Automatic imaging method, system and equipment

Abstract

The invention discloses an automatic imaging method, and a system and equipment thereof. According to the method, a sample interference surface and a sample boundary are positioned, and then non-equidistant three-dimensional OCT image acquisition is carried out. According to the non-equidistant acquisition strategy based on the sample structure characteristics, the acquisition stepping can be automatically adjusted according to the complexity of different areas, the large stepping is adopted in the Non-ROI area sample arm to realize rapid imaging, and the small stepping is adopted in the ROI area sample arm to maintain high-resolution imaging, so that the redundant data acquisition is reduced, and the acquisition efficiency is improved. According to the dynamic acquisition strategy, the total imaging time is remarkably shortened while high-resolution imaging of the region of interest is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of full-field optical coherence tomography, and in particular relates to an automated imaging method and a system and equipment thereof. Background Art

[0002] Optical coherence tomography (OCT) systems, especially full-field optical coherence tomography (FF-OCT), play an important role in the field of biomedical imaging. They have the advantages of high resolution and label-free, and can perform three-dimensional imaging of biological samples at the micron level, providing real-time histological information for biological tissue research. However, traditional OCT systems still face a series of challenges in their application:

[0003] Long imaging time: When imaging complex heterogeneous and irregular samples such as three-dimensional cell culture models, traditional OCT systems use isometric scanning, which makes the imaging process time-consuming. This not only limits the efficiency of real-time imaging, but also may have an adverse effect on sample activity due to the long imaging process, thereby interfering with the accuracy of experimental results.

[0004] Acquisition efficiency and data storage issues: Existing research focuses on improving the imaging speed of OCT systems, but ignores the optimization of acquisition efficiency. Taking the FF-OCT system as an example, when facing complex three-dimensional cell culture models, its fixed step-by-step acquisition method leads to low acquisition efficiency and generates a large amount of data, which brings a heavy burden to subsequent data processing and storage.

[0005] Disadvantages of manual operation of interference surface positioning: The interference surface positioning of traditional OCT systems usually relies on manual operation, and the operator needs to observe the interference fringes with the naked eye to determine the position of the interference surface. This method not only requires high experience and skills of the operator, but also easily introduces human errors due to the large number of layers caused by high-resolution layer-by-layer scanning. In particular, for active biological samples such as cell spheroids or organoids in normal culture, due to their uneven size and distribution, they will float in the culture medium and produce motion artifacts, making it difficult to directly observe the interference fringes, which in turn leads to imaging difficulties, and the accuracy and efficiency of manual positioning are greatly reduced.

[0006] In summary, when faced with imaging of complex samples, traditional OCT systems have problems such as long imaging time, low acquisition efficiency, and inaccurate interference surface positioning. There is an urgent need for a more efficient, accurate, and highly automated imaging method and related systems and equipment to overcome these limitations. Summary of the invention

[0007] The purpose of the present invention is to overcome the above-mentioned deficiencies of the prior art and to provide an automated imaging method and a system and device thereof.

[0008] The present invention is implemented as follows: In a first aspect, the present invention provides an automated imaging method, the method using non-isometric three-dimensional OCT image acquisition; The non-equidistant three-dimensional OCT image acquisition is to reconstruct the OCT image by focusing layer images under different phase shift amounts; according to the grayscale variance of the OCT image, the region of interest and the region of non-interest of the current sample focal plane are screened out, the sample arm is moved to the next focal plane layer with a small step length in the region of interest, and the sample arm is moved to the next focal plane layer with a large step length in the region of non-interest.

[0009] Preferably, in the non-equidistant three-dimensional OCT image acquisition, whether it is still in the current focal plane layer is identified based on the comparison between the number of OCT images and the sliding window size. If the number of OCT images is greater than the sliding window size, the sample arm is moved to the next focal plane layer with a large step size.

[0010] Preferably, before acquiring non-equidistant three-dimensional OCT images, it is necessary to locate the sample interference surface and locate the focal plane layer at the sample boundary.

[0011] Preferably, the positioning sample interference surface comprises: The OCT image is reconstructed by the focal surface layer image under different phase shift amounts; the first sliding window change rate is calculated according to the grayscale variance of the OCT image; if the first sliding window change rate is greater than a preset threshold value ξ1, it is considered that interference occurs in the current sample focal plane; if the first sliding window change rate is less than or equal to the preset threshold value ξ1, the reference arm is moved in the direction of increasing optical path.

[0012] More preferably, in the positioning sample interference surface, based on the comparison between the number of OCT images and the sliding window size, it is identified whether it is still in the current focal plane layer. If the number of OCT images is less than or equal to the sliding window size, the reference arm is moved in the direction of increasing optical path.

[0013] Preferably, the positioning focal plane layer at the sample boundary comprises: The second sliding window change rate is calculated according to the image clarity of the focal surface layer image; If the change rate of the second sliding window is less than the preset threshold ξ3, the current focal plane layer is considered to be the sample boundary. If the change rate of the second sliding window is greater than or equal to the preset threshold ξ3, the sample arm is moved to the next focal plane layer.

[0014] More preferably, the positioned focal plane layer is in the sample boundary, and based on the comparison between the number of images in the current focal plane layer and a preset threshold value ξ2, it is identified whether it is still in the current focal plane layer. If the number of images in the current focal plane layer is less than or equal to the preset threshold value ξ2, the sample arm is moved to the next focal plane layer.

