High-flux multi-view automatic positioning light sheet microscopic imaging method and system

Abstract

The invention provides a high-throughput multi-view automatic positioning light sheet microscopic imaging method. The method comprises the following steps: automatically identifying the size of a sample, presetting parameters, and carrying out preliminary scanning; splicing imaging data in real time; globally and automatically positioning a sample area; planning a shooting path; and multi-target area high-throughput shooting parameters are optimized. A microscope large-view high-flux rapid imaging method based on a real-time image automatic splicing and positioning technology is combined with a real-time splicing function of shot imaging data and an automatic positioning function of a sample position in an image to rapidly search specific position area information of a sample; and shooting imaging and data storage of the experimental sample are realized by automatically optimizing shooting configuration parameters. The invention further provides a high-throughput multi-view automatic positioning light sheet microscopic imaging system adopting the method, and shooting imaging and data processing can be rapidly and accurately carried out on a target large sample.

Description

Technical Field

[0001] The present application relates to the field of optical microscopic three-dimensional imaging, and in particular to a high-throughput multi-field-of-view automatic positioning light sheet microscopic imaging method and system. Background Art

[0002] Light Sheet Fluorescence Microscopy (LSFM) is an advanced microscopic imaging technique that obtains high-resolution images by exposing the sample to a cross-sectional light sheet instead of traditional point-to-point scanning. It is particularly suitable for three-dimensional imaging of biological samples.

[0003] In the related art, light-sheet fluorescence microscopy typically uses a tiled area scanning method to image large samples, with customizable imaging area configuration parameters. However, the irregularities of the sample being imaged and the inability to pre-locate the sample's actual location can result in a large amount of imaging data from non-sample areas during the image capture process, wasting storage space and consuming the capture process. Furthermore, capturing large sample data is inherently time-consuming, and coupled with the effects of phototoxicity during the capture process, this can easily cause the fluorescence signal from the sample data to gradually weaken, compromising the validity of the final captured sample data.

[0004] The above problems need to be solved urgently. Summary of the Invention

[0005] In order to solve the above problems, it is necessary to propose a high-throughput multi-field automatic positioning light sheet microscopy method and system to improve the sample imaging performance in light sheet fluorescence microscopy.

[0006] In order to achieve the above objectives, this application adopts the following technical solutions:

[0007] In one aspect, a high-throughput multi-field light sheet microscopy method with automatic positioning is provided, comprising the following steps:

[0008] Automatically identify sample size and preset parameters for preliminary scanning;

[0009] Real-time stitching of imaging data;

[0010] Global automatic positioning of sample area;

[0011] Plan the shooting route;

[0012] Optimize high-throughput imaging parameters for multiple target areas.

[0013] On the other hand, the present application provides a high-throughput multi-field-of-view automatic positioning light sheet microscopy imaging system, which adopts the high-throughput multi-field-of-view automatic positioning light sheet microscopy imaging method as described above.

[0014] This application is based on a microscope large-field high-throughput rapid imaging method based on real-time image automatic stitching and positioning technology. It combines the real-time stitching function of the captured imaging data and the automatic positioning function of the sample position in the image to quickly search for the specific location area information of the sample, and realizes the shooting imaging and data storage of the experimental sample by automatically optimizing the shooting configuration parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a schematic diagram of an exemplary sample size identification device and identification principle of the present application;

[0016] Figure 2 This is a schematic diagram of an exemplary imaging control system of the present application for automatically identifying and setting sample size parameters;

[0017] Figure 3 Schematic diagram of an exemplary high-throughput observation object area of ​​the present application;

[0018] Figure 4 This is a simple schematic diagram of an exemplary target high-throughput relationship of the present application;

[0019] Figure 5 This is a schematic diagram of an exemplary calculation process of traversing all regional points in this application;

[0020] Figure 6 Schematic diagram of shooting positions of different starting Tile areas of the present application;

[0021] Figure 7 This is an exemplary schematic diagram of the path planning of different Tile areas in this application;

[0022] Figure 8 is a schematic diagram of an exemplary single target high-throughput region of the present application;

[0023] Figure 9 This is a schematic diagram of an exemplary shooting process of the present application;

[0024] Figure 10 This is an exemplary planar sample positioning outline diagram of the present application. DETAILED DESCRIPTION

[0025] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0026] The terms used in this application are for the purpose of describing particular embodiments only and are not intended to limit this application. As used in this application and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

[0027] It should be understood that the words “first”, “second” and similar terms used in the specification and claims of this application do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as “a” or “an” do not indicate a quantity limitation, but rather indicate the presence of at least one; “plurality” indicates a quantity of two or more. Unless otherwise indicated, words such as “front”, “rear”, “lower” and / or “upper” are for ease of description only and are not limited to one position or one spatial orientation. Words such as “include” or “comprising” and similar terms mean that the elements or objects appearing before “include” or “comprising” cover the elements or objects listed after “include” or “comprising” and their equivalents, and do not exclude other elements or objects.

