Single cell printing equipment control method and system based on convolutional neural network optimization

Abstract

The invention discloses a single cell printing equipment control method and system based on convolutional neural network optimization, and relates to the technical field of intelligent control, and the control method comprises the steps: continuously collecting cell suspension images containing a target single cell before jet printing operation is carried out on a cell suspension, so as to determine each frame of cell main region image, after inputting the shallow convolutional network, respectively constructing a pixel gray parameter set, a pixel response parameter set and a pixel mutation parameter set; analyzing the morphological fluctuation degree of each frame of cell main region image to screen out a standard morphological feature image; the integrity degree of a target single cell structure contained in the standard morphological feature image is analyzed, a cell integrity index is determined to judge whether the integrity degree of the target single cell structure in the current standard morphological feature image meets the printing requirement or not, and a printing control instruction of a corresponding grade is sent out and executed; accurate identification of complete cells in the cell suspension is achieved, and the printing yield and stability are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and specifically to a single-cell printing device control method and system based on convolutional neural network optimization. Background Art

[0002] With the rapid development of cutting-edge research in life sciences and cell engineering, single-cell printing technology, as an important supporting means for microscale biomanufacturing, has been widely used in the fields of cell array construction, tissue structure generation, and personalized drug screening. In particular, with the assistance of microfluidic platforms, precise identification and printing of target single cells through image-driven control has become a key link in achieving high-throughput and high-precision biomanufacturing. Therefore, how to accurately identify target single cells and evaluate their structural integrity before printing has become one of the core issues in improving printing quality and stability.

[0003] For the above-mentioned single-cell printing tasks, the current single-cell recognition and printing control methods mostly rely on traditional image processing technology for target extraction and structural judgment, which makes it difficult to accurately analyze the tiny deformations of cells in the image and the complex structure of damaged edges; moreover, the existing printing control usually makes printing judgments based on a single image frame, ignoring the dynamic fluctuations of cell morphology in the time series, which easily leads to judgment errors; at the same time, the integrity of the cell structure lacks multi-dimensional and quantitative indicator support before actual injection, making it difficult to achieve adaptive control decisions based on multi-parameter fusion; these shortcomings directly affect the recognition accuracy and consistency of the printed products during the printing process, limiting the widespread deployment of single-cell printing in high-precision application scenarios.

[0004] The fundamental reason for the above status quo and defects is that the existing single-cell recognition and printing control methods lack deep structural perception capabilities in image parameter modeling, and it is difficult to construct a model mechanism that reflects the synergistic relationship between pixel structure response, edge mutation and grayscale characteristics; especially in single-cell suspensions, the transparency differences between cells are small, the boundaries are blurred, and there is interference from cell damage fragments. Conventional image discrimination models have difficulty distinguishing the slight differences between intact single cells and damaged structures, and are very likely to misjudge the target structure before the printing instruction is issued; the consequences include: damaged cells are mistakenly printed during the injection process, resulting in tissue construction failure; or intact target cells are misjudged as damaged, and the cell suspension is frequently replaced, resulting in reduced operating efficiency, prolonged printing cycle, and system-level failures such as nozzle clogging and microvalve mis-opening due to delayed control; therefore, it is urgent to construct a printing control method that can optimize the intelligent recognition mechanism based on convolutional neural networks and realize quantitative evaluation of cell integrity to improve the stability and accuracy of single-cell printing. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a single-cell printing device control method and system based on convolutional neural network optimization, which solves the problems in the above-mentioned background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a single cell printing device control method based on convolutional neural network optimization, comprising the following steps:

[0007] S1. Before the cell suspension is jet-printed, continuously collect cell suspension images containing target single cells to determine the main cell area image of each frame. After inputting the images into a shallow convolutional network, respectively construct a pixel grayscale parameter set, a pixel response parameter set, and a pixel mutation parameter set.

[0008] S2. Analyze the degree of morphological fluctuation of the main region image of each frame of the cell based on the pixel grayscale parameter set, the pixel response parameter set, and the pixel mutation parameter set to screen out a standard morphological feature image;

[0009] S3. Based on the screened standard morphological feature images, analyzing the integrity of the target single cell structure contained in the standard morphological feature images to determine the cell integrity index;

[0010] S4. Compare and analyze the cell integrity index to determine whether the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, so as to issue and execute corresponding level printing control instructions.

[0011] Preferably, the specific steps of S1 include:

[0012] S11. Before performing the jet printing operation on the cell suspension, continuously capture images of the cell suspension containing the target single cell at a preset sampling frequency using an industrial camera positioned in the center of the microfluidic channel field of view. After grayscale processing, a cell suspension image frame sequence is constructed, and the main cell region image of each frame is determined based on the projection image range of the target single cell in each frame.

[0013] S12. Divide each frame of the cell main region image into pixels according to the pixel resolution of the image, and record the grayscale value of each pixel in each frame of the image to construct a pixel grayscale parameter set, wherein the pixel grayscale data set includes the grayscale value of each pixel in each frame of the cell main region image;

[0014] S13. Input each frame of the main area image of the cell into a pre-trained shallow convolutional network, and perform structural scanning on the main area image of the cell through multiple convolution kernels respectively, extract the maximum value of the structural response output by all convolution kernels at the corresponding pixel position as the structural response value of the corresponding pixel point, and construct a pixel response parameter set, wherein the pixel response data set includes the structural response value of each pixel point in the main area image of each frame of the cell.

[0015] Preferably, S14, based on the constructed pixel grayscale data set, constructing a pixel mutation parameter set by performing local gradient analysis on the pixel grayscale distribution around each pixel point, specifically including:

[0016] With the current pixel as the center, a 3×3 pixel matrix around the current pixel is selected as the local area input, and convolution operations are performed with the Sobel operators in two standard directions. Specifically, the grayscale value of each pixel in the pixel matrix is ​​multiplied by the corresponding Sobel weight and then summed to obtain the horizontal grayscale change value Gx and vertical grayscale change value Gy of the current pixel, where the standard directions include horizontal and vertical directions;

[0017] The horizontal grayscale change value Gx and the vertical grayscale change value Gy of the current pixel are correlated, and after the two-dimensional gradient modulus is calculated, the local edge mutation coefficient of the current pixel is determined, specifically: Where, Indicates the local edge mutation coefficient of the current pixel;

[0018] According to the determined local edge mutation coefficient of the current pixel, the local edge mutation coefficient of each pixel in each frame of the cell main area image is calculated in sequence, and a pixel mutation parameter set is constructed.

[0019] Preferably, the specific steps of S2 include:

[0020] S21, performing feature recognition on the constructed pixel grayscale parameter set, extracting the grayscale value of each pixel in each frame of the cell main region image, and combining it with a statistical averaging algorithm to obtain the grayscale mean value of each frame of the cell main region image;

[0021] S22. Correlate the constructed pixel grayscale parameter set, pixel response parameter set, and pixel mutation parameter set, analyze the degree of morphological fluctuation of the cell main region image in each frame after dimensionless processing, and determine the morphological tension value of the cell main region image in each frame, specifically: Where, Indicates the morphological tension value of the main area image of the cell in the i-th frame, represents the gray value of the jth pixel in the main area image of the cell in the i-th frame, represents the grayscale mean of the main area image of the cell in the i-th frame, represents the structural response value of the jth pixel in the main area image of the cell in the i-th frame, It represents the local mutation coefficient of the edge of the j-th pixel in the main area image of the cell in the i-th frame, Indicates the number of pixels.