[0015] Preferably, the direction of movement of the sample arm during the non-equidistant three-dimensional OCT image acquisition process is opposite to the direction of movement of the sample arm during the process of positioning the focal plane layer at the sample boundary.

[0016] In a second aspect, the present invention provides an automated imaging system, comprising: The sample interference surface positioning module is responsible for positioning the sample interference surface to achieve equal optical path lengths between the sample arm and the reference arm; A sample boundary positioning module is responsible for positioning the focal plane layer at the sample boundary; The image acquisition module is responsible for acquiring non-equidistant three-dimensional OCT images of the sample.

[0017] In a third aspect, the present invention provides an automated imaging device, comprising an illumination module, an imaging module, and a control module, wherein the imaging module comprises a reference arm module, a sample arm module, and a detection arm module; The illumination module provides a low-coherence light source, and then sends the light to the reference arm module and the sample arm module respectively; the reflected light returned by the sample arm module interferes with the reflected light returned by the reference arm module and enters the detection arm module; the detection arm module obtains an interference signal; The lighting module includes a low-coherence light source, a first convex lens f1, an aperture stop AS, a variable field stop FS, a second convex lens f2, and a beam splitting cube prism BS; light emitted by the low-coherence light source is focused by the first convex lens f1 and imaged at the aperture stop AS, and the image point of the light source passes through the variable field stop FS and the rear second convex lens f2 before entering the beam splitting cube prism BS for beam splitting; The sample arm module comprises a first objective lens MO and a first displacement stage; a culture dish for placing the sample to be tested is arranged on the first displacement stage, and the objective lens is located above the culture dish on the first displacement stage; The reference arm module includes a filter ι, a second objective lens MO, a reference mirror, a phase shifter, and a second displacement stage connected in sequence through an optical path; the phase shifter is used to perform phase shifting to achieve the acquisition of current focal plane layer images with different phase shift amounts; the second displacement stage is used to move the reference arm; The detection arm module includes a third convex lens f3 and a camera; the interference light passes through the third convex lens f3 and forms a clear image on the camera; the first translation stage is used to move the sample arm; The control module is used to control the first translation stage, the second translation stage, the phase shifter, and the camera.

[0018] The beneficial effects of the present invention are: The present invention adopts an automated interference surface detection method, which significantly reduces the variability caused by manual operation and human factors, improves the repeatability of focal plane detection and the reproducibility of experiments, and can achieve fast and accurate interference surface positioning with the support of an automatic control system.

[0019] The present invention also proposes a non-equidistant acquisition strategy based on sample structural characteristics, which can automatically adjust the acquisition step according to the complexity of different regions. In the Non-ROI region, the sample arm uses a large step to achieve fast imaging, and in the ROI region, the sample arm uses a small step to maintain high-resolution imaging, thereby reducing redundant data acquisition and improving acquisition efficiency. This dynamic acquisition strategy significantly shortens the total imaging time while ensuring high-resolution imaging of the region of interest.

[0020] In addition, the present invention is not limited to imaging three-dimensional cultured cell models with obvious structural stratification, but is also applicable to imaging of other complex samples with obvious optical scattering characteristics, such as imaging of the internal structure of three-dimensional cultured cell models. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 It is a schematic diagram of the structure of the OCT automated imaging device provided in an embodiment of the present invention.

[0023] Figure 2 It is a general flow chart of the OCT automated imaging method provided by an embodiment of the present invention.

[0024] Figure 3 It is a flow chart of positioning the interference surface of a sample in the OCT automated imaging method provided in an embodiment of the present invention.

[0025] Figure 4 It is a flow chart for locating sample boundaries in the OCT automated imaging method provided in an embodiment of the present invention.

[0026] Figure 5 It is a flow chart of non-isometric three-dimensional OCT image acquisition in the OCT automated imaging method provided in an embodiment of the present invention.

[0027] Figure 6The OCT automated imaging method provided in the embodiment of the present invention is used to collect and verify the results, wherein (a)-(c) are respectively the microscope images and interference images of the bottom surface of the culture dish, the inner epidermis of the onion and the three-dimensional culture model of RG cells; (d) is the grayscale variance change curve of the interference surface detection process; (e) is the Brenner gradient value change curve of the boundary detection process; (f)-(g) are respectively the grayscale variance value change curves of the equidistant acquisition process and the non-equidistant acquisition process. DETAILED DESCRIPTION

[0028] The interference surface positioning of traditional OCT systems relies on manual operation, requiring the interference fringes to be observed with the naked eye to determine the exact position of the interference surface. In addition, due to the need for high-resolution layer-by-layer scanning, there are many layers, the imaging time is long, and it is easy to produce human errors. For active biological samples such as cell spheroids or organoids in normal culture, the size and distribution are uneven, and floating in the culture medium will cause motion artifacts, and it is particularly difficult to image complex samples.

[0029] To solve this problem, the present invention adopts an automated imaging method, which controls the reference arm displacement stage to move automatically and combines the phase shifter to change the phase to collect images, and uses grayscale variance calculation and sliding window change rate to determine the position of the interference surface. For example, in the HepaRG cell three-dimensional culture model experiment, the reference arm displacement stage automatically steps in the direction of increasing the optical path by, for example, 10μm, and accurately finds the interference surface in a short time (within one minute), while traditional manual operation takes longer and the accuracy is easily affected by the operator's experience, effectively reducing human errors and improving positioning efficiency.

