CN114359899B - Cell co-culture model, cell model construction method, computer device, and storage medium - Google Patents

Abstract

The present application relates to a cell co-culture model, a method for constructing a cell model, a computer device and a storage medium. The method for constructing a cell co-culture model includes: real-time acquisition of cells in a cell co-culture process under a preset cell seeding density and cell seeding ratio Image, extract the cell characteristic parameters of each cell at different time nodes in the cell image, and mark the type and number of cells in the cell image; with the co-cultured cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time The node is the input, and the cell characteristic parameters of each cell are used as the output to train the neural network to obtain a cell co-culture model, which can observe the process of cell culture in real time and obtain cells in the process of cell co-culture quickly, effectively and repeatedly. Characteristic parameters are very important for studying the functional state of cells, especially when different types of cells are co-cultured, and the interactions between cells are studied.

Description

Cell co-culture model and cell model construction method, computer equipment and storage medium

technical field

The present application relates to the technical field of deep learning, and in particular, to a cell co-culture model, a method for constructing a cell model, a computer device and a storage medium.

Background technique

Cell co-culture in clinical biological experiments has always been a difficult problem with many limitations. Taking the co-culture of mesenchymal stem cells with other cells as an example, mesenchymal stem cells are a kind of pluripotent stem cells, which have all the commonalities of stem cells, namely self-renewal and multi-directional differentiation ability. It also has the most clinical applications, and the most sourced is bone marrow mesenchymal stem cells (BMSCs), also known as bone marrow stromal fibroblasts in the past. For a variety of interstitial tissues, such as bone, cartilage, fat, bone marrow hematopoietic tissue. The clinical application of mesenchymal stem cells to solve a variety of blood system diseases, cardiovascular diseases, liver cirrhosis, nervous system diseases, partial meniscus resection and repair of knee joints, autoimmune diseases, etc. has made major breakthroughs, and has made major breakthroughs in nervous system repair. Has long-term development prospects.

Since the specific mechanism of mesenchymal stem cells in vivo has not been fully understood, many studies in recent years have found that there are various mechanism effects of mesenchymal stem cells co-cultured with other cells. In the prior art, it has been found that the co-culture of mesenchymal stem cells and natural killer cells can enhance the killing activity of seven pairs of dendritic cells, and it has also been found that the co-culture of mesenchymal stem cells and neutrophils can prolong the lifespan of neutrophils. and maintain its biological activity and so on. Existing studies often emphasize the co-culture effect of mesenchymal stem cells on target cells, but little is known about the characteristic parameters of mesenchymal stem cell co-culture, and it is difficult to control the co-culture during the co-culture of mesenchymal stem cells and target cells. The parameters of culture conditions, and the rapid separation and subsequent analysis of different cells after co-culture are also problems that need to be solved urgently.

The existing multi-cell co-culture methods mainly include direct contact co-culture and indirect contact co-culture. Direct contact co-culture is to put two or more types of cells in the same culture system to make them in direct contact. This method is suitable for adjacent tissue cells in vivo. Since these cells can transmit the cytokines produced by means of communication connection, closed connection and anchor connection in vivo, these connection information can be retained by direct contact culture, so that the cultured cells are closer to the natural state in vivo. But its disadvantage is that it is difficult to separate the two kinds of cells after co-culture. It can only be separated by fluorescently labeling one of the cells by flow sorting, or by labeling one of the cells with immunomagnetic beads and then using a magnetic field. separation, which is very cumbersome and expensive. At the same time, it is difficult to distinguish two morphologically identical cells by general methods without the use of subsequent experimental studies. In indirect contact co-culture, two types of cells are cultured separately and contacted separately through the medium. The two cells are not in direct contact, but cytokines can communicate. The advantage is that it highlights the effect of conditioned cells on the target cells, and it is easier to separate the two types of cells. However, this method does not strictly carry out simultaneous culture, and generally only two types of cells can be co-cultured, but not more. The co-culture of cells, as there is no contact between the two types of cells, does not show that the stromal cells have a significant support function for the parenchymal cells.