[0022] Preferably, S23, performing comparative analysis based on the morphological tension values ​​of each frame of the acquired cell main region image to determine a standard morphological feature image, specifically includes:

[0023] According to the morphological tension values ​​of the main cell area images of each frame obtained and combined with the statistical averaging algorithm, the mean morphological tension value is obtained; the morphological tension values ​​of the main cell area images of each frame are respectively calculated with the morphological tension mean value to obtain the morphological tension difference of the main cell area images of each frame; after feature extraction, the main cell area image corresponding to the minimum morphological tension difference is selected as the standard morphological feature image.

[0024] Preferably, the specific steps of S3 include:

[0025] S31. Based on the determined standard morphological feature image and in combination with the pixel response parameter set, extracting the structural response value of each pixel in the standard morphological feature image, performing an extreme value search operation, and performing a difference calculation between the maximum structural response value and the minimum structural response value in the obtained standard morphological feature image to determine the structural response amplitude of the standard morphological feature image;

[0026] S32. Based on the standard morphological feature image and in combination with the pixel mutation parameter set, extract the edge local mutation coefficient of each pixel in the standard morphological feature image, compare the edge local mutation coefficient of each pixel with a preset mutation threshold, screen out corresponding pixels whose edge local mutation coefficient exceeds the mutation threshold, mark them as edge pixels, and record the grayscale value of each edge pixel;

[0027] S33. After performing a second-order Laplace transform on the grayscale value of each edge pixel, obtain the second-order transform value of each edge pixel, and determine the boundary sharpening index of the standard morphological feature image by combining a statistical averaging algorithm.

[0028] Preferably, in S34, the morphological tension value, structural response amplitude, and boundary sharpness index of the standard morphological feature image are correlated, and after dimensionless processing, the integrity of the target single cell structure contained in the standard morphological feature image is analyzed to determine the cell integrity index, specifically: Where, represents the cell integrity index, Indicates the morphological tension value, represents the structural response amplitude, Indicates the edge sharpness index.

[0029] Preferably, the specific steps of S4 include:

[0030] S41. Compare and analyze the cell integrity index with the preset integrity threshold to determine whether the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, and issue a corresponding level of printing control instruction, specifically including:

[0031] If the cell integrity index exceeds the integrity threshold, it means that the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, and a first-level printing control instruction is issued;

[0032] If the cell integrity index does not exceed the integrity threshold, it means that the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, and a secondary printing control instruction is issued.

[0033] Preferably, S42, performing corresponding print control operations according to the received primary print control instructions and secondary print control instructions, specifically including:

[0034] When a first-level printing control instruction is received, the execution content is as follows: at this time, it indicates that the target single cell structure in the cell suspension is not damaged, and the nozzle structure is adjusted by the piezoelectric microvalve to perform a jet printing operation on the cell suspension containing the intact target single cell;

[0035] When a secondary printing control instruction is received, the execution content is: this indicates that the target single cell structure in the cell suspension is damaged, so the cell suspension is replaced and the S1-S4 process is repeated until a primary printing control instruction is issued, and the nozzle structure is adjusted by the piezoelectric microvalve to perform a jet printing operation on the cell suspension containing the complete target single cell.

[0036] A single-cell printing device control system based on convolutional neural network optimization, including a pixel parameter acquisition module, a feature screening module, a structure analysis module, and a printing control module;

[0037] The pixel parameter acquisition module is used to continuously acquire cell suspension images containing target single cells before the cell suspension is jet-printed, so as to determine the main cell area image of each frame. After inputting the images into the shallow convolutional network, the pixel grayscale parameter set, pixel response parameter set, and pixel mutation parameter set are constructed respectively.

[0038] The feature screening module is used to analyze the degree of morphological fluctuation of the main area image of each frame of the cell based on the pixel grayscale parameter set, the pixel response parameter set and the pixel mutation parameter set to screen out the standard morphological feature image;

[0039] The structure analysis module is used to analyze the integrity of the target single cell structure contained in the standard morphological feature image based on the screened standard morphological feature image, and determine the cell integrity index;

[0040] The printing control module is used to compare and analyze the cell integrity index to determine whether the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, so as to issue and execute the corresponding level of printing control instructions.

[0041] The present invention provides a single-cell printing device control method and system based on convolutional neural network optimization, which has the following beneficial effects:

[0042] (1) By constructing a single-cell printing device control method and system based on convolutional neural network optimization, the accuracy of target single-cell structure recognition and the adaptability of printing control are significantly improved; this method integrates the multi-core structure perception mechanism of shallow convolutional network, and constructs three types of pixel parameter sets: grayscale characteristics, structural response and edge mutation based on continuous image frame sequence before printing operation, and realizes dynamic quantitative expression of cell morphological tension through dimensionless processing; on this basis, not only can the standard morphological feature images be accurately screened, but also the cell integrity index reflecting the integrity of the single-cell structure can be constructed, and graded printing control instructions can be issued accordingly; overall, it effectively breaks through the limitations of traditional methods that rely on single-frame images and are difficult to evaluate structural integrity, and realizes the accurate identification and intelligent decision-making control of intact cells in single-cell suspensions with strong structural complexity, fuzzy boundaries and a lot of interference information, thereby improving the accuracy, stability and operation efficiency of printing, and providing key technical support for single-cell printing control and its high-precision application.

[0043] (2) The proposed pixel parameter acquisition and convolution feature extraction mechanism forms three nonlinear structure perception channels of grayscale, response and mutation in the image acquisition stage, which significantly enhances the ability to resolve microstructures in cell images, especially forming a clear discrimination interface between the target single cell and its surrounding background, damaged fragments or blurred edge areas; traditional image processing methods generally rely on low-dimensional features such as grayscale enhancement and edge sharpening. When faced with image information with high transparency of cell contours and strong morphological fluctuations, over-generalization of edges or mis-extraction of structures will occur, resulting in target cell recognition errors; the present invention introduces a shallow convolutional network structure and combines it with local gradient mutation calculation to extract the mutation coefficient of the cell edge at the pixel level, thereby constructing a more differentiated pixel mutation parameter set; this process significantly improves the system's ability to model the details of the tiny deformation and damaged edges of intact cells, avoids the misidentification problem caused by unclear boundary judgment of traditional methods, and achieves high-confidence and accurate locking of the target single cell, laying a stable foundation for subsequent integrity judgment and printing control.

[0044] (3) By constructing a cell integrity index, the morphological tension value, structural response amplitude and boundary sharpening index are normalized and integrated to form quantifiable and comparable basic parameters for printing decisions, which effectively makes up for the defects of single control strategy and insufficient feedback response in the existing technology; in traditional methods, printing execution mostly relies on single-frame image judgment, which ignores the dynamic fluctuations of cell morphology in the time series and does not evaluate the cell structure state, resulting in printing failure, nozzle mis-spraying and the need to frequently replace cell suspension; while constructing the cell integrity index parameters, this invention also sets an integrity threshold, divides the first and second level printing control levels, and automatically executes printing or replacement operations based on control instructions to achieve a dynamic closed loop of the printing control link; this mechanism can not only continuously evaluate the target cell state under high-frequency sampling, but also make intelligent adjustments based on the current recognition quality to ensure that the execution of printing instructions has higher stability, robustness and real-time response capabilities, meeting the control accuracy requirements in precision manufacturing scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flow chart of a single-cell printing device control method based on convolutional neural network optimization according to the present invention;

[0046] Figure 2 This is a block diagram of the control system of a single-cell printing device optimized based on a convolutional neural network.

