Algorithm for grabbing cells in high-throughput microporous chip by using glass capillary tube

Abstract

The invention belongs to the technical field of biomedical engineering and micro-nano, and provides an algorithm for grabbing cells in a high-throughput microporous chip by using a glass capillary, which comprises the following steps: acquiring image data by using a CCD (Charge Coupled Device), calibrating the microporous chip, moving a first registration point in the chip to the center of the visual field of the CCD by using a moving platform, and moving a second registration point in the chip to the center of the visual field of the CCD; controlling the Z axis to move to enable the CCD to collect clear image information, marking and recording the positions of the registration points, moving to the next registration point, repeatedly executing the steps to obtain coordinate values of nine sets of registration points and corresponding Z axis positions, carrying out calibration algorithm calculation, obtaining a calibration matrix, calculating and selecting the actual position of the Z axis of the micropore, moving the Z axis and the glass capillary, and completing the calibration of the micropore. Controlling the micro-flow pump to extract liquid, sucking the cells into the glass capillary, and then moving out the glass capillary to complete the cell grabbing operation; the problem that the position of the glass capillary extending into the micropore is difficult to determine due to physical distortion of the chip and an included angle and deviation between the chip and the displacement table is solved.

Description

Technical Field

[0001] The invention belongs to the fields of biomedical engineering and micro-nano technology, and specifically relates to an algorithm for grabbing cells in a high-throughput microporous chip using a glass capillary. Background Art

[0002] Single-cell analysis is the key to revealing cell heterogeneity and promoting a deeper understanding of cell fate. In recent years, microfluidics has become an ideal tool for single-cell analysis due to its high throughput and automation advantages. Among various microfluidic platforms, microwell chips have the advantages of simple operation and easy integration with in situ analysis capabilities, making them an ideal technology for single-cell research. Existing microwell chips need to collect relevant signals such as antibody labeling and single-cell staining under a microscopic imaging system. The identified positive target cells need to be picked manually;

[0003] However, the small imaging area of ​​the objective lens results in a small detection function area of ​​the microporous chip and a reduced cell comparison throughput. Increasing the detection function area of ​​the microporous chip will lead to problems such as physical distortion of the chip processing, resulting in the inability of the capillary to be positioned 5μm above all cells with high throughput during the movement of the platform, resulting in failure to capture cells.

[0004] To this end, technicians in this field have proposed an algorithm for using glass capillaries to grab cells in high-throughput microporous chips to solve the problem of being unable to grab cells due to the uncertain distance of the glass capillaries extending into the microporous chip holes caused by physical distortion during the high-throughput microporous chip processing. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides an algorithm for using glass capillaries to capture cells in a high-throughput microporous chip to solve the problems mentioned in the background technology.

[0006] The algorithm for grabbing cells in a high-throughput microwell chip using a glass capillary includes the following steps:

[0007] S1. Place the high-throughput microwell chip on the chip fixture;

[0008] S2, start the software control program and turn on the CCD to collect image data;

[0009] S3, start the calibration operation by clicking the microwell chip calibration button;

[0010] S4, the mobile platform moves the first registration point in the chip to the center of the CCD field of view;

[0011] S5, control the Z-axis movement so that the CCD can collect the clearest image information;

[0012] S6. Mark the position of the registration point and record it in the running memory;

[0013] S7, the mobile platform moves the second registration point in the chip to the center of the CCD field of view;

[0014] S8, controlling the Z-axis focusing, marking, and recording steps again until all nine registration points are operated, and obtaining the coordinate values ​​of nine groups of registration points and the corresponding Z-axis positions;

[0015] S9, performing a calibration algorithm calculation on the three-dimensional coordinates of the nine recorded registration points and the theoretical three-dimensional coordinates of the device system to obtain a calibration matrix;

[0016] S10, calculating the actual Z-axis position of the selected microhole and moving the Z-axis to a position △D above the relative microhole;

[0017] S11, moving the glass capillary by a distance △D to the selected microwell;

[0018] S12, controlling the microfluidic pump to pump liquid so that the glass capillary tube absorbs the cells in the selected well;

[0019] S13. Remove the glass capillary and complete the cell grabbing operation.