At present, a variety of multi-cell co-culture models have been proposed. In the prior art, a six-well cell co-culture plate including several interconnected culture wells, pistons, and polycarbonate membranes has been proposed. Connected culture wells, through holes are set between adjacent culture wells, the through holes can be sealed by pistons, and there is a handle on the pistons for easy loading and unloading. When using, according to the experimental needs, one or more pistons can be removed to make the culture holes. It can communicate with each other to achieve the intercommunication of the culture medium. The disadvantage is that it is one-time, and the cells cannot be separated after mixing.

Rapid and effective observation of cell behavior is also an urgent technical problem to be solved in the co-culture process. At present, the methods of continuous observation of cells for a long time mainly include visual inspection of microscopes. The disadvantage is that the time is long, and the state can only be observed at a specific time point; indirect observation of metabolites is carried out, and a specific chemical substance MTT is added to the cells. The cell state is obtained indirectly by detecting MTT, but the disadvantage is that it will interfere with the normal growth of cells, and it is not sensitive to real-time detection of the amplification process; and the use of real-time sensor observation, including resistance sensors and thermal sensors, has the disadvantage that it may damage cells, and the error is relatively small. large and susceptible to cellular electrochemical processes. However, the state observation of cells is often transient, and the optimal observation time is often missed during the experiment.

In addition, in the actual multi-cell culture process, it is necessary to observe or obtain the growth state and morphological regularity of the cells in the culture dish by moving the culture dish. However, frequently changing the state of the culture dish may affect the survival and development of the cells in the culture dish. . In addition, at present, the main way to distinguish different cells is to use morphological methods to detect, or to use double-labeled immunohistochemical techniques or in situ hybridization techniques, or to use different fluorescent labels, all of which involve direct manipulation of cells. It will affect the natural growth of cells and is not conducive to observing the actual growth law of cells.

There have been many studies on the rapid observation in the process of cell culture in the existing technology, but it is still unable to systematically and comprehensively provide real-time characteristic parameters of different cells in the process of cell co-culture, nor to realize the prediction and evaluation of the growth state of cell co-culture.

In recent years, with the rapid development of technologies such as virtual reality, biomedicine, high-performance computing and artificial intelligence, how to use information technology to empower human life functions and disease treatment, and achieve more in-depth qualitative and quantitative research and prediction, has become 21 a major scientific and technological issue of the century. Human virtual twin aims to build a living human body digital model with realistic geometric shape and behavior, and can support medical and health applications, and use it as a high-level intersection of artificial intelligence, virtual reality, big data, 5G network, biomedicine, and current diagnosis and treatment technologies. The integration and convergence platform provides important information support means for medical diagnosis and research. For human virtual twins, the construction of physiological and biochemical models of various scale units of the human body is the core, including the behavior modeling description at the micro-scale of the human body.

SUMMARY OF THE INVENTION

In order to solve the above technical problems or at least partially solve the above technical problems, the present application provides a cell co-culture model, a cell model construction method, a computer device and a storage medium.

In a first aspect, the present application provides a method for constructing a cell co-culture model, comprising the following steps:

Collect real-time cell images in the process of cell co-culture under the preset cell seeding density and cell seeding ratio, extract the cell characteristic parameters of each cell at different time nodes in the cell image, and mark the type and number of cells in the cell image;

Taking the co-cultured cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time node as input, and using the cell characteristic parameters of each cell as the output to train the neural network, the cell co-culture model is obtained.

Preferably, before the real-time acquisition of cell images in the process of co-culture of cells under the preset cell seeding density and preset cell seeding ratio, the method further includes:

According to the same co-cultivation conditions, different types of cells are co-cultured according to the preset cell seeding density and cell seeding ratio, which are used to collect cell images during the cell co-cultivation process in the next step.

Preferably, the cell images during the co-cultivation process under the preset cell seeding density and cell seeding ratio are collected in real time, and cell characteristic parameters of each cell at different time nodes in the cell image are extracted, including:

Utilize multiple microscope heads to collect real-time cell images in the process of cell co-culture under the preset cell seeding density and preset cell seeding ratio, and send the collected cell images to the projection device;

Use the projection device to deduplicate and summarize all the cell images at the same acquisition time node, and process the received cell images into three-dimensional cell images for display;

Cell characteristic parameters are extracted from the displayed three-dimensional cell images under different acquisition time nodes by using multiple visual capture devices.