[0028] The following describes in detail a high-throughput multi-field automatic positioning light sheet microscopy method and system according to an exemplary embodiment of the present application in conjunction with the accompanying drawings. The features of the following embodiments and implementations may complement or be combined with each other unless they conflict.

[0029] With the continuous advancement and development of science and technology, research in the biomedical field has become increasingly in-depth, and the exploration of microstructures and cellular activities has become a focus of attention. For example, with the development of brain science and technology, the demand for imaging neuronal clusters and even whole-brain slices has increased significantly. The large number of neurons in the brain places higher demands on the imaging field of view and resolution. As super-resolution technology develops towards higher throughput and a larger field of view, researchers are pursuing high-throughput, high-resolution, and large-field imaging, making high-throughput microscopic imaging a hot area. Fluorescence microscopy uses ultraviolet light as a light source to illuminate the object under examination, causing it to fluoresce. The object's shape and location can then be observed under the microscope. Fluorescence microscopy is generally used to study the absorption, transport, distribution, and localization of substances within cells. Some substances in cells, such as chlorophyll, fluoresce when exposed to ultraviolet light. Other substances, while not inherently fluorescent, can also fluoresce when stained with fluorescent dyes or fluorescent antibodies and exposed to ultraviolet light. Fluorescence microscopy is one of the tools used to qualitatively and quantitatively study these substances.

[0030] Light sheet fluorescence microscopy uses light sheet fluorescence microscopy (LSFM), which uses light sheet illumination / orthogonal detection followed by fluorescence imaging. Specifically, LSFM uses a thin light sheet (usually a laser) to illuminate the sample vertically, illuminating only a tiny plane, which can greatly reduce photodamage and photobleaching of the sample. Orthogonal detection means that the detection lens it uses is usually at a 90-degree angle to the light sheet, which can achieve high resolution and fast imaging at the same time. Fluorescence imaging means that the sample is usually labeled with a fluorescent dye or fluorescent protein. When the light sheet passes through the sample, only the part that intersects with the light sheet will emit a fluorescent signal, which is captured by the detection lens.

[0031] Light-sheet fluorescence microscopy, a new fluorescence microscopy technique, can improve the resolution of imaging cell structure and tissue function in biomedical research. In fields such as cell biology, neuroscience, and developmental biology, LSFM can be used to observe the three-dimensional structure and dynamic processes of living cells, tissues, and even entire organisms. Due to its fast imaging speed and minimal sample damage, LSFM is suitable for rapid imaging of large-scale samples, such as tissue sections and small animal models. In tissue engineering, LSFM can be used to observe the growth, movement, and interactions of cells within three-dimensional structures. Furthermore, LSFM can be used in drug screening and toxicology studies to observe the effects of drugs on cells or tissues.

[0032] In summary, this technology has become an indispensable research tool in this field due to its low phototoxicity and photobleaching, high-speed, high-resolution three-dimensional imaging, reduced sample damage, and suitability for in vivo imaging. However, how to quickly and accurately capture and image large target samples and process data is widely needed in the biomedical field.

[0033] Light-sheet fluorescence microscopy, a new fluorescence microscopy technique, significantly improves the resolution of cell structure and tissue function in biomedical research and has become an indispensable research tool in this field. However, due to hardware limitations, a single field of view cannot capture a wide range of images, and currently, most images are taken for a single or small number of target samples. High-throughput targets often require a large field of view, which, in the case of high-resolution images, requires more time and space.

[0034] The light sheet fluorescence microscopy technology involved in this application is a typical optical microscopic three-dimensional imaging technology, which uses a light sheet confined near the focal plane to scan the sample layer by layer to obtain a three-dimensional fluorescence image of the sample. The light sheet used in the light sheet fluorescence microscope can be a flat static light sheet covering the entire imaging field of view generated by a laser beam passing through a cylindrical lens, or it can be a dynamic virtual light sheet formed in the imaging field of view area by a laser beam passing through optical elements such as a scanning galvanometer. In the related art, the sample shooting area is usually manually customized, and the sample area is shot, scanned, imaged, and data stored one by one in a tiled manner. During the shooting and scanning process of each area, the current shooting area is judged to be a non-empty sample area to drive the microscope to further adjust the parameters (adjust the actual sample shooting depth) to re-shoot and obtain the image. However, shooting, imaging, and image storage of data from non-sample areas during the shooting process will cause time-consuming data shooting and waste of storage space, reducing the efficiency of image data acquisition. Traditional image processing methods are used to identify and determine non-empty sample areas, which are inefficient, and due to the diversity of biomedical images, low processing accuracy can easily cause some areas of the sample to be missing.