[0020] Preferably, the formula for calculating the calibration matrix in S9 is:

[0021]

[0022] Among them, Q o , Q m They are the theoretical coordinate positions and actual coordinate positions of the nine registration points respectively;

[0023]

[0024] Among them, S qo , S qm Q o , Q m The mean matrix of each row;

[0025] Q od =Q O -S qo (5)

[0026] Q md =Q m -S qm (6)

[0027]

[0028] Its SO is the covariance matrix;

[0029] SO=U∑V T (8)

[0030] Among them, U is an m*m orthogonal matrix, which contains the left singular vectors of SO; ∑ is an m*n diagonal matrix, which contains the descending singular values ​​of SO; V represents an n*n orthogonal matrix, which contains the singular vectors of SO;

[0031]

[0032] R om =V*M*U T (10)

[0033] T om =S qm -R om *S qo (11)

[0034] This gives the calibration matrix RT:

[0035]

[0036] The r part is R om , the t part is T om .

[0037] Preferably, the actual Z-axis position of the selected microhole is calculated by the calibration matrix, and the formula is:

[0038]

[0039] SC is the source coefficient matrix, which stores the theoretical values ​​of the hole coordinates;

[0040] DT=RT*SC (14)

[0041]

[0042] What is saved in the DT matrix is ​​the actual final XYZ position.

[0043] Preferably, after calculating the actual Z-axis position of the selected micropore, the Z-axis is moved to a position ΔD above the relative micropore, where ΔD is 5 μm.

[0044] Preferably, the glass capillary is moved a distance ΔD to the selected microwell, wherein ΔD is 5 μm.

[0045] Preferably, the step of controlling the Z-axis movement further includes an auto-focus algorithm based on image clarity, which determines the optimal focal length by calculating the gradient change of the image. The auto-focus algorithm of image clarity can be expressed as: Where I is the image intensity function, x i ,y i are the pixel coordinates and N is the total number of pixels in the image.

[0046] Preferably, the step of controlling the microfluidic pump to pump liquid further includes a flow rate regulation algorithm based on cell volume, which can adjust the pumping rate according to the size of the target cells.

[0047]

[0048] Where Base_Rate is the base pumping rate, Cell_Volume is the target cell volume, Reference_Volume is the reference cell volume, and k is the adjustment coefficient.

[0049] Preferably, the step of calculating the calibration matrix also includes a calibration optimization algorithm based on machine learning, which optimizes the calibration process using historical data, Calibration_Matrix=f(Historical_Data), wherein Historical_Data is historical calibration data.

[0050] Preferably, the algorithm further includes an anomaly detection mechanism for identifying and eliminating anomalies during image acquisition or cell capture. The anomaly detection mechanism can be expressed as:

[0051]

[0052] Feature i is the i-th eigenvalue, ω i is the weight coefficient, and M is the total number of features.

[0053] Preferably, the algorithm also includes a user feedback loop, allowing the user to adjust the algorithm parameters according to the experimental results to optimize the cell capture effect.

[0054] Adjusted_Parameter=Current_Parameter+α×(User_Input-Current_Parameter),

[0055] Where Current_Parameter is the current parameter value, User_Input is the parameter adjustment value entered by the user, and α is the learning rate.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] 1. The present invention overcomes the physical distortion, offset and rotation of the chip in the XYZ axis direction of the mobile platform by using a micron-level calibration algorithm for the displacement of the capillary glass tube, and accurately calculates the position information of the capillary glass tube extending into any selected hole in the 196,000-level range.

[0058] 2. The detection area of ​​the microporous chip of the present invention can be increased to a working area of ​​19 mm × 12 mm, covering 196,000 high-throughput micropores. The algorithm solves the problem of high-throughput positioning of capillaries to 5 μm above all micropores during platform movement.

