Cell culture management system based on artificial intelligence

Abstract

The invention discloses a cell culture management system based on artificial intelligence, and the system comprises a data acquisition module which is used for dividing an incubator into regions, and collecting environment parameter data information of a target cell culture dish in each region and metabolism data information of cells in the target cell culture dish; the environment data analysis module is used for analyzing the environment parameter data information and the metabolism data information of the cells to obtain a first analysis result; the cell image analysis module is used for acquiring microscopic cell image data and analyzing the microscopic cell image data to obtain a second analysis result; the abnormity early warning module is used for judging whether the target cell culture dish is abnormal or not according to the first analysis result and the second analysis result, and if yes, early warning is carried out on the abnormal target culture dish; by combining the environment analysis result and the image analysis result, the condition of the cells in the cell culture dish can be accurately judged, the abnormal condition can be early warned in time, and cross infection of normal and abnormal cell culture dishes is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cell culture management, in particular to a cell culture management system based on artificial intelligence. BACKGROUND

[0002] Cell culture refers to the continuous growth, proliferation and maintenance of in vitro cells (such as animal cells, plant cells and microorganisms) in a simulated in vivo physiological environment. It is widely used in the fields of biological medicine and scientific research.

[0003] In the cell culture process, the cells to be cultured are usually placed in culture dishes for culture. When the number of culture dishes is too large, it is difficult to find abnormal conditions of the culture dishes, which may lead to cross infection of the culture dishes. SUMMARY

[0004] In view of the above prior art, the present application provides a cell culture management system based on artificial intelligence, which mainly solves the technical problems existing in the background art.

[0005] To achieve the above purpose, the technical scheme of the embodiment of the present application is as follows: A cell culture management system based on artificial intelligence, comprising: A data acquisition module is used to divide the incubator into regions, and collect environmental parameter data information of target cell culture dishes and metabolic data information of cells in the target cell culture dishes in each region; An environmental data analysis module is used to analyze the environmental parameter data information and the metabolic data information of the cells to obtain a first analysis result; A cell image analysis module is used to obtain microscopic cell image data, analyze the microscopic cell image data, and obtain a second analysis result; An abnormality warning module is used to determine whether the target cell culture dish is abnormal according to the first analysis result and the second analysis result, and to warn the abnormal target culture dish if it is abnormal.

[0006] Optionally, the collection of environmental parameter data information of target cell culture dishes and metabolic data information of cells in the target cell culture dishes in each region comprises: Obtaining cell basic information and culture medium information, generating an RFID tag, and pasting the RFID tag on the outer wall of the target cell culture dish; Obtaining a unique identifier through the RFID tag, binding the identifier with a monitoring unit, and collecting environmental parameter data information and cell metabolic data information in real time.

[0007] Optionally, the analysis of the environmental parameter data information and the metabolic data information of the cells obtains a first analysis result, comprising: Performing static and dynamic joint correlation analysis on the environmental parameter data information and the metabolic data information of the cells obtains a candidate parameter pair. According to the candidate parameter pair, an environmental feature value is obtained. Setting an environmental dynamic threshold, comparing the environmental feature value with the environmental dynamic threshold obtains a comparison result, and the comparison result is taken as the first analysis result.

[0008] Optionally, the static and dynamic joint correlation analysis on the environmental parameter data information and the metabolic data information of the cells obtains a candidate parameter pair, comprising: The environmental parameter data information includes time-series temperature data, time-series humidity data, time-series Concentration data, time-series Concentration data, time-series pH value data. The metabolic data information of the cells includes glucose consumption rate data and lactic acid production rate data. Calculate the Pearson correlation coefficient between each pair of features in the environmental parameter data information, and calculate the Pearson correlation coefficient between the environmental parameter data information and the metabolic data information of the cells to obtain a first calculation result. Calculate the Spearman rank correlation coefficient between each pair of features in the environmental parameter data information, and calculate the Spearman rank correlation coefficient between the environmental parameter data information and the metabolic data information of the cells to obtain a second calculation result. According to the first calculation result and the second calculation result, a heat map matrix is generated, and a strong correlation pair of environment-metabolism is screened according to the heat map matrix to obtain a joint candidate pair. Performing rolling window dynamic analysis on the joint candidate pair, calculating the dynamic correlation coefficient, capturing the time-series change of the environmental parameter data information and the metabolic parameter, and screening to obtain a candidate parameter pair. The candidate parameter pair includes a linear correlation pair and a nonlinear correlation pair.