[0030] In the detection of sample boundaries, the present invention uses image clarity calculation and sliding window change rate to determine the sample boundary. During the automatic movement of the sample arm, the sample boundary can be accurately found by calculating the clarity values ​​of images at different positions. For example, in the boundary detection of the HepaRG cell three-dimensional culture model, the sample arm automatically steps down by, for example, 2 μm, and determines the boundary based on the change trend of the clarity value (the image gradually blurs and the gradient value tends to be stable), ensuring that the subsequent acquisition can obtain the complete three-dimensional structure of the sample.

[0031] The present invention also proposes a non-equidistant acquisition strategy based on the structural characteristics of the sample. According to the structural differences between the ROI and Non-ROI regions, the sample arm step distance is automatically adjusted by analyzing the changes in the grayscale variance value of the OCT image. Fast imaging with large steps (such as 10μm) in the Non-ROI region reduces the amount of data, and high-resolution imaging with small steps (such as 2μm) in the ROI region retains key information. Compared with traditional isometric acquisition, in the imaging of the HepaRG cell three-dimensional culture model, non-equidistant acquisition greatly reduces data acquisition in areas with sparse cell structures, maintains high resolution in key areas such as the bottom cluster structure, effectively balances imaging speed and quality, improves acquisition efficiency, reduces the burden of data storage, and is suitable for imaging of a variety of complex samples with a wide range of applications.

[0032] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0033] like Figure 1 As shown, an automated imaging device provided by an embodiment of the present invention includes a lighting module, an imaging module, and a control module, wherein the imaging module includes a reference arm module, a sample arm module, and a detection arm module; The illumination module provides a low-coherence light source, and then sends the light to the reference arm module and the sample arm module respectively; the reflected light returned by the sample arm module interferes with the reflected light returned by the reference arm module and enters the detection arm module; the detection arm module obtains an interference signal; The illumination module includes a low-coherence light source, a first convex lens f1, an aperture diaphragm AS, a variable field diaphragm FS, a second convex lens f2, and a beam splitting cube prism BS; the illumination module serves as the light source supply part of the entire imaging system, and the light emitted by the core component, the low-coherence light source, first passes through the first convex lens f1 for focusing imaging. This focusing process enables the light emitted by the light source to form a clear image point at the aperture diaphragm AS. The function of the aperture diaphragm AS is to limit the aperture size of the light beam, thereby effectively controlling the light energy entering the subsequent optical system, and helping to reduce the interference of stray light and improve the contrast and clarity of the imaging. Then, the light source image point passing through the aperture diaphragm AS continues to propagate, and through the variable field diaphragm FS, the diaphragm can flexibly adjust the field size according to the actual imaging requirements to adapt to the imaging of samples of different sizes and structures. Finally, the light enters the beam splitting cube prism BS after passing through the rear second convex lens f2 for beam splitting operation, and the split light is sent to the reference arm module and the sample arm module respectively, providing the necessary beam basis for the subsequent interference imaging process.

[0034] The sample arm module includes a first objective lens MO and a first displacement stage; a culture dish for placing the sample to be tested is provided on the first displacement stage, and the first objective lens MO is located above the culture dish on the first displacement stage; the first objective lens MO in the sample arm module can converge and collimate the light transmitted from the illumination module to ensure that the light is evenly irradiated onto the sample to be tested (such as biological samples such as cell balls) placed on the first displacement stage at a suitable angle and intensity. The first displacement stage is a device that can accurately control the position of the sample. By adjusting its position, the sample can be moved in the direction of the optical axis, which is crucial for finding different focal plane layers of the sample and realizing three-dimensional imaging. During the imaging process, the main task of the sample arm module is to accurately transmit the light signal reflected from the sample back to the detection arm module so as to interfere with the light returned by the reference arm module.

[0035] The reference arm module includes a filter ι, a second objective lens MO, a reference mirror RM, a phase shifter, and a second displacement stage connected in sequence through an optical path; the phase shifter is used to perform phase shifting to realize the acquisition of current focal plane layer images with different phase shift amounts; the second displacement stage is used to move the reference arm; the filter ι in the reference arm module is mainly used to screen light in a specific wavelength range to ensure that the spectral characteristics of the reference light meet the requirements of the imaging system and reduce the influence of unnecessary spectral components on the imaging results. The second objective lens MO also plays the role of converging and collimating light, so that the reference light can be irradiated onto the reference mirror RM in the form of parallel light. The role of the second displacement stage in the reference arm module is to accurately control the length of the reference arm, and to achieve optical path matching by changing the length of the reference arm, which is one of the key steps to obtain high-quality interference signals. The phase shifter is a key component in the reference arm module, which can realize the acquisition of current focal plane layer images with different phase shift amounts by changing the phase of the reference light, and provide rich phase information for subsequent image reconstruction and analysis.

[0036] The detection arm module includes a third convex lens f3 and a camera; the interference light passes through the third convex lens f3 and forms a clear image on the camera; the first translation stage is used to move the sample arm; the third convex lens f3 in the detection arm module is responsible for focusing and imaging the light signal returned from the sample arm module and the reference arm module and after interference, so that the interference signal can be clearly imaged on the camera. As the final receiving device of the imaging system, the camera converts the light signal into an electrical signal and records the intensity and phase information of the interference image. These image data will be transmitted to the control module for subsequent processing and analysis.