[0059] 3. The calibration optimization algorithm based on machine learning provided by the present invention significantly improves the calibration accuracy and robustness of cell capture in micro-well chips by comprehensively considering data preprocessing, feature selection, model selection, cross-validation, hyperparameter tuning, model fusion, online learning, feedback loop, model interpretability, deployment and monitoring, etc. The algorithm can adapt to different operating conditions and chip quality differences, realize deep learning and pattern recognition of historical calibration data, and thus optimize the calculation of the calibration matrix. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A flowchart of the algorithm for capturing cells in a high-throughput microwell chip using a glass capillary of the present invention;

[0061] Figure 2 Calibrate the interface for the microwell chip of the present invention;

[0062] Figure 3 The present invention moves to the interface in the micropore;

[0063] Figure 4 The cell interface in the well is selected for capture in the present invention. DETAILED DESCRIPTION

[0064] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0065] As attached Figure 1 To Attachment Figure 4 As shown:

[0066] Embodiment 1: The present invention provides an algorithm for capturing cells in a high-throughput microporous chip using a glass capillary, which specifically includes the following steps:

[0067] S1. Place the high-throughput microwell chip on the chip fixture;

[0068] S2, start the software control program and turn on the CCD to collect image data;

[0069] S3, start the calibration operation by clicking the microwell chip calibration button;

[0070] S4, the mobile platform moves the first registration point in the chip to the center of the CCD field of view, that is, the center of the CCD acquisition area;

[0071] S5, control the Z-axis movement so that the CCD can collect the clearest image information;

[0072] S6. Obtain the coordinate value of the registration point by marking the registration point, and save the coordinate value and the Z-axis position in the running memory;

[0073] S7, the mobile platform moves the second registration point in the chip to the center of the CCD field of view;

[0074] S8, controlling the Z-axis focusing, marking, and recording steps again until all nine registration points are operated, and obtaining the coordinate values ​​of nine groups of registration points and the corresponding Z-axis positions;

[0075] S9, performing a calibration algorithm calculation on the three-dimensional coordinates of the nine recorded registration points and the theoretical three-dimensional coordinates of the device system to obtain a calibration matrix;

[0076] S10, calculating the actual Z-axis position of the selected microhole and moving the Z-axis to a position △D (5 μm) above the relative microhole;

[0077] S11, move the glass capillary by a distance of △D (5 μm) to the selected microwell;

[0078] S12, controlling the microfluidic pump to pump liquid so that the glass capillary tube absorbs the cells in the selected well;

[0079] S13. Remove the glass capillary and complete the cell grabbing operation.

[0080] The formula for calculating the calibration matrix is:

[0081]

[0082] Among them, Q o , Q m They are the theoretical coordinate positions and actual coordinate positions of the nine registration points respectively;

[0083]

[0084] Among them, S qo , S qm Q o , Q m The mean matrix of each row;

[0085] Q od =Q O -S qo (5)

[0086] Q md =Q m -S qm (6)

[0087]

[0088] Its SO is the covariance matrix;

[0089] SO=U∑V T (8)

[0090] Among them, U is an m*m orthogonal matrix, which contains the left singular vectors of SO; ∑ is an m*n diagonal matrix, which contains the descending singular values ​​of SO; V represents an n*n orthogonal matrix, which contains the singular vectors of SO;

[0091]

[0092] R om =V*M*U T (10)

[0093] T om =S qm -R om *S qo (11)

[0094] This gives the calibration matrix RT:

[0095]

[0096] The r part is R om , the t part is T om .

[0097] Preferably, the actual Z-axis position of the selected microhole is calculated by the calibration matrix, and the formula is:

[0098]

[0099] SC is the source coefficient matrix, which stores the theoretical values ​​of the hole coordinates;

[0100] DT=RT*SC (14)

[0101]

[0102] What is saved in the DT matrix is ​​the actual final XYZ position.

[0103] As can be seen from the above, this method and algorithm solve the problem of difficulty in determining the position of the glass capillary extending into the microhole due to physical distortion of the chip and the angle and offset between the chip and the translation stage.

[0104] Verification example: actual verification of the algorithm in the embodiment

[0105] like Figure 2The figure shows the micro-hole chip calibration interface. The position information of the nine registration points in the figure is the real coordinate value of each registration point when it is in the center area of ​​CCD acquisition.