[0009] Optionally, according to the candidate parameter pair, an environmental feature value is obtained, comprising: Input the original time-series data of the linear candidate parameter pair verified dynamically into a vector autoregressive model, perform Granger causality test, and output a P-value matrix. Input the original time-series data of the nonlinear candidate parameter pair verified dynamically into DTW, perform nonlinear similarity calculation and lag correlation identification, and obtain a lag feature matrix. constructing a PLSR model, adjusting a weight coefficient matrix of the PLSR according to the P value matrix and the lag feature matrix, training the PLSR model to obtain a trained PLSR model; calculating the environmental feature value by using the trained PLSR model.

[0010] Optionally, the analysis of the microscopic cell image data to obtain a second analysis result comprises: converting the microscopic cell image into a microscopic cell grayscale image, processing the microscopic cell grayscale image to obtain a cell morphology judgment result; converting the microscopic cell image into a microscopic cell RGB image, processing the microscopic cell RGB image to obtain a cell color judgment result; obtaining a second analysis result according to the cell morphology judgment result and the cell color judgment result.

[0011] Optionally, the processing of the microscopic cell grayscale image to obtain a cell morphology judgment result comprises: processing the microscopic cell grayscale image to obtain a connected domain of each cell in the microscopic cell grayscale image, performing edge processing on the connected domain to obtain all edge pixel points of the connected domain; arbitrarily combining the edge pixel points in pairs to obtain a plurality of edge pixel pairs, for each edge pixel pair, calculating the Euclidean distance between the two edge pixel points in the edge pixel pair, and taking the edge pixel pair with the largest Euclidean distance as a target pixel pair; calculating the distance between the two edge pixel points in the target pixel pair to obtain a first distance, and finding a center point according to the first distance; calculating the distance between the center point and all edge pixel points to obtain a second distance data set, setting a distance threshold, comparing the distance threshold with all second distances in the second distance data set to obtain a cell morphology judgment result.

[0012] Optionally, the processing of the microscopic cell RGB image to obtain a cell color judgment result comprises: converting the microscopic cell RGB image into an HSV microscopic cell image; segmenting the HSV microscopic cell image to obtain a plurality of sub-regions of the HSV microscopic cell image, calculating an average HSV value for each sub-region; comparing the average HSV value with a preset HSV value to obtain a cell color judgment result.

[0013] Optionally, the method further comprises: judging whether the target cell culture dish is abnormal according to the first analysis result and the second analysis result, and if abnormal, warning the abnormal target culture dish, comprising: weighting the first analysis result and the second analysis result to obtain a third analysis result; if the third analysis result is higher than a preset threshold, judging that the target cell culture dish is abnormal, marking the RFID tag of the abnormal target culture dish and warning.

[0014] The cell culture management system based on artificial intelligence provided by the application has the advantages that: after the data acquisition module acquires the environmental parameter data information of the target cell culture dish and the metabolic data information of the cells in the target cell culture dish, the environmental data analysis module is used for data analysis to obtain a first analysis result, the cell image analysis module is used for analysis of the microscopic cell image to obtain a second analysis result, the environmental-metabolic analysis result and the image analysis result are combined to judge whether the cells in the target cell culture dish are in a normal state, and if abnormal, a warning is given; the environmental analysis result and the image analysis result are combined to accurately judge the cell condition in the cell culture dish and timely warn of abnormal conditions, so that the cell culture dish with abnormal cells can be found in time, and cross infection of the abnormal cell culture dish and the normal cell culture dish is avoided. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 FIG. 1 is a structural schematic diagram of the cell culture management system based on artificial intelligence in the embodiments of the application. DETAILED DESCRIPTION

[0016] The technical solutions of the application are further described in detail below in combination with the drawings and specific embodiments. Unless otherwise defined, all the technical and scientific terms used herein have the same meanings as those commonly understood by the person skilled in the art to which the application belongs. The terms used in the specification of the application are only for the purpose of describing the specific embodiments and are not intended to limit the application. In the following description, the expression “some embodiments” describes a subset of all possible embodiments, but it should be understood that “some embodiments” can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict.