[0037] The control module can achieve precise control of the entire imaging system through the host computer and the controller. It can control the moving speed, displacement and moving direction of the first translation stage, so as to achieve precise positioning and movement of the sample arm in three-dimensional space. For the second translation stage, the control module can accurately adjust its displacement to ensure the accurate change of the reference arm length to achieve optical path matching. At the same time, the control module can also control the phase shift amount and speed of the phase shifter, as well as the exposure time, acquisition frame rate and other parameters of the camera to ensure the automation, precision and efficiency of the entire imaging process.

[0038] For example, the light from the sample arm passes through the first objective lens MO to evenly illuminate the sample. The light from the reference arm passes through the filter ι and the second objective lens MO to be emitted as parallel light and illuminate the reference mirror RM. The sample arm and the reference arm are adjusted to make the sample light and the reference light interfere, and the interference light passes through the third convex lens f3 to form a clear image on the camera.

[0039] For example, the automated imaging device is suitable for an OCT system, which may be an imaging system that can realize tomographic scanning, such as FF-OCT, time domain OCT, swept frequency OCT or spectral domain OCT. It is also suitable for layer-by-layer scanning microscope systems such as laser confocal microscope and light sheet microscope, and can be adjusted according to actual conditions.

[0040] This embodiment also provides an automated imaging method based on the above device, taking FF-OCT as an example. Figure 2 The following steps are involved: Step S1, positioning the sample interference surface to achieve equal optical path lengths of the sample arm and the reference arm; Step S2, positioning the focal plane layer at the sample boundary; Step S3, collecting non-equidistant three-dimensional FF-OCT images of the sample; Specifically, in step S1, before the automatic search for the sample interference surface is performed, the strong backscattered signal stripes on the bottom of the culture dish are first used for positioning, and then the first displacement stage of the sample arm is moved to search the sample position and the second displacement stage of the reference arm is moved to find the interference surface of the sample. Figure 3 , the specific steps are as follows: S11. Use the bottom surface of the culture dish to find the interference fringes: First, place the culture dish containing samples such as cell spheroids on the first objective lens, adjust the focal plane to find the cell spheroid sample in the field of view, then move the sample arm first stage to find the focal plane of the bottom surface of the culture dish, and then move the reference arm second stage to find the interference surface of the bottom surface. When fringes are observed, it means that interference occurs, that is, the coherence gate and the focal plane coincide.

[0041] S12. Move the sample arm to find the clear focal plane of the sample: Move the sample arm translation stage to find the clear focal plane of the cell ball. At this time, since the light irradiating the cell ball is refracted by the bottom surface of the culture dish, the actual optical path of the sample arm increases, while the optical path of the reference arm does not change. As a result, a large optical path difference l0 is generated between the sample arm and the reference arm. l0 is greater than the coherence length of the light source, resulting in no interference. Therefore, in order to ensure subsequent stable and clear imaging, the reference arm needs to be adjusted in the direction of increasing the optical path.

[0042] The bottom of the culture dish usually has a relatively flat structure, and its reflected light can form relatively stable and easy-to-observe interference fringes, which provides a reference position for the subsequent search for the sample interference surface. After finding the interference fringes on the bottom of the culture dish, the clear focal plane of the sample is found by moving the sample arm. This process requires precise control of the movement of the sample arm to ensure that the sample is in the best imaging position.

[0043] Compared with planar structures such as the bottom of a culture dish, the interference fringes of a cell sphere are difficult to directly observe due to its complex structure, so the following steps S13-S15 are performed.

[0044] S13, performing phase shifting through the reference arm phase shifter to collect current focal plane layer images with different phase shift amounts; performing static reconstruction through the focal plane layer images with different phase shift amounts to calculate the FF-OCT image of the sample; Since the FF-OCT system acquires layer images, it can directly acquire the current focal plane layer images with different phase shift amounts through phase shift; if other OCT systems are used, such as the time domain OCT system, the scanning mode is A-scan. In order to realize the imaging of layer images, it is necessary to control the galvanometer to scan in the X and Y directions to obtain the surface image. For the swept frequency OCT and spectral domain OCT systems, the scanning mode is B-scan. In order to realize the imaging of layer images, it is necessary to control the galvanometer to scan in the X direction to obtain the surface image.

[0045] S14, update the number of FF-OCT images n=n+1, and then determine whether n is greater than the sliding window size, if so, proceed to step S15, if not, move the reference arm in the direction of increasing optical path with a step length w1, and then return to step S13; in this embodiment, w1 is 10 μm; S15, calculate the grayscale variance of the current focal layer FF-OCT image; calculate the first sliding window change rate based on the grayscale variance, and determine whether the current first sliding window change rate is greater than a preset threshold value ξ1 (the value in this embodiment is 0.05, which can be replaced based on experience). If so, it is considered that interference occurs in the current sample focal plane; otherwise, move the reference arm in the direction of increasing optical path with a step size w1, and then return to step S13.

[0046] Step S13 is as follows: the host computer controls the phase shifter of the reference arm to shift the optical path difference between the sample arm and the reference arm. At the same time, each time the phase is shifted, the camera collects an image of the focal plane layer. The image interference signal is (1) In the formula is the DC term, is the sample FF-OCT image information to be obtained, is the phase difference, is the phase shift introduced by the reference arm phase shifter.

