, which means that as long as the maximum illumination inclination angle of the bright field LED unit is inscribed with the numerical aperture of the objective lens, a maximum of 13 images and a minimum of 5 images are required to restore very accurately The phase distribution of the object. On the contrary, in another case, that is, when the maximum illumination tilt angle of the bright field LED unit and the numerical aperture of the objective lens cannot be inscribed, the phase results recovered from 81, 13, and 5 bright field low-resolution images are shown in Fig. 4(c1), Figure 4(c2), and Figure 4(c3). Obvious low-frequency reconstruction errors can be seen in the figure, indicating that when the maximum illumination tilt angle of the bright field LED unit cannot be inscribed with the numerical aperture of the objective lens, the phase distribution of the object cannot be accurately recovered no matter how many pictures are taken. The simulation results fully demonstrate the innovation and effectiveness of the present invention. Firstly, the device uses an annular LED plate whose illumination angle is inscribed with the numerical aperture of the objective lens, not only to reduce the number of original image captures, but also to improve the phase recovery accuracy . Because it can be seen from Figure 4 that when the maximum illumination tilt angle of the bright field LED unit is inscribed with the numerical aperture of the objective lens, the phase distribution of the object can be recovered very accurately, and a maximum of 13 images are required, which greatly improves the Imaging speed. In contrast, if the maximum illumination tilt angle of the bright field LED unit in the traditional FPM system and the numerical aperture of the objective lens cannot be inscribed, no matter how many images are taken, the phase distribution of the object cannot be accurately recovered.

In order to further test the dynamic measurement speed of the present invention, microlens array samples and Hela living cell samples were also selected for experimental testing. In the experiment, the ring-shaped LED board used contains 13 LED units, and it is used to generate 13 different angles of illumination light. The radius of the ring is 20mm, and the central wavelength of the emitted red light is 632.8nm, and the spectral bandwidth is about 20nm. The numerical aperture of the microscope objective lens used in the system is 0.4, and the magnification is 10x. A low-resolution image captured by lighting up the LED unit in the center of the ring-shaped LED board is shown in Figure 5(a). Select the image area in the box in Figure 5(a), and the bright field images of this area under 9 different illumination angles are shown in Figure 5(b). Figure 5(c) and Figure 5(d) respectively show the high-resolution light intensity distribution and phase distribution recovered after iterative reconstruction in this area, and the microlens converted by comparing the standard parameters of the sample with the measured phase The radius of curvature shows that this method can achieve stable and high-precision quantitative phase imaging.

Figure 6 shows the imaging results of label-free dynamic quantitative phase of Hela cells using this method. The high-resolution phase distribution restored by the full field of view is shown in Fig. 6(a). The image area in the dotted line box in Fig. 6(a) is selected and enlarged, as shown in Fig. 6(b). clear and distinct. Figure 6(c1)-Figure 6(c7) sequentially shows the dynamic quantitative phase imaging results of the real-time division of Hela cells in the selected area at different times (0s, 250s, 500s, 750s, 1000s, 1250s, 1500s,). It can be seen from the figure that the method of the present invention only needs to take 13 bright-field low-resolution pictures, and can very stably and accurately reconstruct the phase distribution of the object with a large field of view and high resolution, which is very suitable for label-free high-speed imaging of living cells. Quantitative Phase Microscopy.