[0035] In summary, the present applicants (they) believe that at this stage, fluorescent light sheet microscopes usually adopt a tiled area scanning shooting method in the large-scale shooting process, and perform high-resolution imaging of samples by customizing the shooting area configuration parameters. The above shooting process has the following problems: 1) Due to the irregularity of the sample and the inability to pre-position the actual position information of the sample, the imaging data of the corresponding non-sample area exists in the shooting process, which wastes storage space and the shooting process is time-consuming. 2) The shooting of the entire data itself is time-consuming, and coupled with the influence of phototoxicity during the shooting process, it is easy to cause quenching of the sample data fluorescence signal, affecting the validity of the final sample shooting data. 3) In high-throughput data shooting, the observed objects may be distributed in hundreds of different areas. In view of this, the present applicants (they) proposed a set of shooting ideas, which is simply "rough scanning-positioning-planning-fine scanning": first use a lower resolution and a larger field of view for rapid rough scanning, automatically locate and identify the location and outline of the sample, use the designed algorithm to plan the optimal fine scanning path, and then switch back to the high-resolution state for the final fine scanning shooting. For the rough scanning process, an imaging control system was designed that automatically identifies the sample size and general outline and uses the system lens parameters to automatically plan the imaging system parameters. This allows for rapid scanning of the entire preset range using lower resolution and larger axial step sizes, and the captured images are stitched together in real time. A fast imaging method was then proposed that automatically locates the object for high-throughput observation based on the stitched large-field-of-view images. This method plans the globally optimal imaging path to reduce the time required for fine scanning, while also reducing the total travel of the expensive high-precision translation stage to minimize wear and tear, achieving low-loss, high-precision imaging and data storage of experimental samples.

[0036] The present application provides a high-throughput multi-field automatic positioning light sheet microscopy method and system, which can realize large-field high-throughput rapid imaging of microscopes based on real-time image automatic stitching and positioning technology. The method combines the real-time stitching function of the captured imaging data and the automatic positioning function of the sample position in the image to quickly search for the specific location area information of the sample, and realizes the imaging and data storage of the experimental sample by automatically optimizing the shooting configuration parameters. The imaging method provided by the present application uses the cooperation of software and hardware to achieve a theoretically unlimited shooting field of view, faster shooting efficiency and less storage space, so as to achieve the imaging capability of the current hot fields using current equipment seamlessly. The following will be explained in detail with examples.

[0037] The high-throughput multi-field automatic positioning light sheet microscopy imaging method comprises the following steps:

[0038] Automatically identify sample size and preset parameters for preliminary scanning;

[0039] Real-time stitching of imaging data;

[0040] Global automatic positioning of sample area;

[0041] Plan the shooting route;

[0042] Optimize high-throughput imaging parameters for multiple target areas.

[0043] Optionally, the "rough scan" stage includes automatically identifying the sample size and presetting parameters for preliminary scanning. Optionally, the optimal rough scan shooting parameters are calculated based on the camera lens parameters, set, and performed. The preset parameters include the width and height of a single shooting area, the percentage of overlapping areas in the X-axis direction, the percentage of overlapping areas in the Y-axis direction, the number of channels, the laser wavelength, the laser intensity, the Z-axis depth, the Z-axis step size, the physical value of the imaging pixel, the number of shooting areas in the X-axis direction, the number of shooting areas in the Y-axis direction, the number of shooting time nodes, and the spacing between shooting time nodes.

[0044] Specifically, use Figure 1 The sample identification device and identification principle diagram shown below captures the entire field of view of the sample and calculates its size. The offset between the stage and the sample container is calculated based on the structure of the entire microscope system, allowing for a reasonable starting position for imaging. Generally, this calculation can be performed using the following method.

[0045]

[0046] Optionally, the imaging control system automatically identifies and sets sample size parameters. Based on the sample container size and the imaging field size at different lens magnifications, the number of rows and columns of the imaging tile area is calculated:

[0047]

[0048] Optionally, an optimal axial step length can be preset by the sample container depth and lens magnification parameters, so as to capture as many layers as possible to show more details while ensuring the imaging efficiency. Optionally, according to the row and column tiling obtained in the above steps, a reasonable serpentine movement can be planned (see Figure 7 Description part), get the number and field of view of each site.