[0047] The static reconstruction described in step S13 can be achieved by using a four-step phase-shifting interferometer method, or a two-step, three-step, five-step or six-step phase-shifting interferometer method. The specific imaging method can select a suitable phase-shifting step according to the system application scenario and in combination with imaging accuracy and efficiency.

[0048] For example, the four-step phase-shift interferometry method is specifically to collect The four pictures under the four phase shift quantities are , then the FF-OCT image of the sample can be expressed as (2) In the process of moving the reference arm translation stage, the four-step phase shifting method can be used to reconstruct the FF-OCT image containing the current layer information of the cell sphere. By calculating and evaluating the effective information of the FF-OCT image, it can be determined whether the interference surface of the cell sphere is found, that is, whether the cell sphere reaches the position where the coherence gate and the focal plane coincide.

[0049] In step S15, the first sliding window change rate is calculated according to the grayscale variance, wherein the grayscale variance is calculated as follows: (3) In the formula and Respectively represent the horizontal and vertical pixel numbers of the image, Indicates the current focal layer FF-OCT image The pixel value at Indicates that all pixels in the image are averaged. If the calculated grayscale variance value changes suddenly, it means that the sample arm and the reference arm interfere with each other at this position, that is, the FF-OCT image contains effective information, indicating that the interference surface of the cell ball is found. At the same time, the more effective information, the larger the grayscale variance value.

[0050] First sliding window change rate The calculation of is as follows: (4) Where abs() represents the absolute value function; Represents the average value of the grayscale variance in sliding window windows1; It represents the average value of the grayscale variance in sliding window windows2. The sizes of sliding windows windows1 and sliding windows windows2 are equal, and the position of window2 is one position forward compared to the position of window1. When sliding, the relative positions of the two windows are kept unchanged and gradually moved to the right.

[0051] After finding the interference surface of the cell sphere, the three-dimensional data of the cell sphere can be collected by moving the sample arm stage and cooperating with the camera and phase shifter. The FF-OCT three-dimensional structure of the cell sphere can be reconstructed by subsequent data processing and analysis. However, at this time, the focal plane of the first objective lens of the sample arm and the coherence gate of the reference arm are located inside the cell sphere rather than at the boundary. Therefore, in order to collect the complete three-dimensional structure of the cell sphere, it is necessary to move the sample arm stage to place the focal plane and coherence gate at the boundary of the cell sphere.

[0052] Specifically, Figure 4 Step S2 specifically includes: S21, collecting the current focal plane layer image; S22, update the number of images in the current focal plane layer m=m+1, and then determine whether m is greater than a preset threshold value ξ2, if so, proceed to step S23, if not, move the sample arm to the next focal plane layer with a step length w2, and then return to step S21; in this embodiment, w2 is 2μm; m=n, n represents the number of FF-OCT images; in this embodiment, the threshold value ξ2 can be ξ2=2Q-1, where Q represents the sliding window size; S23, calculate the image clarity of the current focal plane layer image; calculate the second sliding window change rate based on the image clarity, and determine whether the current second sliding window change rate is less than a preset threshold ξ3 (the value in this embodiment is 0.005, which can be replaced based on experience). If so, the current focal plane layer is considered to be the sample boundary. Otherwise, the sample arm is moved to the next focal plane layer with a step size of w2, and then returns to step S21.

[0053] In this embodiment, the image clarity is measured using the Brenner gradient function. Since the focal plane and the coherence gate position remain unchanged during the movement of the sample arm stage, the focal plane and the coherence gate are placed at their boundaries by moving the cell ball position. At this time, the cell ball gradually becomes out of focus in the image captured by the camera until it becomes blurred. By judging the clarity of the images captured at different positions, it can be determined whether the cell ball has moved to the predetermined position. Each time the camera captures an image, the Brenner gradient value of the image is calculated as

[0054] (5) As the image becomes more and more blurred, the Brenner gradient value calculated by the image tends to be stable. The sliding window is used to find the point where the function value is stable, which is the defocus position.

[0055] For the image clarity evaluation function, the present invention is not limited to the Brenner gradient function, but also includes various image clarity evaluation functions such as the Tenengrad gradient function, the Laplacian gradient function and the grayscale variance function. The optimal evaluation function can be selected according to the application scenario and in combination with accuracy and efficiency.

[0056] Second sliding window change rate The calculation of is as follows: (6) Where abs() represents the absolute value function; Represents the average value of image clarity in sliding window windows1; It represents the average value of the image clarity in sliding window windows2. The sizes of sliding windows windows1 and sliding windows windows2 are equal. The initial position of window1 remains unchanged. When sliding, only window2 is gradually slid.