[0106] Substituting the nine real coordinate values ​​into the above formula (1) we can get Q m Matrix, Q o is a theoretical matrix, and its specific value is:

[0107]

[0108] Substituting into the above formula, the RT matrix value is derived as:

[0109]

[0110] Calculate the real coordinates of holes 2-5 in zone 1. The theoretical coordinates of holes 2-5 in zone 1 (368, 162, 5) show that the SC matrix value is:

[0111]

[0112] The DT matrix is ​​calculated by the SC matrix and the RT matrix, and the final DT matrix value is:

[0113]

[0114] The actual coordinate values ​​of holes 2-5 in the final zone are as follows Figure 3 shown.

[0115] Embodiment 2: This embodiment is substantially the same as Embodiment 1, except that the step of controlling the Z-axis movement further includes an autofocus algorithm based on image clarity, which determines the optimal focal length by calculating the gradient change of the image. The autofocus algorithm based on image clarity can be expressed as: Where I is the image intensity function, x i ,y i is the pixel coordinate, N is the total number of pixels in the image, and the image clarity-based autofocus algorithm further includes an adaptive threshold determination mechanism that can dynamically adjust the calculation parameters of the image gradient according to the brightness and contrast of the current image to ensure that the optimal focal length can be accurately identified under different lighting conditions. In addition, the algorithm also integrates a multi-focus evaluation module, which identifies and optimizes multiple focus areas by performing local area analysis on the acquired image to achieve a more refined image clarity evaluation. At the same time, the autofocus algorithm also includes a prediction model that predicts and automatically adjusts the focal length based on historical focus data and environmental change parameters to adapt to rapidly changing experimental conditions. In addition, the algorithm integrates an image quality evaluation indicator, such as a sharpness metric or a Gaussian blur metric, to quantitatively evaluate the clarity of the image and adjust the focal length with real-time feedback.

[0116] Specifically, the autofocus algorithm integrates a multi-scale analysis framework that can evaluate the clarity of the image at different resolution levels, thereby achieving a balance between fast coarse focus and fine fine-tuning. In addition, the algorithm also includes a user preference-based learning module that can adjust the autofocus strategy to meet the needs of specific users based on the user's subjective evaluation of image clarity.

[0117] Preferably, the step of controlling the microfluidic pump to pump liquid further includes a flow rate regulation algorithm based on cell volume, which can adjust the pumping rate according to the size of the target cells.

[0118]

[0119] Where Base_Rate is the basic pumping rate, Cell_Volume is the target cell volume, Reference_Volume is the reference cell volume, and k is the adjustment coefficient. The flow regulation algorithm based on cell volume further includes a real-time feedback system that can monitor the cell suction state during the pumping process and adjust the pumping rate according to the real-time data of the suction state to ensure the integrity and activity of the cells. In addition, the algorithm also has learning capabilities and can self-optimize and adjust parameters based on historical pumping data to adapt to the pumping needs of different types of cells. At the same time, the flow regulation algorithm includes a cell characteristic database for storing different types of cell volumes and corresponding optimal pumping rates. The algorithm quickly matches and applies the pumping rate that best suits the current target cell by querying the database, thereby improving the success rate and efficiency of cell capture.

[0120] Furthermore, the flow regulation algorithm also includes a dynamic adjustment mechanism, which can dynamically adjust the pumping rate according to the real-time monitored cell status and environmental parameters (such as temperature, pressure, etc.), ensuring the best cell capture effect under different experimental conditions. In addition, the algorithm also includes a prediction model to predict the ideal pumping rate corresponding to different cell volumes, thereby achieving more precise flow control.

[0121] Furthermore, the step of calculating the calibration matrix also includes a calibration optimization algorithm based on machine learning, which uses historical data to optimize the calibration process, Calibration_Matrix = f(Historical_Data), where Historical_Data is historical calibration data. The calibration optimization algorithm based on machine learning further uses deep learning technology to construct a multi-layer neural network model to extract features and recognize patterns from historical calibration data, thereby realizing intelligent optimization of the calibration process. The algorithm also includes an adaptive learning rate adjustment mechanism that can dynamically adjust the learning rate according to the learning progress to accelerate the convergence speed and improve the calibration accuracy. The calibration optimization algorithm also includes a model verification module that tests and verifies the accuracy and generalization ability of the model through cross-validation techniques, such as k-fold cross-validation. In addition, the algorithm also includes a hyperparameter optimization mechanism, such as Bayesian optimization, to automatically adjust the model parameters to achieve the optimal calibration effect.