[0049] In some possible implementations, the initial scan involves scanning the current depth of the current area using a custom-planned imaging path along the XY plane; then moving the sample along the Z-axis in steps to scan the next depth layer, completing each layer of scanning until continuous imaging along the depth direction is complete and a 3D image data stack is formed. The sample then moves to the next imaging area and scans the data layer by layer. The XY plane, defined as the plane encompassing the length of the scanning area along the X-axis and the length of the scanning area along the Y-axis, accurately captures the width and height of the actual sample location.

[0050] Optionally, the preliminary scan includes customizing the shooting path of the light sheet along the XY plane, dividing the lens imaging area into several rows and columns, and scanning and shooting a set depth range for each imaging area: moving the sample along the depth direction according to a larger Z-axis step to perform layer-by-layer scanning. After the layer-by-layer scanning is completed, a three-dimensional image data Stack will be formed, and then moving to the next imaging area to perform layer-by-layer depth data scanning again.

[0051] In some possible embodiments, the moving step length of the Z-axis sample shooting is increased. Such a setting can ensure that the complete information of the sample is scanned in the Z-axis, reduce the amount of shooting data, and improve the efficiency of the initial scanning. In some possible embodiments, the real-time stitching of imaging data includes using the maximum intensity projection method to perform planar projection of the three-dimensional image data stack of the current area in the three axes of X / Y / Z, respectively, to obtain XY plane projection map, YZ plane projection map, and XZ plane projection map, and combining an image alignment method based on preset parameters and the corresponding projection map data information obtained to calculate and align the coordinate information of the image data of each shooting area in the global shooting image data, perform real-time stitching of the corresponding projection maps obtained in each axis, scan to obtain the three-dimensional image data stack of the next shooting area, and repeat the above stitching imaging operation. Among them, the image alignment method based on preset parameters and the corresponding projection map data information obtained includes a method based on converting the position information of the physical shooting area into pixel information of the image coordinate system, a method based on normalized cross-correlation matching, a phase correlation matching method, etc., and the method is not limited.

[0052] Specifically, the Maximum Intensity Projection (MIP) method is a medical image processing technique that processes volumetric data and projects the pixels with the highest pixel intensity to form a two-dimensional image. MIP images can display the entire three-dimensional structure, helping researchers obtain data more accurately.

[0053] Optionally, the operator can perceive the gradual tiling process and the final overall view of the observation area in real time during the long shooting process. Optionally, the real-time stitching of imaging data includes the following steps:

[0054] According to the above preset shooting parameters, the physical coordinate information of each shooting area position can be obtained, and the stitching coordinate information of the images shot in each area in the final global field of view image can be calculated; the camera continuously shoots in the Z direction of the current area position to form a three-dimensional image stack data, and the maximum intensity projection method MIP is used to calculate the projection image of the three-dimensional image stack data in the X / Y / Z axes; the projection image of each axis is downsampled to reduce the size of the single projection image to reduce the amount of stitching parameter calculation, and real-time image stitching is achieved in each axis.

[0055] Optionally, when obtaining the XY plane projection Proj xy 、YZ plane projection diagram Proj yz and XZ plane projection Proj xz Afterwards, the projection images of the imaged areas in each axis are stitched together in real time. For areas where multiple projection images overlap, this application uses the maximum intensity projection method to recalculate the pixel values ​​at each location in the overlapping area. The scan continues to obtain the 3D image data stack of the next area, and the above operation is repeated until the scanning of each area is completed, and the complete projection image stitching image along each axis is obtained.

[0056] Optionally, the automatic positioning of high-throughput multi-observation objects includes automatically positioning high-throughput observation objects based on the global stitched projection images of the X / Y / Z axes obtained in the above implementation, and the specific steps include the following:

[0057] The maximum inter-class variance method is used to automatically calculate the foreground / background threshold segmentation parameters of the projection image and obtain a global binary image based on the global pixel value and pixel number distribution information. The binary image is searched for a continuous pixel point set to detect the target outer contour information and summarize the contour endpoints of each group of outer contours. The corresponding contour endpoint sets of each high-throughput observation object are traversed, and the maximum and minimum coordinates in the X / Y axis are calculated according to the coordinate values ​​of the pixel points corresponding to each endpoint in the set, and the position information of each outer contour enclosing rectangle is obtained, finally obtaining the coordinate position of each high-throughput observation object in the image.

[0058] Specifically, the global automatic positioning of the sample area includes stitching images based on the projection images of the X / Y / Z three axes, and automatically positioning the sample area position for each axial stitching image. In some possible implementations, the global automatic positioning of the sample area includes using the maximum inter-class variance method to calculate the image foreground and background threshold segmentation parameter threshold, obtaining a binary image of each axial stitching image, performing an edge detection operation on the binary image of each axial stitching image to obtain edge points, summarizing the edge points into groups of contour points according to set rules, traversing the coordinate values ​​of the pixel points corresponding to each contour point, and respectively calculating the maximum and minimum coordinates (xmin, ymin, xmax, ymax) on the X / Y axis of the image to obtain the position information of the sample in the image data, wherein the contour is screened (according to the rules set by area and length) and a convex polygon is fitted during the summarization process.