Preferably, the co-cultured cell inoculation density, cell inoculation ratio, cultured cell type and cell co-culture time node are used as inputs, and the cell characteristic parameters of each cell are used as outputs to train a neural network to obtain a cell co-culture model, including:

Summarize the cell type of each cell and the cell characteristics under each cell co-culture time node under the preset co-culture cell inoculation density and cell inoculation ratio, and obtain the cell characteristic data set under each cell co-culture time node;

The cell feature data set under each cell co-culture time node is randomly divided into two parts: cell feature training set and cell feature test set;

The co-cultured cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time node in the cell characteristic training set are used as input, and the cell characteristic parameters of each cell are used as output to train the neural network to obtain the initial cell co-culture model;

The co-culture cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time node in the cell characteristic test set are used as input, and the cell characteristic parameters of each cell are used as output to test the initial cell co-culture model, which is used to adjust the initial cell co-culture model. The weight value of the culture model is used to obtain a cell co-culture model with a prediction accuracy greater than a preset threshold.

Preferably, the co-cultured cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time node in the cell characteristic test set are used as input, and the cell characteristic parameters of each cell are used as output to test the initial cell co-culture model through dynamic Time node regularization algorithm implementation.

Preferably, the cell characteristic parameters include at least one of cell membrane morphology, cell nucleus morphology, coordinates of cells relative to the center point of the co-culture dish, cell size, cell area, average cell growth rate, and the number of cell surface markers and metabolites , wherein the number of the cell surface markers and metabolites is calculated from the images of cells marked with different fluorescent substances or isotopic substances on the cells.

In a second aspect, the present application provides a method for constructing a cell model, comprising the following steps:

Obtain the growth curve and morphological characteristic data of cells at different time nodes, and determine the degree of cell differentiation according to the cell morphological characteristic data;

The cell culture characteristic parameter database is constructed according to the cell growth curve, morphological characteristic data, the degree of cell differentiation and the collected cell physical and biochemical characteristic data at different time nodes;

Use 3D modeling technology to construct a virtual culture environment including culture vessel, culture medium, and culture condition parameters, and based on the cell culture characteristic parameter database, use motion planning technology to generate virtual target cells and add them to the virtual culture environment to use rendering The engine generates frame-by-frame cell information and culture environment information, and builds a cell model under the simulated cell culture system based on the frame-by-frame cell information and culture environment information.

Preferably, when the virtual culture environment is a co-culture virtual environment, the growth curves and morphological characteristic data of the cells at different time nodes are obtained based on a pre-built cell co-culture model.

In a third aspect, the present application provides a computer device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

memory for storing computer programs;

The processor is configured to implement the steps of the above-mentioned method for constructing a cell co-culture model or the above-mentioned method for constructing a cell model when executing the program stored in the memory.

In a fourth aspect, the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned method for constructing a cell co-culture model or the above-mentioned method for constructing a cell model are realized. .

Compared with the prior art, the above-mentioned technical solutions provided in the embodiments of the present application have the following advantages:

In the method provided in the embodiment of the present application, the cell images during the co-culture process of cells under the preset cell seeding density and cell seeding ratio are collected in real time, the cell characteristic parameters of each cell at different time nodes in the cell image are extracted, and the cell image The type and number of cells are labeled; the co-culture cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time node are used as inputs, and the cell characteristic parameters of each cell are used as the output to train a neural network to obtain a cell co-culture model. , can observe the process of cell culture in real time, and can quickly, effectively and reproducibly obtain cell characteristic parameters in the process of cell co-culture. role is very important.

Description of drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description serve to explain the principles of the invention.

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. In other words, on the premise of no creative labor, other drawings can also be obtained from these drawings.

FIG. 1 is a schematic flowchart of a method for constructing a cell co-culture model provided in the embodiment of the present application;

FIG. 2 is a schematic diagram of a microscope when the mesenchymal stem cells of the embodiment of the present application are cultured alone on the 4th day;

3 is a schematic flowchart of a method for constructing a cell co-culture model provided by another embodiment of the present application;

FIG. 4 is a specific flowchart of step S1 in the embodiment of the present application;

5 is a schematic diagram of a device for culturing cells and capturing images for modeling in an embodiment of the present application;

FIG. 6 is a schematic flowchart of a specific flow of step S2 in this embodiment of the present application;

FIG. 7 is a schematic flowchart of a method for constructing a cell model according to an embodiment of the present application. As shown in Figure 7, the cell model construction method of the present application;

8 is a schematic flowchart of a method for constructing a cell co-culture model and acquiring cell state information using a virtual culture environment as a co-culture virtual environment and the co-culture cells include two types of cells as an example;

9 is a schematic structural diagram of a device for constructing a cell co-culture model and obtaining cell state information according to an embodiment of the present invention;

FIG. 10 is a schematic structural diagram of a computer device according to an embodiment of the present invention.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments It is a part of the embodiments of this application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present application.