[0122] Preferably, the calibration optimization algorithm also includes an adaptive feature selection module that can automatically identify the features that have the greatest impact on the calibration accuracy and optimize the model training process accordingly. In addition, the algorithm also includes a model fusion framework that improves the accuracy and robustness of the calibration by combining the prediction results of multiple different models.

[0123] Embodiment 3: This embodiment is substantially the same as Embodiment 1, except that the algorithm further includes an abnormality detection mechanism for identifying and eliminating abnormalities during image acquisition or cell capture. The abnormality detection mechanism can be expressed as:

[0124]

[0125] Feature i is the i-th eigenvalue, ω i is the weight coefficient, M is the total number of features. The anomaly detection mechanism further includes a multi-dimensional feature analysis module, which can comprehensively analyze various features in the image acquisition process, such as brightness, contrast, texture, etc., as well as cell morphology, size and other features in the cell capture process. By setting reasonable weight coefficients, rapid identification and response to abnormal situations can be achieved. The anomaly detection mechanism also includes an adaptive update module, which can continuously update the anomaly detection model according to new experimental data to adapt to changes in experimental conditions and new abnormal patterns. In addition, the mechanism also includes a multi-level alarm system, which provides different levels of alarms and recommended countermeasures according to the severity of the anomaly.

[0126] Preferably, the algorithm also includes a user feedback loop, allowing the user to adjust the algorithm parameters according to the experimental results to optimize the cell grabbing effect.

[0127] Adjusted_Parameter=Current_Parameter+α×(User_Input-Current_Parameter),

[0128] Where Current_Parameter is the current parameter value, User_Input is the parameter adjustment value input by the user, α is the learning rate, and the user feedback loop further includes an intelligent parameter recommendation system, which uses data mining technology to analyze the user's operating habits and preferences based on the user's historical adjustment records and experimental results, and provides personalized parameter adjustment suggestions for the user. In addition, the feedback loop also includes an effect evaluation module, which can evaluate the effect of the parameters adjusted by the user and provide feedback to assist the user in more accurate parameter optimization. The user feedback loop also includes a user behavior analysis module, which identifies the user's operating mode and preference by analyzing the user's operation log and feedback to provide more personalized parameter adjustment suggestions. In addition, the feedback loop also includes an effect tracking system for recording and analyzing the effect changes after parameter adjustment, providing a basis for continuous improvement for users and systems. The user feedback loop also includes an intelligent recommendation system, which can recommend the best parameter adjustment strategy based on user feedback and experimental data. In addition, the feedback loop also includes a long-term learning module, which can track the trend of user adjustment parameters and automatically update the recommendation algorithm accordingly to achieve more intelligent parameter recommendation.

[0129] Additionally, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment (ie, those features that are not relevant to the best mode presently contemplated for carrying out the invention or those that are not relevant to implementing the invention) may not be described.

[0130] It should be understood that in the development of any actual implementation, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but for those of ordinary skill having the benefit of this disclosure, without undue experimentation, the development effort will be a routine task of design, fabrication, and production.

[0131] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0132] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0133] In the drawings of the embodiments disclosed in the present invention, only the structures involved in the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other.

[0134] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An algorithm for capturing cells in high-throughput microwell chips using glass capillaries, characterized in that: The specific steps include: S1. Place the high-throughput microwell chip on the chip fixture; S2, start the software control program and turn on the CCD to collect image data; S3, start the calibration operation by clicking the microwell chip calibration button; S4, the mobile platform moves the first registration point in the chip to the center of the CCD field of view; S5, control the Z-axis movement so that the CCD can collect the clearest image information; S6. Mark the position of the registration point and record it in the running memory; S7, the mobile platform moves the second registration point in the chip to the center of the CCD field of view; S8, controlling the Z-axis focusing, marking, and recording steps again until all nine registration points are operated, and obtaining the coordinate values ​​of nine groups of registration points and the corresponding Z-axis positions; S9, performing a calibration algorithm calculation on the three-dimensional coordinates of the nine recorded registration points and the theoretical three-dimensional coordinates of the device system to obtain a calibration matrix; S10, calculating the actual Z-axis position of the selected microhole and moving the Z-axis to a position △D above the relative microhole; S11, moving the glass capillary by a distance △D to the selected microwell; S12, controlling the microfluidic pump to pump liquid so that the glass capillary tube absorbs the cells in the selected well; S13. Remove the glass capillary and complete the cell grabbing operation.