[0059] The maximum inter-class variance method is also known as the Otsu method. This method is an algorithm that automatically determines the threshold for image binarization segmentation. It divides the image into two parts, background and target, based on the grayscale characteristics of the image. By finding an optimal threshold, the inter-class variance between the two parts is maximized, thereby achieving the goal of minimizing the probability of misclassification. This method is simple and fast to calculate and is not affected by image brightness and contrast. Specifically, the maximum inter-class variance method traverses all possible thresholds, calculates the intra-class variance that divides the image into foreground and background according to the threshold, and determines the optimal threshold based on the minimum value of the intra-class variance. This threshold maximizes the difference between the foreground and background areas, thereby achieving the best image segmentation effect. Edge detection is performed on the binary images of each axial stitching image to obtain edge points, and the edge points are summarized into groups of contour points according to set rules. During the summary process, the contours are screened (rules can be set based on area, length, etc.) and convex polygons are fitted. The calculated contour point data of each group are traversed to the coordinate values ​​of the pixel points corresponding to each contour point, and the maximum and minimum coordinates (xmin, ymin, xmax, ymax) in the X / Y axis of the image are calculated respectively to obtain the position information of the sample in the image data.

[0060] Optionally, the planned shooting path is intended to reduce shooting time and the travel and wear of the expensive high-precision translation stage, including the following steps: according to the above results, calculate the image position coordinate information corresponding to each high-throughput observation object, record the region ROI of each high-throughput observation object i The corresponding image position coordinates {x i ,y i , z i}, where x i ,y i , z iRespectively represent the X / Y / Z axis coordinate values ​​in the image coordinate system; the distance between the target high-throughput observation object areas ROI in the image coordinate system is calculated according to the following formula: Among them, d (i，j) Represents ROI i and ROI j The image coordinate distance between them is traversed through each observation object area, and finally the planning path with the shortest path distance is selected as the planning path for the instrument shooting.

[0061] Specifically, such as Figure 2 A schematic diagram of an exemplary high-throughput observation area is shown. As shown, A, B, C, and D represent four observation areas. The left diagram shows a schematic diagram of each area under the full-field scanning method, with the background squares representing individual scan areas. The right diagram shows a schematic diagram of the distances between each area.

[0062] Taking the imaging of four high-throughput observation regions A, B, C, and D as an example, d(·) represents the distance between two regions in the image coordinate system. The search for the optimal path planning (minimizing the total distance) can be abstracted into a Hamiltonian path planning problem: each observation region is divided into a set of "visited" regions and a set of "unvisited" regions. Starting from the leftmost observation region, the backtracking method recursively searches all "unvisited" regions, marks them as "visited" regions, and searches for the next region. This search method traverses each observation region, and finally selects the planned path with the shortest path distance as the planned path for the instrument imaging.

[0063] Specifically, the two scanning methods of "global optimal path scanning" and "full field of view line-by-line scanning" mentioned in this application are illustrated as follows. Figure 2 For example, assuming a global capture area with a width and height of 7*7, and a single capture area with a width and height of 100um*100um, the two displacement capture methods are compared as follows:

[0064] (1) “Full field of view line by line scanning” method: Figure 2 Taking the left picture as an example, the scanning is performed line by line from top to bottom, and the displacement between the upper and lower lines is performed in a "snake-like" manner. The total shooting displacement path is 4800um.

[0065] (2) “Global Optimal Path Scanning” Method: Figure 2 Taking the image on the right as an example, the distances between each high-throughput observation object are calculated, resulting in a globally optimal imaging path: C->D->B->A, with a total imaging displacement of 2300 μm. This calculation shows that the proposed global optimal path scanning method reduces the imaging path displacement by 52.1% compared to the full-field line-by-line scanning method, effectively improving imaging efficiency, shortening the imaging distance, and reducing imaging hardware wear.

[0066] Specifically, optimizing the high-throughput imaging parameters for multiple target regions includes calculating image coordinate position information corresponding to each target high-throughput region, calculating the distance between each target high-throughput region ROI in the image coordinate system according to a formula, and calculating an optimal path for continuous imaging between each target high-throughput region. Optionally, the optimal path for continuous imaging between each target high-throughput region is calculated using a Hamiltonian path algorithm.