[0017] In the following description, a large number of specific details are given in order to provide a more thorough understanding of the application. However, it is obvious to the person skilled in the art that the application can be implemented without one or more of these details. In other cases, some technical features known in the art are not described in order not to obscure the application.

[0018] It is to be understood that the application can assume various alternative orientations and step sequences, except where expressly imposed by legal requirements or implied to exclude equivalent technical or legal solutions. It is also to be understood that the following description of the embodiments along with the accompanying drawings is provided for the purpose of demonstrating the technical solutions of the present application and is not intended to limit the present application. The terminology used herein is for the purpose of describing specific embodiments only and is not intended to limit the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising", when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. As used herein, the term "and / or" includes any and all combinations of associated items.

[0019] It is also to be noted that, as used in this specification and the appended claims, the singular forms "a", "an", and "the" include plural referents unless the content clearly dictates otherwise. Thus, for example, reference to "a component" includes a combination of two or more components, and the like.

[0020] In order to thoroughly understand the present application, detailed structures will be presented in the following description in order to illustrate the technical solutions proposed by the present application. The optional embodiments of the present application are described in detail as follows, however, in addition to these detailed descriptions, the present application can have other embodiments.

[0021] Embodiments Please refer to the accompanying drawings Figure 1 The application provides a cell culture management system based on artificial intelligence, comprising: A data acquisition module is configured to divide a culture box into regions, and collect environmental parameter data information of a target cell culture dish and metabolic data information of cells in the target cell culture dish in each region. An environmental data analysis module is configured to analyze the environmental parameter data information and the metabolic data information of the cells to obtain a first analysis result. A cell image analysis module is configured to acquire microscopic cell image data, and analyze the microscopic cell image data to obtain a second analysis result. An abnormality warning module is configured to determine whether the target cell culture dish is abnormal according to the first analysis result and the second analysis result, and to issue a warning for the abnormal target culture dish if the target culture dish is abnormal.

[0022] Specifically, a plurality of cell culture dishes are usually placed in the incubator, and each cell culture dish is cultured with one type of cells. In order to facilitate management, the incubator is divided into a plurality of areas, each area has the same environmental parameters, and a certain number of cell culture dishes are placed in each area. Since the environmental parameters of different cells may be the same or different, the cells cultured in the cell culture dishes in each area may be of the same type or different types. For example, fibroblasts, epithelioid cells and lymphoblast-like cells are different cell types. Fibroblasts, epithelioid cells and lymphoblasts are placed in different cell culture dishes, and the cell culture dishes are placed in the same area in the incubator. The cell environment is crucial to cell growth, and the cell environment parameters can directly affect cell metabolism, proliferation, etc. Different types of cells may have the same or different environmental parameters. By obtaining the environmental parameter data information of the target cell culture dish and the metabolic data information of the cells in the target cell culture dish, the environmental data and the cell metabolic data are monitored and analyzed, and the monitoring and analysis results can provide a basis for judging whether the cell culture is abnormal. Under the observation of the microscopic cell image, normal cells and abnormal cells may be different in morphology, results, quantity, etc. Through the cell image analysis module, the cell image data is analyzed in different dimensions to obtain the cell image analysis result, and whether the cell is abnormal is judged. The abnormality of the environmental data does not necessarily lead to the pollution of the cell environment and the abnormal growth of the cells. Therefore, by combining the environmental analysis result and the analysis result of the cell image, a more accurate judgment result is obtained. If the cell state is abnormal, a warning is given.