2. The algorithm for capturing cells in a high-throughput microwell chip using a glass capillary as claimed in claim 1, characterized in that: The formula for calculating the calibration matrix in S9 is: Among them, Q o , Q m They are the theoretical coordinate positions and actual coordinate positions of the nine registration points respectively; Among them, S qo , S qm Q o , Q m The mean matrix of each row; Q od =Q o -S qo (5) Q md =Q m -S qm (6) Its S0 is the covariance matrix; SO=U∑V T (8) Among them, U is an m*m orthogonal matrix, which contains the left singular vector of S0; ∑ is an m*n diagonal matrix, which contains the descending singular values ​​of S0; V represents an n*n orthogonal matrix, which contains the singular vector of S0; R om =V*M*U T (10) T om =S qm -R om *S qo (11) This gives the calibration matrix RT: The r part is R om , the t part is T om .

3. The algorithm for capturing cells in a high-throughput microwell chip using a glass capillary as claimed in claim 2, characterized in that: The actual Z-axis position of the selected microhole is calculated by the calibration matrix, and the formula is: SC is the source coefficient matrix, which stores the theoretical values ​​of the hole coordinates; DT=RT*SC (14) What is saved in the DT matrix is ​​the actual final XYZ position.

4. The algorithm for capturing cells in a high-throughput microwell chip using a glass capillary as claimed in claim 1, characterized in that: After calculating the actual Z-axis position of the selected microhole, move the Z-axis to a position △D above the relative microhole, where △D is 5 μm.

5. The algorithm for capturing cells in a high-throughput microwell chip using a glass capillary as claimed in claim 1, characterized in that: The glass capillary was moved to the selected microwell by a distance of ΔD, where ΔD was 5 μm.

6. The algorithm for capturing cells in a high-throughput microwell chip using a glass capillary as claimed in claim 1, characterized in that: The step of controlling the Z-axis movement also includes an auto-focus algorithm based on image clarity, which determines the optimal focal length by calculating the gradient change of the image. The auto-focus algorithm of image clarity can be expressed as: Where I is the image intensity function, x i ,y i are the pixel coordinates and N is the total number of pixels in the image.

7. The algorithm for capturing cells in a high-throughput microwell chip using a glass capillary as claimed in claim 1, characterized in that: The step of controlling the microfluidic pump to pump liquid also includes a flow regulation algorithm based on cell volume, which can adjust the pumping rate according to the size of the target cell. Where Base_Rate is the basic pumping rate, Cell_Volume is the target cell volume, Reference_Volume is the reference cell volume, and k is the adjustment coefficient.

8. The algorithm for capturing cells in a high-throughput microwell chip using a glass capillary as claimed in claim 1, characterized in that: The step of calculating the calibration matrix also includes a calibration optimization algorithm based on machine learning, which optimizes the calibration process using historical data, Calibration_Matrix=f(Historical_Data), where Historical_Data is historical calibration data.

9. The algorithm for capturing cells in a high-throughput microwell chip using a glass capillary as claimed in claim 1, characterized in that: The algorithm also includes an anomaly detection mechanism for identifying and eliminating anomalies in the image acquisition or cell grabbing process. The anomaly detection mechanism can be expressed as: Feature i is the i-th eigenvalue, ω i is the weight coefficient, and M is the total number of features.

10. The algorithm for capturing cells in a high-throughput microwell chip using a glass capillary as claimed in claim 1, characterized in that: The algorithm also includes a user feedback loop that allows the user to adjust algorithm parameters based on experimental results to optimize cell grasping. Adjusted_Parameter=Current_Parameter+α×(User_Input-Current_Parameter), Where Current_Parameter is the current parameter value, User_Input is the parameter adjustment value entered by the user, and α is the learning rate.

Images