[0067] In some possible implementations, a depth-first traversal method is used to calculate and traverse all region points, visit the starting region point, and mark it as "visited". A deep visit is performed on the first "unvisited" region of the region point, and after each visit, the adjacent points are marked as "visited" to avoid repeated visits. When there is no "unvisited" region, the previous "unvisited" region is backtracked and a deep visit is performed again until all region points are traversed. Specifically, the depth-first traversal method can be used to calculate and solve the Hamiltonian path, including the following steps:

[0068] ① Set the leftmost area as the starting point and mark it as "visited";

[0069] ② Using the currently visited point, select one of the remaining unvisited areas for the next visit and add the path length between the two points to the total path length;

[0070] ③Mark the newly visited area points in ② as the current "visited" status (to avoid repeated visits);

[0071] ④ Repeat steps ②③ until there are no "unvisited" areas, and get the total path length under the current access order;

[0072] ⑤Backtrack to the previously "unvisited" area and select other remaining unvisited points and repeat ②③④ to obtain the total path length in different visit orders;

[0073] ⑥ Find the access order with the optimal total path length.

[0074] Among them, the depth-first traversal method includes commonly used search path optimization and pruning optimization.

[0075] like Figure 4 The following is a simple diagram of an exemplary target high flux relationship. As shown in the figure, the image coordinate position information corresponding to each target high flux is calculated, and the image position coordinates corresponding to each target high flux area are recorded as {x i ,y i , z i}, where x i / y i / z i Respectively represent the upward coordinate values ​​of the X / Y / Z axes in the image coordinate system.

[0076] The distance between each target high-throughput region ROI in the image coordinate system is calculated according to the formula:

[0077]

[0078] Among them, d (i，j) Represents ROI i and ROI j The image coordinate distance between them.

[0079] Specifically, such as Figure 4 As shown in the figure, taking the four target high-flux areas A, B, C, and D as an example, d(·) represents the distance between two areas in the image coordinate system. The depth-first traversal method is used to calculate and traverse all area points and is implemented recursively, including the following steps:

[0080] ① Set the leftmost area as the starting point and mark it as "visited"

[0081] ②Using the current "visited" point, select one of the remaining "unvisited" areas for the next visit and add the path length between the two points to the total path length

[0082] ③Mark the newly visited area points in ② as the current "visited" status (to avoid repeated visits)

[0083] ④ Repeat steps ②③ until there are no “unvisited” areas, and get the total path length under the current access order.

[0084] ⑤ Backtrack to the previously "unvisited" area and select other remaining unvisited points and repeat ②③④ to obtain the total path length in different visit orders

[0085] ⑥ Find the access order with the optimal total path length.

[0086] Optional, such as Figure 5 As shown, taking area A as the traversal starting point as an example, the following sub-steps are included:

[0087] like Figure 5 As shown in (a), we start from vertex A and do a depth-first traversal. Black indicates visited areas, and gray indicates unvisited areas. We traverse the neighboring points of area A. Its neighboring points are C, D, and B respectively. In the next step, we recursively visit area C first.

[0088] like Figure 5 As shown in (b), the neighboring points of area C are traversed. Its neighboring points are A, B, and D respectively. Area A is in the "visited" state, so the next step is to recursively visit area B;

[0089] like Figure 5As shown in (c), the neighboring points of area B are traversed, and their neighboring points are A, C, and D respectively. Among them, area A and area C are in the "visited" state, so the next step is to recursively visit area D;

[0090] like Figure 5 As shown in (d), after visiting area D, the visit to all areas is completed, and the order of visiting each area is recorded;

[0091] Using the leftmost point as the starting area point, repeat the above steps and record all path sequence information. Finally, calculate the area access path with the minimum distance, which is the optimal area access path. In summary, the optimal path for continuous shooting between each target high-throughput area can be calculated.

[0092] Optionally, the above steps are implemented to mainly optimize the path planning between different target high-throughput areas. In some possible implementations, the shooting path within a single target high-throughput area may be further optimized.

[0093] Specifically, such as Figure 6 Shown is a schematic diagram of the shooting position of different starting Tile areas, such as Figure 7 The figure shows a schematic diagram of path planning for different tile areas. As shown in the figure, for a single target high-throughput area, four different starting tile area shooting positions are provided: top left, top right, bottom left, and bottom right; four different tile area path planning directions are provided: snake by rows, straight by rows, snake by columns, and straight by cols. In other words, for a single target area, it still contains one or more lens imaging areas, divided into several rows or columns of lens imaging areas. Different starting shooting positions include top left, top right, bottom left, and bottom right, and different path planning directions include snake by row, straight by row, snake by column, and straight by column. As shown in the figure, the starting tile area shooting position and the ending tile area shooting position can be customized.