[0023] As an optional implementation, the collection of the environmental parameter data information of the target cell culture dish in each area and the metabolic data information of the cells in the target cell culture dish comprises: Obtain cell basic information and culture medium information, generate an RFID tag, and paste the RFID tag on the outer side wall of the target cell culture dish. Obtain a unique identifier through the RFID tag, bind the identifier with a monitoring unit, and collect environmental parameter data information and cell metabolic data information in real time.

[0024] Specifically, the cell basic information includes cell type, generation number and initial inoculation density, and the culture medium information includes culture medium composition. An RFID electronic tag is assigned to each cell culture dish, and the corresponding RFID electronic tag is pasted on the side wall of the corresponding cell culture dish. The RFID tag stores unique ID number, cell type, culture medium batch, operation date and other basic data information, which facilitates subsequent management. By generating and pasting the RFID tag, a unique digital identity is given to each target cell culture dish, and the target cell culture dish is associated with the corresponding cultured cell type, so that in the process of cell culture management, if an abnormal cell culture condition occurs, the target culture dish can be quickly identified and located; since each RFID tag has a unique identifier, the identifier is bound to the cell monitoring unit, and the cell metabolic data information and environmental parameter data information of the cultured cells in the target culture dish corresponding to the identifier can be obtained through the cell monitoring unit, which facilitates subsequent identification of cell abnormalities.

[0025] As an optional implementation, the analysis of the environmental parameter data information and the metabolic data information of the cells to obtain a first analysis result comprises: performing static and dynamic joint correlation analysis on the environmental parameter data information and the metabolic data information of the cells to obtain a candidate parameter pair; obtaining an environmental feature value according to the candidate parameter pair; setting an environmental dynamic threshold, comparing the environmental feature value with the environmental dynamic threshold to obtain a comparison result, and taking the comparison result as the first analysis result.

[0026] As an optional implementation, the static and dynamic joint correlation analysis on the environmental parameter data information and the metabolic data information of the cells to obtain a candidate parameter pair comprises: the environmental parameter data information comprises time-series temperature data, time-series humidity data, time-series concentration data, time-series concentration data, and time-series pH value data; the metabolic data information of the cells comprises glucose consumption rate data and lactic acid production rate data; calculating the Pearson correlation coefficients of the features in the environmental parameter data information two by two, and calculating the Pearson correlation coefficients between the environmental parameter data information and the metabolic data information of the cells to obtain a first calculation result; calculating the Spearman rank correlation coefficients of the features in the environmental parameter data information two by two, and calculating the Spearman rank correlation coefficients between the environmental parameter data information and the metabolic data information of the cells to obtain a second calculation result; generating a heat map matrix according to the first calculation result and the second calculation result, screening an environment-metabolism strong correlation pair according to the heat map matrix, and obtaining a joint candidate pair; performing rolling window dynamic analysis on the joint candidate pair, calculating a dynamic correlation coefficient, capturing the time-series changes of the environmental parameter data information and the metabolic parameter, and screening a candidate parameter pair; The candidate parameter pairs include linearly related pairs and nonlinearly related pairs.