[0057] Specifically, step S3 is based on the sample-related ROI non-equidistant three-dimensional FF-OCT synchronous timing automated acquisition protocol to achieve improved acquisition efficiency. The acquisition process is controlled by the host computer at the same time. The sample arm first displacement stage, the reference arm phase shifter and the camera work together to synchronize the timing operation. The specific steps include:

[0058] S31, collecting current focal plane layer images with different phase shift amounts; performing static reconstruction through the focal plane layer images with different phase shift amounts to calculate the FF-OCT image of the sample; the static reconstruction method here can refer to step S13; S32, update the number of FF-OCT images n=n+1, and then determine whether n is greater than the sliding window size, if so, proceed to step S33, if not, move the sample arm to the next focal plane layer with a step length w3, and then return to step S31; in this embodiment, w3 takes a value of 10 μm; S33, calculate the grayscale variance of the current focal plane FF-OCT image; construct a grayscale variance value change curve of the FF-OCT image according to the grayscale variance, screen out the region of interest (ROI) and the non-region of interest of the current sample focal plane, move the sample arm to the next focal plane layer with a step length of w4 in the region of interest, and then return to step S31; move the sample arm to the next focal plane layer with a step length of w3 in the non-region of interest (Non-ROI), and then return to step S31; in this embodiment, w4 takes a value of 2μm; The change rate of the third sliding window is calculated based on the grayscale variance, and it is determined whether the current change rate of the third sliding window is greater than a threshold value ξ4 (the value in this embodiment is 0.001, which can be replaced based on experience). If so, the current sample focal plane reaches the other boundary of the sample and image acquisition stops; otherwise, the sample arm is moved to the next focal plane layer with a step size of w4, and then returns to step S31.

[0059] The third sliding window change rate The calculation of is as follows: (7) The settings of sliding window windows1 and sliding window windows2 have the same change rate as the first sliding window. The sizes of sliding window windows1 and sliding window windows2 are equal. The position of window2 is one position forward compared to the position of window1. When sliding, the relative positions of the two windows are kept unchanged and gradually moved to the right.

[0060] The moving direction of the sample arm to the next focal plane layer in the above steps S32 and S33 is opposite to the moving direction of the sample arm to the next focal plane layer in step S2.

[0061] In step S3, after determining the coincidence of the coherence gate and the focal plane and the position of the sample boundary, four different phase images of the current position are collected, and then the reference arm displacement stage moves in a large step toward the other end of the sample. Each time the camera and the phase shifter move one step, they will collaboratively collect four different phase images, and at the same time, the four-step phase shifting method is used for static reconstruction to obtain the FF-OCT image and calculate the grayscale variance of the FF-OCT image. When the sample coincides with the coherence gate, the FF-OCT image will contain the structural information of the current layer of the sample. At this time, the grayscale variance value calculated by the FF-OCT image will mutate compared to the previous one, and the grayscale variance value will increase as the FF-OCT image contains more structural information. Therefore, for biological samples with complex structures, such as organoids, non-equidistant ROI acquisition of the biological sample structure can be achieved according to the amount of effective information contained in the FF-OCT images of different layers. Large steps are collected in the Non-ROI area to increase the acquisition speed, and small steps are collected in the ROI area to retain more effective information.

[0062] In order to realize non-equidistant 3D acquisition, it is necessary to automatically distinguish the boundary between ROI and Non-ROI, and determine the boundary between ROI and Non-ROI by using the structural difference between ROI and Non-ROI regions, that is, the change in the grayscale variance value of FF-OCT images acquired in ROI and Non-ROI regions. Figure 6 Middle (b) - Figure 6(d) shows three FF-OCT image grayscale variance value change curves from the cell ball model detection experiment, and the black dots represent the boundaries of ROI and Non-ROI. In step S33, the grayscale variance value change curve of the FF-OCT image is constructed based on the grayscale variance, and the region of interest (ROI) and non-region of interest of the current sample focal plane are screened out. Specifically, according to the grayscale variance value change curve of the FF-OCT image, it is checked whether each grayscale variance value meets one of the boundary change conditions (5). If it does, it is considered that the current sample focal plane reaches the boundary of the ROI, and the sample arm is moved to the next focal plane layer with a step length of w4, and then returns to step S31; if it does not meet, the sample arm is moved to the next focal plane layer with a step length of w3, and then returns to step S31;

[0063] The boundary change conditions are specifically: (8) In the formula , Respectively represent the grayscale variance value of the xth OCT image The first and second order derivatives of .

[0064] If condition 1 above is met, then When the current grayscale variance value change curve is a descending mark point, the change curve begins to show a downward trend; if the above condition 2 is met, then When the current grayscale variance value change curve is a horizontal mark point, the change curve begins to show a horizontal and gentle trend; if the above condition 3 is met, then When the current grayscale variance value change curve is a rising mark point, the change curve begins to show an upward trend.

[0065] The present invention utilizes the above method to accurately find the boundary between different structures, that is, the boundary between the ROI and Non-ROI areas, and then change the moving step length of the first translation stage of the sample arm to achieve non-equidistant three-dimensional FF-OCT image acquisition.

[0066] Therefore, the above-mentioned imaging method is used to image the three-dimensional cell culture model. Since the developmental state of the three-dimensional cell culture model can be characterized by morphological characteristics, in order to more accurately quantify the morphological characteristics of the three-dimensional cell culture model, the obtained FF-OCT image needs to be preprocessed. The FF-OCT image is automatically segmented using the pre-trained YOLOv8 neural network model to obtain a precisely segmented three-dimensional cell culture model image. The FF-OCT image data is reconstructed in three dimensions using Amira software to reproduce the basic structure of the three-dimensional model and accurately evaluate the morphological characteristics. At the same time, based on the connected domain calculation, the volume, surface area and other morphological parameters of the reconstructed three-dimensional cell culture model are quantified. The specific steps are:

[0067] (1) Preprocess the FF-OCT image in step S3, binarize each XY image after preprocessing, then fill the holes, and save the filled image as three-dimensional image data.