[0047] Figure 3 This is a logical thinking diagram of the single-cell printing device control method based on convolutional neural network optimization of the present invention;

[0048] Figure 4 This is a simplified diagram of the control flow of the single-cell printing device optimized based on convolutional neural networks in the present invention. DETAILED DESCRIPTION

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0050] Example 1

[0051] See also Figure 1 、 Figure 3 and Figure 4 The present invention provides a single cell printing device control method based on convolutional neural network optimization, comprising the following steps:

[0052] S1. Before the cell suspension is jet-printed, continuously collect cell suspension images containing target single cells to determine the main cell area image of each frame. After inputting the images into a shallow convolutional network, respectively construct a pixel grayscale parameter set, a pixel response parameter set, and a pixel mutation parameter set.

[0053] S2. Analyze the degree of morphological fluctuation of the main region image of each frame of the cell based on the pixel grayscale parameter set, the pixel response parameter set, and the pixel mutation parameter set to screen out a standard morphological feature image;

[0054] S3. Based on the screened standard morphological feature images, analyzing the integrity of the target single cell structure contained in the standard morphological feature images to determine the cell integrity index;

[0055] S4. Compare and analyze the cell integrity index to determine whether the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, so as to issue and execute corresponding level printing control instructions.

[0056] In this embodiment, the single-cell printing device control method based on convolutional neural network optimization is provided, which effectively solves the key technical bottlenecks of the existing technology in single-cell identification and printing control, such as insufficient recognition accuracy, one-sided structure judgment, and lack of multi-parameter fusion decision-making; first, by introducing continuous frame image acquisition and shallow convolution network analysis before printing through S1, it no longer relies on single-frame image judgment, and effectively improves the comprehensive perception of temporal fluctuations in cell morphology; the pixel grayscale parameter set, pixel response parameter set and pixel mutation parameter set constructed on this basis form a three-dimensional feature expression channel for cell structure, which significantly enhances the recognition and discrimination ability of fuzzy cell boundary areas, small deformations and damaged structures; through S2 fusion analysis of the above three types of parameter sets, the image can be quantitatively extracted. The morphological tension value of the cells was obtained and the most representative standard morphological feature images were screened out, avoiding the judgment deviation caused by local abnormal frames. Subsequently, S3 further introduced the construction mechanism of the cell integrity index, which realized the quantitative evaluation of structural integrity before printing for the first time, providing a stable and reliable judgment basis for subsequent printing operations. Finally, in S4, the integrity index was compared and analyzed with the preset threshold, and hierarchical printing control instructions were issued, enabling closed-loop control capabilities based on cell state adaptive decision-making. Overall, this method significantly improved the accuracy and stability of pre-printing recognition, making up for the shortcomings of traditional methods in insufficient image modeling depth, delayed structural damage judgment, and single control decision-making, providing a systematic and intelligent printing control path for high-precision single-cell printing.

[0057] Example 2

[0058] Please refer to Figure 1 and Figure 3 , specifically: S1 specific steps include:

[0059] S11. Before performing the jet printing operation on the cell suspension, continuously capture images of the cell suspension containing the target single cell at a preset sampling frequency using an industrial camera positioned in the center of the microfluidic channel field of view. After grayscale processing, a cell suspension image frame sequence is constructed, and the main cell region image of each frame is determined based on the projection image range of the target single cell in each frame.

[0060] The main cell area image of each frame refers to the local image area containing the main structure of the target single cell extracted for each frame of the continuously collected cell suspension image frame sequence, representing the spatial area in the image that is most closely related to the actual existence and structural characteristics of the target single cell; its function is to eliminate irrelevant background or other interfering cellular components in the image, so that the subsequent image parameter extraction and structural analysis can be more focused on the target cell, thereby improving the recognition accuracy and processing efficiency; the area is obtained by first continuously shooting the suspension at a high frequency through an industrial camera set in the central area of ​​the microfluidic channel to obtain a series of grayscale image frames; then, in each frame, the projection range of the target single cell is identified and located, and the entire frame image is cropped according to the projection range to form the main cell area image; this processing not only ensures that the image parameter construction and convolutional network analysis focus on the actual effective cell structure part, but also provides a unified spatial reference for the subsequent construction of grayscale, structural response and mutation parameter sets;

[0061] S12. Divide each frame of the cell main region image into pixels according to the pixel resolution of the image, and record the grayscale value of each pixel in each frame of the image to construct a pixel grayscale parameter set, wherein the pixel grayscale data set includes the grayscale value of each pixel in each frame of the cell main region image;

[0062] The grayscale value of each pixel in each frame of the main area image of the cell refers to the brightness intensity value corresponding to each pixel in each frame of the main area image of the cell, usually expressed as an integer between 0 and 255, which is used to reflect the brightness and darkness of the pixel in the image; the role of these grayscale values ​​is to accurately characterize the internal density, contour boundary and morphological changes of the target single cell structure, and is the basic data source for subsequent image structure recognition, response analysis and mutation detection; the acquisition process is: first, each frame of the main area image is gridded according to the pixel resolution of the image, and it is disassembled into a matrix composed of several pixels; then, through the grayscale processing algorithm, the original color image is converted into a single-channel grayscale image, and the grayscale value of each pixel is read point by point and recorded in sequence, and finally a pixel grayscale parameter set containing the grayscale information of all pixels in all frame images is constructed; this parameter set not only reflects the spatial intensity distribution characteristics of the cell image, but also provides a stable numerical basis for subsequent morphological fluctuation analysis and standard image screening;

[0063] S13. Input each frame of the main area image of the cell into a pre-trained shallow convolutional network, and perform structural scanning on the main area image of the cell through multiple convolution kernels respectively, extract the maximum structural response output of all convolution kernels at the corresponding pixel position as the structural response value of the corresponding pixel point, and construct a pixel response parameter set, wherein the pixel response data set includes the structural response value of each pixel point in the main area image of each frame of the cell.

[0064] The pre-trained shallow convolutional network is used to extract multi-scale structural features from the main region images of each frame of cells. Its core function is to perceive and quantify the local structural patterns, edge continuity and texture features inside the cell by sliding the convolution kernel across the image, thereby providing a highly sensitive discrimination basis for the subsequent generation of structural response values. Compared with traditional image processing methods, shallow convolutional networks have the advantages of being lightweight, low-latency, and suitable for real-time processing. In addition, the network depth is controlled within a limited range, which is more conducive to retaining subtle structural changes in cell images rather than generalized learning. The training process of this convolutional network collects a large number of single-cell image samples with labeled information. Under the supervised learning framework, the structural recognition accuracy or feature response consistency is used as the optimization goal. Combined with cross-entropy or mean squared error, the convolution kernel weights are iteratively updated through the gradient backpropagation algorithm, and finally a stable network parameter set that can effectively capture the typical structural response characteristics of single-cell images is obtained, so that in actual use, it has strong recognition adaptability and robustness to unknown images.