[0027] Specifically, the metabolic parameters of the cells can directly reflect the metabolic state of the cells in the cell culture process, and the environmental parameter data information is a key intermediate variable affecting the cell state; first, after obtaining the metabolic data information of the cells and the environmental parameter data information in the same time period, the metabolic data information of the cells and the environmental parameter data information need to be time-aligned, and the metabolic data information of the cells and the environmental parameter data information are standardized; the Pearson correlation coefficient and the Spearman rank correlation coefficient are calculated for the standardized metabolic data information of the cells and the environmental parameter data information. The Pearson correlation coefficient can measure the linear correlation between two variables, and the range is between -1 and 1. The closer the calculated Pearson correlation coefficient is to 1, the greater the correlation between the two variables. By calculating the Pearson correlation coefficient between each pair of features in the environmental parameter data information, the correlation between different environmental parameter data information is mined, the linear relationship between the environmental parameter data information is evaluated, and it is identified which environmental parameter data information has significant correlation; and the Pearson correlation coefficient between the environmental parameter data information and the metabolic data information of the cells is calculated, from which the influence of different environmental parameter data information on cell metabolism is mined, and it is determined which environmental parameter data information has a linear relationship with the metabolic data information of the cells, to obtain a first calculation result; The Spearman rank correlation coefficient can capture the nonlinear relationship between different parameter pairs. By calculating the Spearman rank correlation coefficient between each pair of features in the environmental parameter data information, and calculating the Spearman rank correlation coefficient between the environmental parameter data information and the metabolic data information of the cells, a second calculation result is obtained. Based on the first calculation result and the second calculation result, a joint heat map matrix is generated, which includes a heat map matrix of linearly related pairs obtained from the first calculation result and a heat map matrix of nonlinearly related pairs obtained from the second calculation result. The joint heat map matrix displays the correlation coefficients in the form of color depth. Dark color area represents high correlation, and light color represents low correlation. The correlation strength between different parameter pairs can be intuitively compared. Therefore, the joint heat map matrix can identify the joint candidate pairs of environmental parameters and metabolic parameters and the joint candidate pairs between environmental parameters. According to the multiple joint candidate pairs, an environmental-metabolic joint candidate list is generated. The multiple joint candidate pairs in the environmental-metabolic joint candidate list are verified for time sequence correlation. The time sequence correlation verification specifically includes: designing a rolling time window to calculate a dynamic correlation coefficient for the joint candidate pairs in the environmental-metabolic joint candidate list, and capturing the time sequence change characteristics. The rolling window analysis is to move a fixed size window on the time series data, calculate the correlation coefficient in each window, and capture the timing changes between parameters; for example, the relationship between temperature and metabolic products may be different at different time periods, and the rolling window analysis can reveal the trend of changes in this dynamic relationship; according to the results of the rolling window dynamic analysis, the candidate parameter pairs are screened out; The calculation of Pearson correlation coefficient and Spearman rank correlation coefficient is a static analysis of data, while the rolling window analysis of joint candidate pairs is a dynamic time series analysis of data. Combining static analysis and dynamic time series analysis can better capture the change trend of environment and metabolism in time series. Through static analysis and dynamic time series analysis, candidate parameter pairs with higher correlation are screened out.

[0028] As an optional implementation, the environment feature value is obtained according to the candidate parameter pair, including: The original time series data of the linear candidate parameter pair that has passed the dynamic verification is input into a vector autoregressive model, Granger causality test is performed, and a P value matrix is output; The original time series data of the nonlinear candidate parameter pair that has passed the dynamic verification is input into DTW, nonlinear similarity calculation and lag correlation identification are performed, and a lag feature matrix is obtained; A PLSR model is constructed, the weight coefficient matrix of the PLSR is adjusted according to the P value matrix and the lag feature matrix, the PLSR model is trained, and a trained PLSR model is obtained; The environment feature value is calculated by using the trained PLSR model.

[0029] Specifically, the direct input of Granger test is the original time series data, not the "linear correlation pair" itself. The linear correlation pair (such as temperature and pH) is a pre-screening result, which is used to determine which variable pairs need to enter the Granger test process.

[0030] The variables with linear correlation are screened out by combining static analysis and dynamic analysis, which indicates that the variables have high correlation. For example, after the linear candidate parameter pair of temperature and pH is screened out, the original time series temperature data and the original time series pH value data are input into the vector autoregressive model to test whether the lag term of temperature can significantly predict the current value of pH, i.e., whether the past value of temperature can effectively predict the current value of pH. During the cell growth process, the change of one parameter in the environmental parameter data information not only affects the cell metabolism, but also may affect other environmental parameters. For example, the increase of temperature in the environmental parameters will enhance the enzyme activity and accelerate the metabolic rate, thereby causing the accumulation of acidic metabolites (such as lactic acid) and further causing the decrease of pH value. In this temperature and pH change process, the temperature change precedes the pH change, i.e., the lag term of temperature can significantly predict the current value of pH. Therefore, by Granger causality test, it is determined that the lag term of temperature significantly predicts the current value of pH, which can guide the PLSR model to give higher weight to temperature and improve the prediction accuracy of the cell metabolic state. The P-value matrix can identify the causality direction and significance between variables. If it is determined that the change of temperature is the Granger cause of the change of pH, it has statistical significance. The vector autoregressive model is used for multivariate time series analysis and can capture the dynamic relationship of multivariate time series. By using the vector autoregressive model to analyze the screened candidate parameter pairs, the dynamic time series interaction relationship between environmental parameters and between environmental parameters and metabolic parameters is captured, and the Granger test is performed based on the vector autoregressive model coefficients to confirm the causality direction and guide the variable weight distribution of the PLSR model. The Granger test uses the results of the VAR model to determine whether the environmental parameters have predictive ability for the metabolic parameters, thereby determining the causality. The result of this step will guide the weight distribution of the variables in the PLSR model to ensure that the model focuses on the variables with significant causality.