Claims (10)

1. a high-speed Fourier stack imaging device based on ring illumination, is characterized in that comprising ring LED plate (1), stage (2), condenser (3), sample to be measured (4), microscope objective lens (5) ), an imaging tube lens (6), a camera (7), wherein the light emitted by the i-th lighted LED unit (8) on the annular LED board (1) is converged by the condenser lens (3) and becomes parallel light and illuminates the On the sample (4), the sample to be tested (4) is placed on the stage (2), part of the diffracted light passing through the sample to be tested (4) is collected by the microscope objective lens (5), and passes through the imaging tube lens (6) Converging and illuminating the imaging plane of the camera (7), and the formed light intensity map is recorded by the camera (7). 2. The high-speed Fourier stack imaging device based on ring illumination according to claim 1, characterized in that said ring-shaped LED board (1) is placed on the front focal plane position of the condenser lens (3), and the ring-shaped LED board (1) The LED unit (8) is located on the optical axis, and the rest of the LED units are evenly arranged on a ring with the center LED unit as the center radius of d; the rear focal plane of the microscope objective lens (5) and the imaging tube lens (6) The front focal plane of the camera (7) is placed on the rear focal plane position of the imaging tube lens (6), and the sample to be measured (4) on the stage (2) is adjusted to the microscope objective lens (5) during imaging. ) of the front focal plane position constitutes an infinity-corrected imaging system. 3. The high-speed Fourier stack imaging device based on ring illumination according to claim 1, characterized in that for an inverted microscope imaging system, the ring LED plate (1) and condenser lens (3) need to be installed on the sample to be tested (4) above. 4. the high-speed Fourier stack imaging device based on annular illumination according to claim 1, characterized in that the numerical aperture of the microscopic objective lens (5) is NA _obj , and the distance from the annular LED plate (1) to the sample is f , the radius of the LED ring in the ring-shaped LED board (1) is d, the light emitted by the LED unit (8) is converged by the condenser lens (3) and becomes parallel light with an angle θ _i about the optical axis, and satisfies 5. The high-speed Fourier stack imaging device based on ring illumination according to claim 1, characterized in that the illumination wavelength of the ring LED plate (1) is λ, the magnification of the microscopic objective lens (5) is Mag, and the pixel The size is Δx _cam and satisfies 6. A high-speed Fourier stack imaging device based on annular illumination, characterized in that it comprises an annular LED plate (1), a stage (2), a sample to be measured (4), a microscopic objective lens (5), and an imaging tube lens (6), camera (7), wherein the light that the LED unit (8) that lights up on the ring-shaped LED plate (1) is directly irradiated on the sample to be tested (4), and the sample to be tested (4) is captured Placed on the stage (2), part of the diffracted light passing through the sample to be measured (4) is collected by the microscope objective lens (5), and converged by the imaging tube lens (6) to illuminate the imaging plane of the camera (7), forming The light intensity map of is recorded by camera (7). 7. according to claim 1 or 6 described high-speed Fourier stack imaging devices based on ring illumination, it is characterized in that camera (7) uses color or monochrome camera, if monochrome camera, let ring LED plate ( In 1), each LED unit (8) emits red or green or blue monochromatic light respectively, and it is enough to record all monochrome images in turn with a monochrome camera (7); if it is a color camera, use the ring-shaped LED board Each LED unit (8) in (1) emits red, green and blue three-color light at the same time, and records all color images with a color camera (7). 8. A high-speed Fourier stack imaging reconstruction method based on ring illumination, characterized in that the steps are as follows: Step 1, LED unit brightness calibration: The ring-shaped LED board is used as the illumination source of the microscope, and each LED unit is lit in turn, and after irradiating a blank sample to be tested, use a high-power objective lens to collect corresponding images and calculate the red and green brightness of each LED unit. The normalized brightness correction coefficients corresponding to the three blue bands; Step 2, LED position calibration and correction: the resolution board is used as the sample to be tested, and the ring-shaped LED board is used as the illumination source of the microscope. Each of the LED units is lit in sequence, and after irradiating the sample to be tested, the under-focus and defocus h distances are collected respectively. Download the corresponding image, and then calculate the illumination angle corresponding to each LED unit in the three bands of red, green and blue through the sub-pixel registration algorithm, and then determine the position of the ring-shaped LED board through nonlinear regression; if there is an LED unit on the ring If the inclined plane wave is not inscribed with the numerical aperture of the objective lens, then the position of the annular LED board is changed by adjusting the fine-tuning translation stage, and the second step is repeated until the inclined plane waves generated by the LED units on all rings are inscribed with the numerical aperture of the objective lens; Step 