[0065] The structural response value of each pixel in each frame of the main region of the cell is the value formed by taking the maximum value from a set of structural feature responses output by multiple convolution kernels after the main region image of the cell is input into a pre-trained shallow convolutional neural network. After perceiving and extracting features from the surrounding structure at each pixel location, multiple convolution kernels are used to output the maximum value. This value reflects the intensity of the pixel's response to the structural pattern in the local area. Its function is to capture the potential deep-level features in the cell image, such as structural boundaries, morphological contours, and organelle distribution. It effectively overcomes the limitation of traditional grayscale values ​​that only reflect brightness and lack structural perception, and provides more discriminative feature support for identifying cell integrity, boundary continuity, and deformation degree. The process of obtaining this structural response value is as follows: first, each frame of the main region image is input into the trained shallow convolutional network model. Then, the image is slidably scanned using multiple convolution kernels of different sizes or different directional perception capabilities. Each convolution kernel calculates a response value for each pixel and its neighborhood. Finally, at each pixel location, the maximum value from the output responses of all convolution kernels is selected as the structural response value of that point. Based on this, a set of pixel response parameters is constructed to achieve structural perception modeling of all pixels in the cell region.

[0066] S14. Based on the constructed pixel grayscale dataset, a pixel mutation parameter set is constructed by performing a local gradient analysis on the pixel grayscale distribution around each pixel point, specifically including:

[0067] With the current pixel as the center, a 3×3 pixel matrix around the current pixel is selected as the local area input, and convolution operations are performed with the Sobel operators in two standard directions. Specifically, the grayscale value of each pixel in the pixel matrix is ​​multiplied by the corresponding Sobel weight and then summed to obtain the horizontal grayscale change value Gx and vertical grayscale change value Gy of the current pixel, where the standard directions include horizontal and vertical directions;

[0068] The horizontal grayscale change value Gx and vertical grayscale change value Gy of the current pixel are calculated based on the spatial distribution of grayscale values ​​in the horizontal and vertical directions, respectively, with the pixel as the center. These values ​​are key indicators for measuring the magnitude of the grayscale gradient of the pixel in the two main directions. They reflect the degree of edge change and directional characteristics of local regions in the image. In cell images, they are particularly useful for identifying cell boundaries, damaged edges, or mutation points, effectively assisting in determining the integrity and edge continuity of the target structure. The method for obtaining these values ​​is as follows: first, a grayscale value matrix of the 3×3 pixels surrounding the current pixel is extracted. This matrix is ​​then convolved with two standard-direction Sobel operators (classic convolution kernels used for edge detection). This means that the grayscale value of each pixel in the matrix is ​​multiplied by the value of the corresponding position in the Sobel weight matrix and the summed results are used to obtain the grayscale change value of the current pixel in the horizontal direction Gx and the vertical direction Gy. The combination of these two values ​​can be further used to calculate the degree of local mutation, providing basic data for the subsequent construction of the mutation parameter set.

[0069] The horizontal grayscale change value Gx and the vertical grayscale change value Gy of the current pixel are correlated, and after the two-dimensional gradient modulus is calculated, the local edge mutation coefficient of the current pixel is determined, specifically: Where, Indicates the local edge mutation coefficient of the current pixel;

[0070] It should be noted that the above formula is an important process for calculating the edge local mutation coefficient of the current pixel in the image, wherein Gx and Gy represent the grayscale change values ​​of the pixel in the horizontal and vertical directions, respectively. The formula comprehensively evaluates the pixel grayscale change intensity through the two-dimensional gradient modulus, thereby accurately characterizing its local edge mutation characteristics; this calculation plays a key role in the present invention: by accurately extracting the grayscale mutation characteristics of the boundary area in the image, the ability to recognize edge structures in single-cell images is significantly improved, especially in the context of blurred boundaries and complex damage interference of cell suspensions mentioned in the background technology, it can effectively solve the problem that traditional methods are difficult to distinguish between intact cells and broken structures; the edge local mutation coefficient, as a parameter characterizing the sensitivity of pixel boundaries in cell images, can be used to screen and locate structural edge pixels, and can also provide a quantitative basis for the subsequent construction of boundary sharpening index and cell integrity evaluation, thereby playing a basic supporting role in achieving high-precision structure recognition, controlling injection decisions and improving printing consistency, and is a key parameter throughout the core image analysis process of the printing control method.

[0071] According to the determined local edge mutation coefficient of the current pixel, the local edge mutation coefficient of each pixel in each frame of the cell main area image is calculated in sequence, and a pixel mutation parameter set is constructed.

[0072] In this embodiment, through the multi-step design of S1, a high-dimensional pixel parameter construction mechanism integrating continuous image acquisition, deep structure recognition and edge mutation analysis is established before single-cell printing, which significantly improves the image expression quality and microstructure recognition ability of target cells, and breaks through the problem that the existing technology only relies on single-frame images and lacks deep structural perception; first, S11 uses an industrial camera to perform continuous high-frequency sampling in the core area of ​​the microfluidic channel, effectively capturing the overall picture and dynamic characteristics of cells in the suspension during movement, providing a time-stable image foundation for subsequent structural analysis; then, S12 constructs a pixel grayscale parameter set to completely record the grayscale distribution details in the main area of ​​the cell, laying a statistical foundation for subsequent morphological change trend analysis; S13 introduces a pre-trained shallow convolutional network, and provides a multi-convolution kernel structure scanning mechanism to improve the quality of image expression and microstructure recognition. Taking the response intensity of each pixel at the structural level significantly enhances the ability to recognize the local structural characteristics, contour clarity and morphological differences of cells; in S14, the Sobel operator is combined to perform bidirectional gradient convolution analysis on the local 3×3 pixel area to accurately capture the mutation behavior of the cell edge, and finally form a pixel mutation parameter set, which makes up for the defect that traditional grayscale analysis is difficult to quantify local boundary changes; overall, this group of steps not only breaks the previous technical limitations of extracting cell information from a single image dimension or static frame, but also establishes an image parameter system that integrates grayscale characteristics, structural response and edge mutation. It can still achieve high-precision modeling of the target single cell structure under complex background interference and high-transparency cell boundary fuzzy conditions, providing a reliable, accurate and multi-source feature basis for subsequent integrity judgment and printing control.

[0073] Example 3

[0074] Please refer to Figure 1 and Figure 3 , specifically: S2 specific steps include:

[0075] S21, performing feature recognition on the constructed pixel grayscale parameter set, extracting the grayscale value of each pixel in each frame of the cell main region image, and combining it with a statistical averaging algorithm to obtain the grayscale mean value of each frame of the cell main region image;

[0076] S22. Correlate the constructed pixel grayscale parameter set, pixel response parameter set, and pixel mutation parameter set, analyze the degree of morphological fluctuation of the cell main region image in each frame after dimensionless processing, and determine the morphological tension value of the cell main region image in each frame, specifically: Where, Indicates the morphological tension value of the main area image of the cell in the i-th frame, represents the gray value of the jth pixel in the main area image of the cell in the i-th frame, represents the grayscale mean of the main area image of the cell in the i-th frame, represents the structural response value of the jth pixel in the main area image of the cell in the i-th frame, It represents the local mutation coefficient of the edge of the j-th pixel in the main area image of the cell in the i-th frame, Indicates the number of pixels.