[0031] The similarity of the original time series data of the nonlinear candidate parameter pairs that have passed dynamic verification is calculated by DTW. The smaller the distance is, the more similar it is. The lag feature matrix of DTW is output. The P-value matrix of the VAR model Granger test and the lag feature matrix are horizontally spliced to form the input feature matrix X. The metabolic parameters are used as the target variable Y for supervised PLSR model training. PLSR projects X to a low-dimensional latent space by iterative solution, maximizes the covariance of X and Y, and adjusts the weight coefficients of PLSR by the P-value matrix and the lag feature matrix. The PLSR model is trained until a trained PLSR model is obtained. For real-time environmental parameter data information, the environmental parameter data information is projected to a low-dimensional space by the trained PLSR model, and environmental characteristic values are output. The environmental characteristic values quantify the overall influence intensity of the environmental parameters on the metabolic parameters. A preset environmental dynamic threshold is compared with the environmental characteristic values, and the comparison result is taken as the first analysis result.

[0032] As an optional implementation, the analysis of the microscopic cell image data to obtain the second analysis result comprises: The microscopic cell image is converted into a microscopic cell grayscale image, and the microscopic cell grayscale image is processed to obtain a cell morphology judgment result. The microscopic cell image is converted into a microscopic cell RGB image, and the microscopic cell RGB image is processed to obtain a cell color judgment result. The second analysis result is obtained according to the cell morphology judgment result and the cell color judgment result.

[0033] As an optional implementation, the processing of the microscopic cell grayscale image to obtain the cell morphology judgment result comprises: The microscopic cell grayscale image is processed to obtain a connected domain of each cell in the microscopic cell grayscale image, and edge processing is performed on the connected domain to obtain all edge pixel points of the connected domain. The edge pixel points are combined in any two-by-two combination to obtain a plurality of edge pixel pairs. For each edge pixel pair, the Euclidean distance between the two edge pixel points in the edge pixel pair is calculated, and the edge pixel pair with the maximum Euclidean distance is taken as a target pixel pair. The distance between the two edge pixel points in the target pixel pair is calculated to obtain a first distance, and a center point is found according to the first distance. The distances between the center point and all edge pixel points are calculated to obtain a second distance data set. A distance threshold is set, and the distance threshold is compared with all second distances in the second distance data set to obtain the cell morphology judgment result.

[0034] Specifically, the microscopic cell image can be obtained by industrial camera cooperating with image acquisition software (such as OpenCV) to realize real-time image capture and format conversion. During the cell culture process, if the culture environment of the cells is contaminated, abnormal morphology and color change of the cells may occur. After converting the microscopic cell image to a grayscale image, a thresholding method is applied to the grayscale image to obtain the connected components of each cell. Then, the Canny edge detection algorithm is used to obtain all edge pixels of each connected component. The thresholding method and the Canny edge detection algorithm are existing technologies and will not be elaborated upon here. All edge pixels are then paired to calculate the Euclidean distance. The smaller the Euclidean distance, the closer the two edge pixels are in space; the larger the Euclidean distance, the farther apart the two edge pixels are in space. Therefore, the edge pixel pair with the largest Euclidean distance is selected as the target pixel pair, i.e., the two farthest pixels are obtained. Then, the target... The distance between two edge pixels in a pixel pair is calculated, and its center point is also the center point of the cell. Then, the distance from the center point to all edge pixels is calculated to quantify the uniformity of cell shape and determine whether it is close to a circle or other regular shape. By using a preset distance threshold, the distance from the center point to all edge pixels is compared with the distance threshold. The cell morphology is different for different cells and different growth stages, so the distance threshold can be set according to the actual cell type and growth stage. For round cells, one distance threshold can be set, while for elliptical or irregular shapes, multiple thresholds can be set. Finally, the cell morphology judgment result is obtained based on the comparison result.