[0094] like Figure 8Shown is a schematic diagram of a single target high-throughput area sample. Due to the irregular morphology of biological samples, there are still non-sample areas in the acquired single target area (rectangular area). Based on the foreground binarization map of each target high-throughput area image obtained in the above implementation, the present application proposes to determine whether the current Tile area is a sample area by calculating the intersection area between each Tile area and the corresponding foreground binarization map before shooting a single target high-throughput area, and skip the non-sample area during the shooting process. As shown in the figure, taking the 3 rows and 3 columns area as an example, the gray gradient area represents the sample area, where the Tile area (0, 2) does not have an intersection or overlap with the sample area, and the sample shooting is skipped during the actual shooting process, so as to further reduce the sample shooting time and improve the efficiency of storage space.

[0095] Based on the above steps, the automatic parameters of this application include the following:

[0096] (1) The sequence of shooting paths for each target high-throughput area;

[0097] (2) The starting shooting position of a single target high-throughput area, the shooting path between tile areas, the number of tile areas shot in the X / Y axis, and the area overlap percentage;

[0098] Based on the multi-target area high-throughput shooting parameters optimized in the above steps, the shooting and data storage of experimental samples are realized.

[0099] In some possible implementations, for a single target area, by calculating the intersection area between each lens imaging area and the corresponding target foreground binary image, it is determined whether the current imaging area can capture the target to set whether to skip the shooting step.

[0100] like Figure 9 The figure shows an exemplary shooting process diagram of the present application, including configuration parameters on the left and right sides of the software interface, and the middle view showing the image imaging effect during the shooting process.

[0101] Figure 10 This is an exemplary plane sample positioning outline diagram of the present application, which represents the XY / YZ / XZ plane sample stitching image and the field of view position positioning effect diagram from left to right.

[0102] The present application provides a system for a high-throughput multi-field automatic positioning light sheet microscopy imaging method using the above-mentioned method, which can quickly and accurately shoot, image and process data of large target samples. It can improve the resolution imaging capability of cell structure and tissue function in biomedical research, and can be widely used in cell biology, neuroscience, developmental biology and other fields. The imaging control system designed in the present application can automatically identify the sample size, automatically preset the shooting parameters according to the lens parameters for rough scanning, save manual operation steps and improve operation efficiency. The real-time image stitching technology of the present application can process the image data shot in the current area in real time, and combine the maximum intensity projection MIP and downsampling operation technology to achieve efficient real-time image stitching, so that users can perceive the full picture of high-throughput observation objects in real time. The high-throughput multi-observation object automatic positioning technology of the present application searches for observation objects in the global stitching projection map of each axis, automatically locates the position coordinate information of each observation object and visualizes it, so that users can intuitively feel the position distribution of high-throughput observation objects in the sample. The global optimal path planning of the present application can search and plan the global optimal path of the fine scanning stage based on the position information of the high-throughput observation object, so that the lens displacement distance during the shooting process is the shortest, reducing the shooting time and the travel and wear of the expensive high-precision translation stage.

[0103] In summary, the high-throughput, large-field-of-view, rapid imaging method and control system proposed in this application automatically identifies the sample size and sets the optimal shooting parameters in combination with the lens parameters for rough scanning. Combined with real-time image stitching technology and high-throughput automatic positioning of the contours of multiple observation objects, users can perceive the overall picture of the observation object in real time; and by optimizing the moving path of the fine-scanning translation stage to reduce instrument and equipment loss and improve the shooting and scanning speed, high-throughput, large-sample data can be quickly imaged and saved.

[0104] The above description is merely a preferred embodiment of the present application and does not constitute any form of limitation to the present application. Although the present application has been disclosed as above with preferred embodiments, it is not intended to limit the present application. Any technician familiar with this profession can make slight changes or modifications to equivalent embodiments with equivalent changes using the technical contents disclosed above without departing from the scope of the technical solution of the present application. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present application without departing from the content of the technical solution of the present application are still within the scope of the technical solution of the present application.

Claims

1. A high-throughput multi-field automatic positioning light sheet microscopy imaging method, characterized in that: The steps include: Automatically identify sample size and preset parameters for preliminary scanning; Real-time stitching of imaging data; Global automatic positioning of sample area; Plan the shooting route; Optimize high-throughput imaging parameters for multiple target areas.