[0077] It should be noted that the formula in S22 is the core expression for calculating the morphological tension value of the main area of ​​the cell image in each frame. By weighted coupling of the pixel grayscale deviation value, the structural response value, and the edge local mutation coefficient, a quantitative measurement of the overall morphological stability and change amplitude of the image is formed. In this formula, represents the grayscale mean value of the main area image of the cell in the i-th frame, reflecting the grayscale deviation degree of the pixel point; Represents the edge local mutation coefficient of the jth pixel in the main area image of the cell in the i-th frame. It reflects the edge sensitivity and structural mutation status of the image. The morphological tension value obtained through the composite calculation of these three types of parameters accurately depicts the stability and dynamic fluctuation of the microstructure of each frame of the image. The calculation logic is closely related to the problem that the traditional method pointed out in the background technology is difficult to capture the dynamic fluctuation of cell morphology in the time series. This formula can identify small but critical deformation trends in multi-frame sequence images, effectively extract morphological fluctuation feature images, and provide an accurate basis for the subsequent screening of the most stable and representative standard morphological images. At the same time, the morphological tension value, as a measure of the degree of structural change of a single frame image, can not only assist in identifying abnormal image frames, but also serve as the core intermediate quantity of multi-parameter fusion, playing a key supporting role in improving the overall recognition accuracy and control decision-making intelligence, and is an important bridge throughout the dynamic structure analysis and judgment execution.

[0078] In the study of image processing and cell morphology recognition, traditional methods rely solely on grayscale values ​​or edge detection indicators to measure structural changes. However, in complex cell images, single grayscale statistics or boundary sharpening often cannot simultaneously reflect the overall structural balance and local mutation intensity. Therefore, a new measurement indicator is needed that can comprehensively reflect local pixel deviation, structural response degree, and boundary disturbance. The morphological tension value was proposed in this context. The theoretical basis for the calculation of this formula comes from the idea of ​​coupling the tensor field with the local response of the image. The pixel grayscale deviation is regarded as the tensor strain, the convolution response result is mapped into the structural force, and the boundary disturbance coefficient is used as the local constraint. The product term formed by the coupling of the three is normalized and averaged across all pixels to obtain the morphological tension value, which is used to describe the morphological stability of the main area image of the cell in each frame. Therefore, by analogy with mechanical tension, the calculation method of the morphological tension value is derived from the three dimensions of image grayscale discreteness (deviation), structural contribution (response), and boundary disturbance (constraint).

[0079] The specific application of the calculation formula of the above-mentioned morphological tension value in the present invention is as follows:

[0080] Image acquisition stage output: After each image frame is acquired, grayscale extraction and structural analysis are performed on the main area, and the morphological tension value is constructed;

[0081] Tension value calculation and screening execution: Calculate the morphological tension value for each frame of the cell main area image, which is used to screen the standard morphological feature image;

[0082] The basis for selecting the standard morphological feature image is as follows: the morphological tension value of each frame of the cell main area image is calculated with the morphological tension mean value, and the cell main area image corresponding to the minimum morphological tension difference is selected as the standard morphological feature image.

[0083] Specifically, S23, performing comparative analysis based on the morphological tension values ​​of the acquired cell main region images of each frame to determine a standard morphological feature image, specifically includes:

[0084] According to the morphological tension values ​​of the main cell area images of each frame obtained and combined with the statistical averaging algorithm, the mean morphological tension value is obtained; the morphological tension values ​​of the main cell area images of each frame are respectively calculated with the morphological tension mean value to obtain the morphological tension difference of the main cell area images of each frame; after feature extraction, the main cell area image corresponding to the minimum morphological tension difference is selected as the standard morphological feature image.

[0085] The standard morphological feature image is a frame of image that is closest to the overall morphological stability trend and has the most representative structure after morphological tension value analysis and comparison among multiple frames of continuously acquired cell main area images. It reflects the most complete structure and the most balanced morphology of the target single cell in the current time series. Its function is to provide a unified and reliable image basis for subsequent structural integrity assessment, avoiding judgment deviations caused by sudden blur, edge interference or local anomalies in single-frame images. The specific selection process is as follows: first calculate the morphological tension value of all frame images and find the average value, then perform the difference operation between the tension value of each frame image and the average value. The smaller the difference, the closer the image morphology is to the overall trend and the smaller the fluctuation. Finally, the image with the smallest morphological tension difference is selected as the standard morphological feature image. Through this screening mechanism, the best analysis object is automatically determined from the time series images, effectively improving the stability of structural analysis and the accuracy of judgment.

[0086] In this embodiment, the grayscale, response and mutation fusion analysis and morphological tension evaluation mechanism set by S2 effectively breaks through the key technical shortcomings of traditional single-cell printing control that only uses a single image frame to make judgments and ignores temporal morphological fluctuations, and realizes the quantitative evaluation of the dynamic stability of cell structure and the automatic screening of the optimal image; specifically, S21 establishes the overall brightness baseline of each frame image as the starting parameter for measuring the changes in cell morphological distribution through full-frame extraction and statistical mean analysis of the pixel grayscale parameter set; and in S22, the three types of parameters, grayscale value, structural response value and edge mutation coefficient, are normalized under the same index system for the first time, and the morphological tension value is comprehensively constructed to achieve a unified measurement of the intensity of image morphological fluctuations; this value can keenly reflect whether the cell exists in continuous frame images. Abnormal behaviors such as mutation, blur, damage or offset can be detected, thus breaking through the limitations of traditional methods that rely on local edge features and are easily misjudged by background interference; the subsequent S23 uses the statistical mean as a reference benchmark, compares the morphological tension difference of each frame image, and automatically selects the frame image with the most stable morphology and the structure that best meets the expected standards as the standard morphological feature image; the setting of this step chain not only realizes the automatic judgment of the best quality frame in the image group, but also effectively avoids the risk of wrong decision-making due to single-frame abnormalities, and has stronger robustness and accuracy in dealing with actual application scenarios with complex cell morphology and significant structural fluctuations; therefore, this module demonstrates unique technical advantages in dealing with background technical difficulties such as the difficulty in identifying structural damage and the lack of quantitative basis for dynamic image changes, and significantly improves the decision-making basis for subsequent structural judgment and printing control.

[0087] Example 4

[0088] Please refer to Figure 1 and Figure 3 , specifically: S3 specific steps include:

[0089] S31. Based on the determined standard morphological feature image and in combination with the pixel response parameter set, extracting the structural response value of each pixel in the standard morphological feature image, performing an extreme value search operation, and performing a difference calculation between the maximum structural response value and the minimum structural response value in the obtained standard morphological feature image to determine the structural response amplitude of the standard morphological feature image;

[0090] The structural response amplitude is the difference between the maximum and minimum structural response values ​​of all pixels in the standard morphological feature image. This value reflects the intensity fluctuation range of the structural features in the spatial distribution of the image and is an important indicator for measuring the overall structural clarity and configuration integrity of the image. Its function is to characterize whether the target single cell in the current image has clear and continuous structural boundaries and significant internal configuration differences, thereby assisting in determining whether the cell has structural integrity suitable for printing. The acquisition method is: first, in the standard morphological feature image, combined with the pixel response parameter set, the structural response values ​​of all pixels are extracted; then, through extreme value search, the maximum and minimum response values ​​are found, and the difference between the two is calculated to obtain the structural response amplitude; the larger the value, the richer the structural details and the clearer the boundaries in the image, which helps to accurately distinguish the target cells from the surrounding background or interfering structures. It is one of the key components of the cell integrity index.