[0035] As an optional implementation, the processing of the RGB image of the microscopic cells to obtain the cell color determination result includes: Convert the RGB image of the microscopic cells into an HSV image of the microscopic cells; The HSV microscopic cell image is segmented to obtain multiple sub-regions of the HSV microscopic cell image, and the average HSV value is calculated for each sub-region. The average HSV value is compared with the preset HSV value to obtain the cell color judgment result.

[0036] Specifically, the color features of the RGB image of microscopic cells are decoupled from the three channels of "red (R), green (G), blue (B)" into three independent dimensions of "hue (H), saturation (S), and brightness (V)" to obtain the HSV microscopic cell image. The HSV pixels are clustered using the K-means algorithm, and the HSV microscopic cell image is segmented according to the clustering results to obtain multiple sub-regions. The average HSV value is calculated for each sub-region. The average HSV value of each sub-region is the sum of the HSV values ​​of each pixel in that region and divided by the number of pixels in that image region. The preset HSV value is a preset HSV threshold range. The average HSV value is compared with the preset HSV value to obtain the cell color judgment result.

[0037] According to the judgment result of the cell morphology and the judgment result of the cell color, a second analysis result is obtained, specifically, a final result is calculated according to the judgment result of the cell morphology and the judgment result of the cell color by weighting, whether the cell in the target cell culture dish is abnormal is judged based on the calculated result, and the result is taken as the second analysis result.

[0038] As an optional embodiment, the method further comprises the following steps: The first analysis result and the second analysis result are allocated with weights, and a third analysis result is calculated; If the third analysis result is higher than a preset threshold, it is judged that the target cell culture dish is abnormal, and the RFID tag of the abnormal target culture dish is marked and warned.

[0039] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. The protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A cell culture management system based on artificial intelligence, characterized in that, include: The data acquisition module is used to divide the incubator into regions and collect environmental parameter data and metabolic data of the target cell culture dishes in each region. The environmental data analysis module analyzes the environmental parameter data and the metabolic data of the cells to obtain a first analysis result; The cell image analysis module is used to acquire microscopic cell image data, analyze the microscopic cell image data, and obtain a second analysis result; The anomaly warning module is used to determine whether the target cell culture dish has an anomaly based on the first analysis result and the second analysis result, and to issue an early warning for the abnormal target culture dish if an anomaly occurs.

2. The artificial intelligence-based cell culture management system according to claim 1, characterized in that, The collection of environmental parameter data and metabolic data of cells in the target cell culture dish in each region includes: Acquire basic cell information and culture medium information, generate RFID tags, and affix the RFID tags to the outer wall of the target cell culture dish; A unique identifier is obtained through the RFID tag, and the identifier is bound to the monitoring unit to collect environmental parameter data and cell metabolism data in real time.

3. The artificial intelligence-based cell culture management system according to claim 1, characterized in that, The analysis of the environmental parameter data and the cellular metabolic data to obtain a first analysis result includes: Static and dynamic joint correlation analysis was performed on the environmental parameter data and the cellular metabolic data to obtain candidate parameter pairs; Based on the candidate parameter pairs, environmental feature values ​​are obtained; Set an environmental dynamic threshold, compare the environmental feature value with the environmental dynamic threshold, obtain the comparison result, and use the comparison result as the first analysis result.