3, original image acquisition: the ring-shaped LED board is used as the illumination source of the microscope, and each LED unit is sequentially lit, and the corresponding original low-resolution bright field image is collected after irradiating the sample to be tested; Step 4, original image preprocessing, including threshold denoising and brightness correction. Firstly, threshold denoising is performed on the original low-resolution images taken according to the average value of dark current noise of the camera, and then each image is divided by The normalized brightness correction coefficient obtained in step 1; Step 5, high-resolution image initialization: average all captured low-resolution bright-field images, and then initialize the amplitude and phase of the high-resolution images by upsampling; Step 6, iterative reconstruction: use Fourier stack imaging technology based on pixel merging to perform synthetic aperture calculation on each image collected in the frequency domain one by one, and gradually reduce the update coefficient, and use the cost function value as the criterion. When it is less than a given threshold, the iteration is stopped, and the amplitude and phase of the high-resolution image at this time are the finally obtained large-field high-resolution microscopic image. 9. The high-speed Fourier stack imaging reconstruction method based on ring illumination according to claim 8, characterized in that in step 2: the process of calibrating the position of the ring LED board is as follows: the resolution board is used as the sample to be tested, and the ring LED board is used as the sample to be tested. The illumination source of the microscope contains N LED units in total, N≤13, and each of the LED units is lit in sequence, and after irradiating the sample to be tested, the corresponding focused images under the focus and defocus h distances are respectively collected and defocused image According to the angular spectrum diffraction theory, the focused image corresponding to the central LED unit Propagate the h distance numerically along the optical axis to obtain a numerical defocused image Each out-of-focus image is then computed by a sub-pixel registration algorithm Relative to numerically defocused images offset (PY _x , PY _y ), then the spatial frequency vector of the illumination light corresponding to the i-th LED unit is <mrow><msub><mi>u</mi><mi>i</mi></msub><mo>=</mo><mrow><mo>(</mo><msub><mi>u</mi><mi>x</mi></msub><mo>,</mo><msub><mi>u</mi><mi>y</mi></msub><mo>)</mo></mrow><mo>=</mo><mfrac><mrow><mn>2</mn><mi>&amp;pi;</mi></mrow><mi>&amp;lambda;</mi></mfrac><mrow><mo>(</mo><mfrac><mrow><msub><mi>PY</mi><mi>x</mi></msub></mrow><msqrt><mrow><msubsup><mi>PY</mi><mi>x</mi><mn>2</mn></msubsup><mo>+</mo><msubsup><mi>PY</mi><mi>y</mi><mn>2</mn></msubsup><mo>+</mo><msup><mi>h</mi><mn>2</mn></msup></mrow></msqrt></mfrac><mo>,</mo><mfrac><mrow><msub><mi>PY</mi><mi>y</mi></msub></mrow><msqrt><mrow><msubsup><mi>PY</mi><mi>x</mi><mn>2</mn></msubsup><mo>+</mo><msubsup><mi>PY</mi><mi>y</mi><mn>2</mn></msubsup><mo>+</mo><msup><mi>h</mi><mn>2</mn></msup></mrow></msqrt></mfrac><mo>)</mo></mrow></mrow> Among them, (u _x , u _y ) is the spatial frequency along the xy direction, and λ is the wavelength of the illuminating light; Finally, determine the position of the annular LED board through nonlinear regression, the formula is: <mrow><mi>Q</mi><mrow><mo>(</mo><msub><mi>u</mi><mi>i</mi></msub><mo>,</mo><msup><mi>&amp;theta;</mi><mn>0</mn></msup><mo>,</mo><msup><mi>&amp;Delta;x</mi>mi><mn>0</mn></msup><mo>,</mo><msup><mi>&amp;Delta;y</mi><mn>0</mn></msup><mo>,</mo><msup><mi>f</mi><mn>0</mn></msup><mo>)</mo></mrow><mo>=</mo><munder><mo>&amp;Sigma;</mo><mi>i</mi></munder><mo>|</mo><msub><mi>u</mi><mi>i</mi></msub><mo>-</mo><msubsup><mi>u</mi><mi>i</mi><mn>0</mn></msubsup><msup><mo>|</mo><mn>2</mn></msup></mrow> x _i ＝d[cos(θ)m _i +sin(θ)n _i ]+Δx y _i ＝d[-sin(θ)m _i +cos(θ)n _i ]+Δy <mrow><msub><mi>u</mi><mi>i</mi></msub><mo>=</mo><mfrac><mrow><mn>2</mn><mi>&amp;pi;</mi></mrow><mi>&amp;lambda;</mi></mfrac><mfrac><msub><mi>x</mi><mi>i</mi></msub><mi>f</mi></mfrac></mrow> <mrow><msub><mi>v</mi><mi>i</mi></msub><mo>=</mo><mfrac><mrow><mn>2</mn><mi>&amp;pi;</mi></mrow><mi>&amp;lambda;</mi></mfrac><mfrac><msub><mi>y</mi><mi>i</mi></msub><mi>f</mi></mfrac></mrow> Among them, Q(...) is the objective function of the nonlinear regression method, (θ,Δx,Δy,f) are the four position parameters of the updated circular LED board, which are the rotation error, the translation error in the x direction, The translation error and focus error in the y direction, (θ ⁰ ,Δx ⁰ ,Δy ⁰ ,f ⁰ ) are the initial position parameters of the ring LED board, Indicates the nonlinear regression operation, d is the distance between two adjacent LED units of the ring-shaped LED board, ( _xi , y _i ) indicates the spatial position coordinates of the i-th LED unit, λ is the wavelength of the illumination light, (m _i ,n _i ) is the row number and column number corresponding to the i-th LED unit. 10. The high-speed Fourier stack imaging reconstruction method based on ring illumination according to claim 9, wherein if the four position parameters of the updated ring LED board are respectively θ=0, Δx=0, Δy=0 , There is no need to perform position correction, otherwise it is necessary to change the position of the annular LED board by adjusting the fine-tuning translation stage, and repeat step 2 until the inclined plane waves generated by the LED units on all the rings are inscribed by the numerical aperture of the objective lens, that is, to achieve