[0091] S32. Based on the standard morphological feature image and in combination with the pixel mutation parameter set, extract the edge local mutation coefficient of each pixel in the standard morphological feature image, compare the edge local mutation coefficient of each pixel with a preset mutation threshold, screen out corresponding pixels whose edge local mutation coefficient exceeds the mutation threshold, mark them as edge pixels, and record the grayscale value of each edge pixel;

[0092] The mutation threshold is determined by statistically analyzing the grayscale gradient characteristics of edge and non-edge areas in a large number of cell images, combining the principle of optimal edge recognition accuracy, and using empirical judgment or optimization algorithm to adaptively set;

[0093] Edge pixels refer to pixels with drastic local grayscale changes and significant edge features in the standard morphological feature image. They are specifically marked and identified by pixels whose local edge mutation coefficient exceeds the preset mutation threshold, and are usually located at the boundary contour of the target single cell structure. Their function is to accurately locate the outer edge area of ​​the cell, facilitate the assessment of whether the cell structure is complete and continuous, and provide a key basis for subsequent boundary sharpening analysis and integrity calculation. The acquisition process is as follows: the edge local mutation coefficient of all pixels in the standard image is extracted from the pixel mutation parameter set, that is, the local change intensity index calculated by grayscale gradient analysis; then the mutation coefficient of each pixel is compared with the mutation threshold set by the system, and the pixels whose edge local mutation coefficient exceeds the mutation threshold are considered to have significant edge mutation and are determined to be edge pixels; finally, the grayscale values ​​of these pixels are recorded for the subsequent calculation of the boundary sharpening index; this step can effectively eliminate the interference of the homogeneous area inside the image and focus on the extraction of the change characteristics of the contour area. It is an important basis for identifying cell structure damage, boundary discontinuity or abnormal deformation;

[0094] S33. After performing a second-order Laplace transform on the grayscale value of each edge pixel, obtain the second-order transform value of each edge pixel, and determine the boundary sharpening index of the standard morphological feature image by combining a statistical averaging algorithm.

[0095] The edge sharpness index (BSI) is an indicator derived by performing a second-order Laplace transform on edge pixels in a standard morphological feature image and then averaging the second-order transform values ​​of each edge point. It quantifies the clarity and sharpness of cell edge contours in an image. Its core function is to assess the integrity and continuity of the target single-cell structure boundary, determining whether it possesses a clear, undamaged geometric boundary morphology. This is a key factor in determining whether a cell structure is suitable for printing. The index is obtained by first extracting the edge pixels screened in the previous step and obtaining their corresponding grayscale values. These grayscale values ​​are then subjected to a second-order Laplace transform, leveraging their sensitivity to the second-order derivative of grayscale changes to enhance the response of the contour region in the image, thereby highlighting the structural boundary. Finally, the Laplace transform results of all edge pixels are averaged, resulting in the BSI. A larger value indicates a more dramatic grayscale change in the edge region and a clearer and more complete boundary, while a smaller value indicates blurred or damaged cell edges, helping the system automatically identify printing risks and make accurate control decisions.

[0096] Specifically, S34, the morphological tension value, structural response amplitude, and boundary sharpness index of the standard morphological feature image are correlated, and after dimensionless processing, the integrity of the target single cell structure contained in the standard morphological feature image is analyzed to determine the cell integrity index, which is specifically: Where, represents the cell integrity index, Indicates the morphological tension value, represents the structural response amplitude, represents the boundary sharpening index, where Represents a logarithmic function that reflects the cell morphological tension and is used to quantify the degree of structural fluctuation. The exponential attenuation factor of the structural response amplitude is expressed. The larger the structural response amplitude, the more dramatic the internal structure difference and the lower the cell integrity. It represents the inverse measure of the boundary sharpness index. The larger the boundary sharpness index is, the clearer the boundary is and the higher the cell integrity is.

[0097] It should be noted that the formula in S34 is a key quantitative indicator for calculating the integrity of cell structure, namely the cell integrity index, which evaluates whether the target single-cell structure in the standard morphological feature image is printable. The formula comprehensively considers the three dimensions of the degree of image morphological change, the intensity of structural response, and the quality of edge sharpening to form a basis for judging intelligent structural recognition. As mentioned in the background technology, traditional recognition methods are difficult to quantitatively evaluate whether the structure of a single cell is complete, and often lead to misjudgment due to the lack of multi-dimensional data fusion. By coupling and calculating multiple key visual parameters through this formula, the problem of difficulty in judging damaged cell structures and misjudging intact structures is effectively solved. By measuring the overall morphological tension, structural response, and edge clarity of the cell in the image, the index accurately reflects the integrity of the cell structure, thus serving as the core quantitative basis for printing decisions. The cell integrity index is a core indicator for comprehensively evaluating whether the target single cell identified in the standard morphological feature image has structural integrity, edge clarity, and texture stability. This indicator provides a quantitative standard before judging whether to execute printing. This mechanism significantly improves the recognition accuracy and printing control intelligence, and is a basic evaluation mechanism to support high-throughput, high-reliability single-cell printing processes.

[0098] The logic of constructing the formula of the above-mentioned cell integrity index is derived from the technical requirements of the parallel evaluation of three key factors in the process of cell structure integrity analysis: the first is the overall grayscale tension of the image, that is, whether the fluctuation amplitude of the pixel grayscale in the cell area reflects continuous and regular changes; the second is the deep structure perception of the image, that is, whether the activation response of the convolutional neural network to the input structure in the middle and back layers is stable and balanced; the third is the sharpness level of the boundary, that is, whether the cell edge has a continuous and clear pixel gradient change trend; among these three, grayscale tension is a characteristic indicator at the static image level, the structural response amplitude reflects the characteristics of the deep feature extraction module, and the edge sharpening index is an image sharpening indicator. Like the analytical performance of microscopic local areas; in traditional image evaluation, these three are often judged separately, while the fusion algorithm of the cell integrity index is used to respond to the above three core characteristics simultaneously, completing the indicator construction without relying on weights or introducing normalization coefficients; the natural logarithm function, exponential function and fractional expression are introduced into the formula, so that the effect of each parameter on the final result can maintain the nonlinear response characteristics without dominating the output in absolute value; this type of function structure has balance and attenuation characteristics, and is the theoretical mathematical basis for realizing the reasonable evaluation of complex structural responses of images. It has the composite processing capability of fusing the output of the neural network feature layer with low-order image features;

[0099] Among them, the three parts constitute a step-by-step suppression product function structure, each item is positive and has a nonlinear response relationship; its derivation path is: By compressing and mapping the morphological tension value, the output effect is kept convergent when the tension is high; the second The internal response mutation is constrained on the completeness index through the exponential decay mechanism, which intuitively prevents the image of the structural mutation area in the activation layer from being misjudged as a complete image; the third The inverse expression of the edge response term is constructed using the boundary convergence ratio. The value of this term decreases as edge sharpening increases. By adding the constant term 2 to the denominator, the problem of near-zeroing caused by the minimum value of Zrh is avoided, ensuring that edge-blurred images are not mistakenly judged as intact. The entire formula avoids the use of fixed weighting coefficients. All factors are independent but respond in combination, which is mathematically differentiable and continuous, and can be directly used for subsequent threshold judgment and strategy selection.