2. The high-throughput multi-field automatic positioning light sheet microscopy imaging method according to claim 1, characterized in that: Automatically identify the sample size and set the best preset shooting parameters for preliminary scanning in combination with the lens parameters. The preset parameters include the width and height of a single shooting area, the percentage of overlapping areas in the X-axis direction, the percentage of overlapping areas in the Y-axis direction, the number of channels, the laser wavelength, the laser intensity, the Z-axis depth, the Z-axis step size, the physical value of the imaging pixel, the number of shooting areas in the X-axis direction, the number of shooting areas in the Y-axis direction, the number of shooting time nodes, and the shooting time node spacing.

3. The high-throughput multi-field automatic positioning light sheet microscopy imaging method according to claim 2, characterized in that: The preliminary scan includes customizing the shooting path of the light sheet along the XY plane, dividing the lens imaging area into several rows and columns, and scanning and shooting a set depth range for each imaging area: moving the sample along the depth direction according to a larger Z-axis step to scan layer by layer. After the layer-by-layer scanning is completed, a three-dimensional image data Stack will be formed, and then moving to the next imaging area to perform layer-by-layer depth data scanning again.

4. The high-throughput multi-field automatic positioning light sheet microscopy imaging method according to claim 3, characterized in that: The real-time stitching of imaging data includes using the maximum intensity projection method to perform plane projection on the three-dimensional image data Stack of the current area in the three axes of X / Y / Z, respectively, to obtain an XY plane projection map, a YZ plane projection map and an XZ plane projection map, combining an image alignment method based on preset parameters and the obtained corresponding projection map data information, calculating the coordinate information of the image data of each shooting area in the global shooting image data, performing real-time stitching on the obtained corresponding projection maps in each axis, scanning to obtain the three-dimensional image data Stack of the next shooting area, and repeating the above stitching imaging operation.

5. The high-throughput multi-field automatic positioning light sheet microscopy imaging method according to claim 4, characterized in that: The global automatic positioning of the sample area includes stitching images based on the projection images in the three axes of X / Y / Z, and automatically positioning the position of the sample area for each axial stitching image.

6. The high-throughput multi-field automatic positioning light sheet microscopy imaging method according to claim 5, characterized in that: The global automatic positioning of the sample area includes using the maximum inter-class variance method to calculate the image foreground and background threshold segmentation parameter threshold, obtaining the binary image of each axial stitching image, performing edge detection operations on the binary image of each axial stitching image to obtain edge points, summarizing the edge points into groups of contour points according to set rules, traversing the coordinate values ​​of the pixel points corresponding to each contour point, and respectively calculating the maximum and minimum coordinates (xmin, ymin, xmax, ymax) on the X / Y axis of the image to obtain the position information of the sample in the image data, wherein the contour is screened and convex polygons are fitted during the summarization process.

7. The high-throughput multi-field automatic positioning light sheet microscopy imaging method according to claim 1, characterized in that: The optimization of the high-throughput shooting parameters of multiple target areas includes calculating the image coordinate position information corresponding to each target high-throughput, calculating the distance between each target high-throughput area ROI in the image coordinate system, and solving and calculating the optimal path for continuous shooting between each target high-throughput area according to the Hamiltonian path algorithm.

8. The high-throughput multi-field automatic positioning light sheet microscopy imaging method according to claim 7, characterized in that: The depth-first traversal method is used to calculate and solve the Hamiltonian path, including: ① Set the leftmost area as the starting point and mark it as "visited"; ② Using the current "visited" point, select one of the remaining "unvisited" areas for the next visit and add the path length between the two points to the total path length; ③Mark the newly visited area points in ② as the current "visited" status (to avoid repeated visits); ④ Repeat steps ②③ until there is no "unvisited" area, and get the total path length under the current access order; ⑤Backtrack the previously "unvisited" area and select other remaining unvisited points again to repeat ②③④ to obtain the total path length in different visit orders; ⑥ Find the access order with the optimal total path length; Among them, the depth-first traversal method includes commonly used search path optimization and pruning optimization.

9. The high-throughput multi-field automatic positioning light sheet microscopy imaging method according to claim 8, characterized in that: For a single target area, it still includes one or more lens imaging areas, which are divided into several rows or columns of lens imaging areas. Different starting shooting positions include upper left, upper right, lower left and lower right. Different path direction planning includes serpentine direction by row, straight direction by row, serpentine direction by column and straight direction by column.

10. The high-throughput multi-field automatic positioning light sheet microscopy imaging method according to claim 9, characterized in that: For a single target area, by calculating the intersection area between each lens imaging area and the corresponding target foreground binary map, it is determined whether the current imaging area can capture the target to set whether to skip the shooting step.

11. A high-throughput multi-field automatic positioning light sheet microscopy imaging system, characterized in that: The high-throughput multi-field automatic positioning light sheet microscopy imaging method adopts the light sheet fluorescence microscopy rapid imaging method as claimed in any one of claims 1 to 10.

Images