[0068] (2) The connected domains of the three-dimensional data after hole filling are marked, and an independent label is generated for each connected domain.

[0069] (3) Calculate the actual volume pixel number and actual surface area pixel number of each connected domain.

[0070] (4) The single-pixel volume and single-pixel surface area are obtained through the lateral resolution and longitudinal resolution, and then the actual volume and actual surface area are calculated.

[0071] Since FF-OCT has great advantages in the detection of three-dimensional cell culture models, the acquired images can be used to analyze various morphological parameters of the three-dimensional cell culture model, including volume, surface area, etc. The calculation formula of the main morphological parameters is: (1) Volume of a single 3D cell culture model (9) in, Indicates The actual volume of the 3D cell culture model, Indicates A three-dimensional cell culture model The number of volume pixels collected with this step size, Indicates The step length, Indicates the lateral resolution.

[0072] (2) Surface area of ​​a single 3D cell culture model (10) in, Indicates The actual surface area of ​​the 3D cell culture model, Indicates A three-dimensional cell culture model The number of surface area pixels collected with this step size.

[0073] (3) Total volume of 3D cell culture model (11) in, Represents the total volume of the 3D cell culture model.

[0074] (4) Total surface area of ​​3D cell culture model (12) in, Represents the total surface area of ​​the 3D cell culture model.

[0075] In order to evaluate the universality of the automated interference surface detection and non-equidistant three-dimensional OCT image acquisition of the OCT automatic imaging method, a FF-OCT imaging experiment was first performed on a three-dimensional culture model of HepaRG cells grown in microporous hydrogels. This example focuses on verifying the difference between the method of the present invention and the traditional manual method of finding interference surfaces, and compares the difference in efficiency between the traditional isometric acquisition and non-equidistant acquisition of the FF-OCT system.

[0076] The backscattered absorption signal of the HepaRG cell three-dimensional culture model is weak, and the interference fringes are difficult to observe with the human eye. Figure 6 Middle (a)- Figure 6 (c) shows the comparison of the bottom surface of the culture dish, the inner epidermis of the onion, and the three-dimensional model of HepaRG cells cultured in microporous hydrogel before and after interference. It can be seen that the interference fringes of the three-dimensional culture model of HepaRG cells are almost impossible to directly observe. Using the automated interference surface detection algorithm, according to Figure 3 Schematic diagram of the steps, based on the reference of interference on the bottom surface of the culture dish, the sample arm is moved to a focal plane in the HepaRG cell 3D culture model. The reference arm translation stage automatically steps in the direction of increasing the optical path by 10 μm, and the interference surface of the HepaRG cell 3D culture model can be located within one minute. Figure 6 (d) shows the grayscale variance curve of the FF-OCT image collected during the automated interferometer positioning process. The two sub-images represent the FF-OCT images at the start and end of the interferometer detection process. By comparison, it is found that the grayscale variance value of the FF-OCT at the end position changes suddenly because it contains structural information. Therefore, it is judged that the focal plane and the interferometer at this position coincide. Furthermore, in order to collect the complete three-dimensional structural information of the HepaRG cell three-dimensional culture model, it is necessary to start collecting from the model boundary. The sample arm automatically steps down at 2μm. Figure 6 (e) shows the changing trend of the Brenner gradient value of the image during boundary detection. The sub-figures represent the image clarity at the beginning and end of boundary detection. It can be clearly seen that as the sample arm moves, the image of the HepaRG cell 3D culture model gradually blurs. The sliding window algorithm detects that the Brenner gradient value no longer decreases significantly, indicating that the model boundary has been reached. The red dotted line indicates that the boundary has been detected and stopped.

[0077] This example further evaluates the efficiency of the non-equidistant acquisition strategy compared to the traditional equidistant sampling. The non-equidistant acquisition strategy dynamically adjusts the sample arm step distance according to the complexity of the sample structure. In the Non-ROI area with less structural changes, a step distance of 10μm is used to reduce unnecessary data collection, while a step distance of 2μm is used in the ROI area with complex structure to maintain high resolution. Figure 6 Middle (e), Figure 6 (f) shows the Brenner gradient value curves at different depths of the same HepaRG cell 3D culture model using traditional isometric sampling and non-isometric acquisition. The red dotted lines represent the boundary and acquisition end position of the bottom cluster structure and the top sparse structure area, respectively. It can be seen that in the cell sparse structure area, the image data collected using the non-isometric acquisition protocol is greatly reduced. Figure 6 The sub-image in (e) is the FF-OCT image of the sparse structure of HepaRG cells attached to the upper part of the microwell in the three-dimensional culture model. The green arrow indicates the sparsely dispersed cells at this depth; Figure 6 The sub-image in (g) is an FF-OCT image of a cell cluster at a certain depth in the HepaRG cell 3D culture model, which clearly shows the more complex cell structure at the bottom. The cluster structure at the bottom has a high cell density and more information exchange between cells, which can better reflect the biological characteristics of the 3D model.

[0078] In addition, this embodiment also provides an OCT automated imaging system, including: The sample interference surface positioning module is responsible for positioning the sample interference surface to achieve equal optical path lengths between the sample arm and the reference arm; A sample boundary positioning module is responsible for positioning the focal plane layer at the sample boundary; The image acquisition module is responsible for acquiring non-equidistant three-dimensional OCT images of the sample.