4. The artificial intelligence-based cell culture management system according to claim 3, characterized in that, The static and dynamic joint correlation analysis of the environmental parameter data and the cellular metabolic data yields candidate parameter pairs, including: The environmental parameter data includes time-series temperature data, time-series humidity data, and time-series... Concentration data, time series Concentration data, time-series pH data; The metabolic data of the cells include glucose consumption rate data and lactate production rate data; The Pearson correlation coefficients are calculated pairwise for the features in the environmental parameter data information, and the Pearson correlation coefficients between the environmental parameter data information and the metabolic data information of the cells are calculated to obtain the first calculation result; The Spearman rank correlation coefficients are calculated pairwise for the features in the environmental parameter data information, and the Spearman rank correlation coefficients between the environmental parameter data information and the metabolic data information of the cells are calculated to obtain a second calculation result; Based on the first calculation result and the second calculation result, a heatmap matrix is ​​generated, and strong correlation pairs between environment and metabolism are screened based on the heatmap matrix to obtain joint candidate pairs; A rolling window dynamic analysis is performed on the joint candidate pairs to calculate the dynamic correlation coefficient, capture the time-series changes of the environmental parameter data and the metabolic parameters, and screen out candidate parameter pairs. The candidate parameter pairs include linear correlation pairs and nonlinear correlation pairs.

5. The artificial intelligence-based cell culture management system according to claim 4, characterized in that, The step of obtaining environmental feature values ​​based on the candidate parameter pairs includes: The original time series data of the dynamically validated linear candidate parameter pairs are input into the vector autoregressive model, and Granger causality test is performed to output the p-value matrix. The original time series data of the dynamically validated nonlinear candidate parameter pairs are input into DTW to perform nonlinear similarity calculation and lag association identification, and the lag feature matrix is ​​obtained. Construct a PLSR model, adjust the weight coefficient matrix of the PLSR according to the P-value matrix and the hysteresis feature matrix, train the PLSR model, and obtain a trained PLSR model. The environmental feature values ​​are calculated using the trained PLSR model.

6. The cell culture management system based on artificial intelligence according to claim 1, characterized in that, The analysis of the microscopic cell image data to obtain a second analysis result includes: The microscopic cell image is converted into a microscopic cell grayscale image, and the microscopic cell grayscale image is processed to obtain the cell morphology judgment result. The microscopic cell image is converted into a microscopic cell RGB image, and the microscopic cell RGB image is processed to obtain the cell color judgment result; Based on the results of the cell morphology and cell color determination, a second analysis result is obtained.

7. The artificial intelligence-based cell culture management system according to claim 6, characterized in that, The process of processing the grayscale image of the microscopic cells to obtain the cell morphology determination result includes: The grayscale image of the microscopic cells is processed to obtain the connected components of each cell in the grayscale image of the microscopic cells. The connected components are then processed to obtain all edge pixels of the connected components. The edge pixels are arbitrarily paired to obtain multiple edge pixel pairs. For each edge pixel pair, the Euclidean distance between the two edge pixels in the edge pixel pair is calculated, and the edge pixel pair with the largest Euclidean distance is taken as the target pixel pair. Calculate the distance between two edge pixels in the target pixel pair to obtain a first distance, and find the center point based on the first distance; Calculate the distance between the center point and all edge pixels to obtain a second distance dataset. Set a distance threshold and compare the distance threshold with all second distances in the second distance dataset to obtain the cell morphology judgment result.

8. The artificial intelligence-based cell culture management system according to claim 6, characterized in that, The process of processing the RGB image of the microscopic cells to obtain the cell color determination result includes: Convert the RGB image of the microscopic cells into an HSV image of the microscopic cells; The HSV microscopic cell image is segmented to obtain multiple sub-regions of the HSV microscopic cell image, and the average HSV value is calculated for each sub-region. The average HSV value is compared with the preset HSV value to obtain the cell color judgment result.

9. A cell culture management system based on artificial intelligence according to claim 2, characterized in that, The step of determining whether the target cell culture dish has an abnormality based on the first analysis result and the second analysis result, and issuing an early warning for the abnormal target culture dish if an abnormality occurs, includes: The first and second analysis results are weighted and then the third analysis result is calculated. If the third analysis result is higher than the preset threshold, it is determined that the target cell culture dish is abnormal, and the RFID tag of the abnormal target culture dish is marked and an early warning is issued.

Images