[0100] In this embodiment, through deep structural analysis and edge detail enhancement processing of standard morphological feature images, multi-dimensional quantitative expression of single cell structural integrity is achieved for the first time, providing a core evaluation index with discrimination and adaptability for high-precision printing control; in S31, based on the pixel response parameter set, the standard image is searched for extreme values, and the difference between the maximum and minimum structural response values, namely the structural response amplitude, is extracted and calculated. This parameter can reflect the overall structural fluctuation and morphological stability of the cell in the main area image, and is particularly suitable for identifying local rupture, collapse or abnormal areas; subsequently, S32 and S33 further introduce the pixel mutation parameter set, perform grayscale extraction and second-order Laplace transform on the edge pixels in the image that exceed the mutation threshold, and finally generate a sharp edge. Index, which can keenly capture the clarity and boundary integrity of cell edges, and effectively distinguish intact structures from damaged fragments or interference artifacts; on this basis, S34 integrates the morphological tension value, structural response amplitude and boundary sharpness index in a dimensionless manner to form a unified integrity index system, realizing the numerical characterization of the target single-cell structural state; this mechanism effectively solves the problems of traditional methods that are difficult to quantify structural integrity, fuzzy judgment criteria, and difficult to integrate multiple types of parameters for joint decision-making. It can not only make accurate judgments on intact cells, but also adapt to complex morphological changes through continuity indicators, significantly improving the reliable judgment ability before printing and the system automation level, and meeting the needs of high-end single-cell printing scenarios that attach equal importance to structural recognition accuracy and control response sensitivity.

[0101] Example 5

[0102] Please refer to Figure 1 and Figure 3 , specifically: S4 specific steps include:

[0103] S41. Compare and analyze the cell integrity index with the preset integrity threshold to determine whether the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, and issue a corresponding level of printing control instruction, specifically including:

[0104] If the cell integrity index exceeds the integrity threshold, it means that the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, and a first-level printing control instruction is issued;

[0105] If the cell integrity index does not exceed the integrity threshold, it means that the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, and a secondary printing control instruction is issued.

[0106] The integrity threshold is a benchmark value used to determine whether the target single-cell structure meets the requirements of inkjet printing. Its function is to provide a quantitative decision-making standard to clearly classify the cell integrity index, thereby distinguishing between structurally intact and potentially damaged cell states, and guiding the issuance of printing control instructions. The setting of this threshold is based on a large amount of experimental data statistics and expert experience. The specific method includes: during the sample training phase, the integrity index is calculated for multiple cell images with known intact and damaged structures. Combined with image quality grading and printing success rate assessment, statistical analysis methods (ROC curve analysis or minimum error rate method) are used to select the critical value that best distinguishes the two types of structural states as the integrity threshold.

[0107] Specifically, S42, performing corresponding print control operations according to the received primary print control instructions and secondary print control instructions, specifically including:

[0108] When a first-level printing control instruction is received, the execution content is as follows: at this time, it indicates that the target single cell structure in the cell suspension is not damaged, and the nozzle structure is adjusted by the piezoelectric microvalve to perform a jet printing operation on the cell suspension containing the intact target single cell;

[0109] When a secondary printing control instruction is received, the execution content is: this indicates that the target single cell structure in the cell suspension is damaged, so the cell suspension is replaced and the S1-S4 process is repeated until a primary printing control instruction is issued, and the nozzle structure is adjusted by the piezoelectric microvalve to perform a jet printing operation on the cell suspension containing the complete target single cell.

[0110] In this embodiment, by introducing a hierarchical printing control mechanism with the cell integrity index as the core criterion, an intelligent printing decision-making system with self-judgment, self-adjustment and self-closed-loop execution capabilities is constructed, which significantly solves a series of systemic defects in the existing technology caused by the lack of quantitative basis for structural state judgment, single printing control decision and delayed operation after misjudgment; specifically, S41 uses the preset integrity threshold as a benchmark to compare and analyze the cell integrity index, clearly divides whether the printing conditions are met, and issues a first-level printing control instruction or a second-level printing control instruction accordingly. It has dynamic adjustment capabilities when facing target cells of different quality states, breaking through the traditional technical bottleneck of relying on static judgment and being difficult to adapt to the complexity of the actual suspension. S42 further forms a closed-loop linkage between the control instruction and the execution action, in which the first-level printing control instruction directly drives the piezoelectric microvalve injection operation , ensuring that high-quality cells are smoothly printed under structural integrity conditions. When a secondary printing control instruction is issued, the suspension replacement and full-process re-inspection are automatically triggered until the printing conditions are met and the operation is performed again, reducing the probability of damaged cells being mistakenly sprayed or unqualified printing loop redundancy; more importantly, the mechanism establishes a closed-loop path of "structural quality identification-control judgment-execution feedback-structural re-judgment", which effectively avoids the system failure risks of tissue construction failure, reduced printing efficiency, nozzle clogging and microvalve mis-opening due to misjudgment, and achieves a triple improvement in printing process stability, control accuracy and response efficiency; therefore, S4 not only completes the closed-loop closure of "identification-judgment-control" in the technical path, but also gives the system a high degree of adaptability and fault tolerance to single-cell printing tasks in practical applications. It is an important guarantee mechanism for promoting precision biomanufacturing to high reliability and high consistency.

[0111] Example 6

[0112] Please refer to Figure 1 and Figure 2 ,Specifically: A single-cell printing device control system based on convolutional neural network optimization,,including pixel parameter acquisition module, feature screening module, structure analysis module and printing control module;

[0113] The pixel parameter acquisition module is used to continuously acquire cell suspension images containing target single cells before the cell suspension is jet-printed, so as to determine the main cell area image of each frame. After inputting the images into the shallow convolutional network, the pixel grayscale parameter set, pixel response parameter set, and pixel mutation parameter set are constructed respectively.

[0114] The feature screening module is used to analyze the degree of morphological fluctuation of the main area image of each frame of the cell based on the pixel grayscale parameter set, the pixel response parameter set and the pixel mutation parameter set to screen out the standard morphological feature image;

[0115] The structure analysis module is used to analyze the integrity of the target single cell structure contained in the standard morphological feature image based on the screened standard morphological feature image, and determine the cell integrity index;

[0116] The printing control module is used to compare and analyze the cell integrity index to determine whether the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, so as to issue and execute the corresponding level of printing control instructions.

[0117] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A single-cell printing device control method based on convolutional neural network optimization, characterized by: The following steps are involved: S1. Before the cell suspension is jet-printed, continuously collect cell suspension images containing target single cells to determine the main cell area image of each frame. After inputting the images into a shallow convolutional network, respectively construct a pixel grayscale parameter set, a pixel response parameter set, and a pixel mutation parameter set. S2. Analyze the degree of morphological fluctuation of the main region image of each frame of the cell based on the pixel grayscale parameter set, the pixel response parameter set, and the pixel mutation parameter set to screen out a standard morphological feature image; S3. Based on the screened standard morphological feature images, analyzing the integrity of the target single cell structure contained in the standard morphological feature images to determine the cell integrity index; S4. Compare and analyze the cell integrity index to determine whether the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, so as to issue and execute corresponding level printing control instructions.