[0079] In summary, the OCT automated imaging device of the present invention integrates lighting, imaging and control modules, and the internal components of each module work together. The lighting module provides a suitable light source and splits the beam through precise optical path control; the sample arm and reference arm modules realize optical signal transmission and phase control through objective lenses, displacement stages and phase shifters; the detection arm module accurately receives interference signals for imaging; the control module accurately controls the actions of each component, such as controlling the coordinated work of the first displacement stage, the second displacement stage, the phase shifter and the camera, to realize automated operations such as image acquisition, phase shift, and sample position adjustment, thereby improving the overall stability and reliability of the imaging system and ensuring efficient and accurate imaging.

[0080] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. An automated imaging method, characterized in that: The method uses non-isometric three-dimensional OCT image acquisition; The non-equidistant three-dimensional OCT image acquisition is to reconstruct the OCT image from the focal surface layer images under different phase shift amounts; According to the grayscale variance of the OCT image, the region of interest and the region of no interest of the current sample focal plane are screened out, and the sample arm is moved to the next focal plane layer with a small step size in the region of interest, and the sample arm is moved to the next focal plane layer with a large step size in the region of no interest.

2. The method according to claim 1, characterized in that: In the non-equidistant three-dimensional OCT image acquisition, the comparison between the number of OCT images and the sliding window size is used to identify whether the image is still in the current focal plane layer. If the number of OCT images is greater than the sliding window size, the sample arm is moved to the next focal plane layer with a large step length.

3. The method according to claim 1, characterized in that: Before acquiring non-equidistant three-dimensional OCT images, it is necessary to locate the sample interference surface and position the focal plane layer at the sample boundary.

4. The method according to claim 3, characterized in that: The positioning sample interference surface comprises: The OCT image is reconstructed by the focal surface layer image under different phase shift amounts; the first sliding window change rate is calculated according to the grayscale variance of the OCT image; if the first sliding window change rate is greater than a preset threshold value ξ1, it is considered that interference occurs in the current sample focal plane; if the first sliding window change rate is less than or equal to the preset threshold value ξ1, the reference arm is moved in the direction of increasing optical path.

5. The method according to claim 4, characterized in that: In the positioning sample interference surface, based on the comparison between the number of OCT images and the sliding window size, it is identified whether it is still in the current focal plane layer. If the number of OCT images is less than or equal to the sliding window size, the reference arm is moved in the direction of increasing optical path.

6. The method according to claim 3, characterized in that: The positioning focal plane layer at the sample boundary includes: The second sliding window change rate is calculated according to the image clarity of the focal surface layer image; If the change rate of the second sliding window is less than the preset threshold ξ3, the current focal plane layer is considered to be the sample boundary. If the change rate of the second sliding window is greater than or equal to the preset threshold ξ3, the sample arm is moved to the next focal plane layer.

7. The method according to claim 6, characterized in that: The positioning focal plane layer is in the sample boundary, and whether it is still in the current focal plane layer is identified based on the comparison between the number of images in the current focal plane layer and the preset threshold ξ2. If the number of images in the current focal plane layer is less than or equal to the preset threshold ξ2, the sample arm is moved to the next focal plane layer.

8. The method according to claim 5, characterized in that: The direction in which the sample arm moves during the non-equidistant three-dimensional OCT image acquisition process is opposite to the direction in which the sample arm moves during the process of positioning the focal plane layer at the sample boundary.

9. An automated imaging system for implementing the method according to any one of claims 1 to 8, characterized in that include: The sample interference surface positioning module is responsible for positioning the sample interference surface to achieve equal optical path lengths between the sample arm and the reference arm; A sample boundary positioning module is responsible for positioning the focal plane layer at the sample boundary; The image acquisition module is responsible for acquiring non-equidistant three-dimensional OCT images of the sample.

10. An automated imaging device for implementing the method according to any one of claims 1 to 8, comprising an illumination module, an imaging module, and a control module, wherein the imaging module comprises a reference arm module, a sample arm module, and a detection arm module; The illumination module provides a low-coherence light source, and then sends the light to the reference arm module and the sample arm module respectively; the reflected light returned by the sample arm module interferes with the reflected light returned by the reference arm module and enters the detection arm module; the detection arm module obtains an interference signal; It is characterized in that The lighting module comprises a low-coherence light source, a first convex lens, an aperture diaphragm, a variable field diaphragm, a second convex lens, and a beam splitting cubic prism; light emitted by the low-coherence light source is focused by the first convex lens and imaged at the aperture diaphragm, and the image point of the light source passes through the variable field diaphragm and the rear second convex lens and then enters the beam splitting cubic prism for beam splitting; The sample arm module comprises a first objective lens and a first translation stage; a culture dish for placing the sample to be tested is arranged on the first translation stage, and the objective lens is located above the culture dish on the first translation stage; The reference arm module includes a filter, a second objective lens, a reference mirror, a phase shifter, and a second displacement stage connected in sequence through an optical path; the phase shifter is used to perform phase shifting to achieve the acquisition of current focal plane layer images with different phase shift amounts; The second translation stage is used to move the reference arm; The detection arm module includes a third convex lens and a camera; the interference light passes through the third convex lens to form a clear image on the camera; the first translation stage is used to move the sample arm; The control module is used to control the first translation stage, the second translation stage, the phase shifter, and the camera.

Images