2. The single-cell printing device control method based on convolutional neural network optimization according to claim 1, characterized in that: The specific steps of S1 include: S11. Before performing the jet printing operation on the cell suspension, continuously capture images of the cell suspension containing the target single cell at a preset sampling frequency using an industrial camera positioned in the center of the microfluidic channel field of view. After grayscale processing, a cell suspension image frame sequence is constructed, and the main cell region image of each frame is determined based on the projection image range of the target single cell in each frame. S12. Divide each frame of the cell main region image into pixels according to the pixel resolution of the image, and record the grayscale value of each pixel in each frame of the image to construct a pixel grayscale parameter set, wherein the pixel grayscale data set includes the grayscale value of each pixel in each frame of the cell main region image; S13. Input each frame of the main area image of the cell into a pre-trained shallow convolutional network, and perform structural scanning on the main area image of the cell through multiple convolution kernels respectively, extract the maximum value of the structural response output by all convolution kernels at the corresponding pixel position as the structural response value of the corresponding pixel point, and construct a pixel response parameter set, wherein the pixel response data set includes the structural response value of each pixel point in the main area image of each frame of the cell.

3. The single-cell printing device control method based on convolutional neural network optimization according to claim 2, characterized in that: S14. Based on the constructed pixel grayscale dataset, a pixel mutation parameter set is constructed by performing a local gradient analysis on the pixel grayscale distribution around each pixel point, specifically including: With the current pixel as the center, a 3×3 pixel matrix around the current pixel is selected as the local area input, and convolution operations are performed with the Sobel operators in two standard directions. Specifically, the grayscale value of each pixel in the pixel matrix is ​​multiplied by the corresponding Sobel weight and then summed to obtain the horizontal grayscale change value Gx and vertical grayscale change value Gy of the current pixel, where the standard directions include horizontal and vertical directions; The horizontal grayscale change value Gx and the vertical grayscale change value Gy of the current pixel are correlated, and after the two-dimensional gradient modulus is calculated, the local edge mutation coefficient of the current pixel is determined, specifically: Where, Indicates the local edge mutation coefficient of the current pixel; According to the determined local edge mutation coefficient of the current pixel, the local edge mutation coefficient of each pixel in each frame of the cell main area image is calculated in sequence, and a pixel mutation parameter set is constructed.

4. The single-cell printing device control method based on convolutional neural network optimization according to claim 3 is characterized in that: The specific steps of S2 include: S21, performing feature recognition on the constructed pixel grayscale parameter set, extracting the grayscale value of each pixel in each frame of the cell main region image, and combining it with a statistical averaging algorithm to obtain the grayscale mean value of each frame of the cell main region image; S22. Correlate the constructed pixel grayscale parameter set, pixel response parameter set, and pixel mutation parameter set, analyze the degree of morphological fluctuation of the cell main region image in each frame after dimensionless processing, and determine the morphological tension value of the cell main region image in each frame, specifically: Where, Indicates the morphological tension value of the main area image of the cell in the i-th frame, represents the gray value of the jth pixel in the main area image of the cell in the i-th frame, represents the grayscale mean of the main area image of the cell in the i-th frame, represents the structural response value of the jth pixel in the main area image of the cell in the i-th frame, It represents the local mutation coefficient of the edge of the j-th pixel in the main area image of the cell in the i-th frame, Indicates the number of pixels.

5. The single-cell printing device control method based on convolutional neural network optimization according to claim 4 is characterized in that: S23, performing comparative analysis based on the morphological tension values ​​of the acquired cell main region images in each frame to determine a standard morphological feature image, specifically including: According to the morphological tension values ​​of the main cell area images of each frame obtained and combined with the statistical averaging algorithm, the mean morphological tension value is obtained; the morphological tension values ​​of the main cell area images of each frame are respectively calculated with the morphological tension mean value to obtain the morphological tension difference of the main cell area images of each frame; after feature extraction, the main cell area image corresponding to the minimum morphological tension difference is selected as the standard morphological feature image.

6. The single-cell printing device control method based on convolutional neural network optimization according to claim 5, characterized in that: The specific steps of S3 include: S31. Based on the determined standard morphological feature image and in combination with the pixel response parameter set, extracting the structural response value of each pixel in the standard morphological feature image, performing an extreme value search operation, and performing a difference calculation between the maximum structural response value and the minimum structural response value in the obtained standard morphological feature image to determine the structural response amplitude of the standard morphological feature image; S32. Based on the standard morphological feature image and in combination with the pixel mutation parameter set, extract the edge local mutation coefficient of each pixel in the standard morphological feature image, compare the edge local mutation coefficient of each pixel with a preset mutation threshold, screen out corresponding pixels whose edge local mutation coefficient exceeds the mutation threshold, mark them as edge pixels, and record the grayscale value of each edge pixel; S33. After performing a second-order Laplace transform on the grayscale value of each edge pixel, obtain the second-order transform value of each edge pixel, and determine the boundary sharpening index of the standard morphological feature image by combining a statistical averaging algorithm.

7. The single-cell printing device control method based on convolutional neural network optimization according to claim 6, characterized in that: S34. Correlate the morphological tension value, structural response amplitude, and boundary sharpness index of the standard morphological feature image. After dimensionless processing, analyze the structural integrity of the target single cell contained in the standard morphological feature image to determine the cell integrity index, which is specifically: Where, represents the cell integrity index, Indicates the morphological tension value, represents the structural response amplitude, Indicates the edge sharpness index.

8. The single-cell printing device control method based on convolutional neural network optimization according to claim 7, characterized in that: The specific steps of S4 include: S41. Compare and analyze the cell integrity index with the preset integrity threshold to determine whether the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, and issue a corresponding level of printing control instruction, specifically including: If the cell integrity index exceeds the integrity threshold, it means that the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, and a first-level printing control instruction is issued; If the cell integrity index does not exceed the integrity threshold, it means that the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, and a secondary printing control instruction is issued.

9. The single-cell printing device control method based on convolutional neural network optimization according to claim 8, characterized in that: S42: Execute corresponding print control operations according to the received primary print control instructions and secondary print control instructions, specifically including: When a first-level printing control instruction is received, the execution content is as follows: at this time, it indicates that the target single cell structure in the cell suspension is not damaged, and the nozzle structure is adjusted by the piezoelectric microvalve to perform a jet printing operation on the cell suspension containing the intact target single cell; When a secondary printing control instruction is received, the execution content is: this indicates that the target single cell structure in the cell suspension is damaged, so the cell suspension is replaced and the S1-S4 process is repeated until a primary printing control instruction is issued, and the nozzle structure is adjusted by the piezoelectric microvalve to perform a jet printing operation on the cell suspension containing the complete target single cell.

10. A single-cell printing device control system based on convolutional neural network optimization, used to implement the single-cell printing device control method based on convolutional neural network optimization according to any one of claims 1 to 9, characterized in that: It includes pixel parameter acquisition module, feature screening module, structure analysis module and printing control module; The pixel parameter acquisition module is used to continuously acquire cell suspension images containing target single cells before the cell suspension is jet-printed, so as to determine the main cell area image of each frame. After inputting the images into the shallow convolutional network, the pixel grayscale parameter set, pixel response parameter set, and pixel mutation parameter set are constructed respectively. The feature screening module is used to analyze the degree of morphological fluctuation of the main area image of each frame of the cell based on the pixel grayscale parameter set, the pixel response parameter set and the pixel mutation parameter set to screen out the standard morphological feature image; The structure analysis module is used to analyze the integrity of the target single cell structure contained in the standard morphological feature image based on the screened standard morphological feature image, and determine the cell integrity index; The printing control module is used to compare and analyze the cell integrity index to determine whether the integrity of the target single cell structure in the current standard morphological feature image meets the printing requirements, so as to issue and execute the corresponding level of printing control instructions.

Images