FIG. 1 is a schematic flowchart of a method for constructing a cell co-culture model provided in an embodiment of the present application. As shown in Figure 1, the cell co-culture model construction method of the present application includes the following steps:

S1, collect the cell images in the process of co-cultivation of cells under the preset cell seeding density and cell seeding ratio in real time, extract the cell characteristic parameters of each cell in the cell image at different time nodes, and analyze the type and number of cells in the cell image. mark;

In practical applications, the cell characteristic parameters include at least one of cell membrane morphology, cell nucleus morphology, coordinates of cells relative to the center point of the co-culture dish, cell size, cell area, average cell growth rate, and the number of cell surface markers and metabolites One, wherein the amounts of the cell surface markers and metabolites are calculated from cell images marked with different fluorescent substances or isotopic substances on the cells.

In one embodiment, the co-cultured cells are umbilical cord mesenchymal stem cells and brain tumor stem cells, respectively, and are resuspended in growth factor-free SFM medium to prepare a single cell suspension with a concentration of 10 ⁶ cells/ml for later use. The rest of the operations were the same as those in Example 1. At the same time, part of the co-cultured cells were taken on the 2nd, 4th, 6th, 8th, and 10th days of culture, and CD133 (brain tumor stem cell marker) and CD29 (mesenchymal stem cell marker) were measured by flow cytometry. Marker) expression, wherein the schematic diagram of the microscope when the mesenchymal stem cells were cultured alone on the 4th day is shown in Figure 2.

In the above steps, in addition to brain tumor stem cells, the target cells can also be other types of stem cells, such as epidermal stem cells, hematopoietic stem cells or adipose stem cells, etc., and can also be conventional differentiated cells, such as fibroblasts, epithelial cells, etc.

In the above cell co-culture, the amount of target cells added can be adjusted as needed, for example, the ratio of mesenchymal stem cells:target cells can be 1:5-5:1.

The cell capture camera used to collect cell images can be an ordinary optical camera, a fluorescence imaging camera, or an electron microscope imaging camera. The above two types of cells can be marked with different fluorescent substances or marked with isotopic substances. The above cell capture camera can be moved and Rotate and capture cell images at different angles, and analyze and process the cell images to obtain cell morphological structures for display or storage.

S2, take the co-cultured cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time node as the input, and use the cell characteristic parameters of each cell as the output to train the neural network to obtain the cell co-culture model.

FIG. 3 is a schematic flowchart of a method for constructing a cell co-culture model according to another embodiment of the present application. As shown in Figure 3, the cell co-culture model construction method of the present application, in addition to steps S1 and S2, also includes the following steps:

S31 , according to the same co-cultivation conditions, cell co-cultivation is performed on different types of cells according to the preset cell seeding density and cell seeding ratio, which is used for collecting cell images during the cell co-cultivation process in the next step.

In practical applications, taking the co-culture of bone marrow mesenchymal stem cells and neural stem cells as an example, according to the same co-culture conditions, different types of cells are co-cultured according to the preset cell seeding density and cell seeding ratio, including the following steps:

The first step, recovery and passage of BMSCs: Take out the frozen mesenchymal stem cells, resuspend them in 10ml L-DMEM complete medium, inoculate on a 10mm cell culture dish, place in 5% CO ₂ , 37 ℃ in a constant temperature incubator. The medium was changed every other day, washed three times with sterile PBS solution, and the medium was changed every 2-3 days thereafter. Cells are passaged when the confluency reaches 75%-85%;

The second step, the recovery and passage of neural stem cells: take the cryopreserved neural stem cells and centrifuge briefly to remove the supernatant, add 2 ml of serum-free neural stem cell complete medium to resuspend and adjust the cell density to 105 cells/ml, inoculate in a T25 cell culture flask, and put it in a T25 cell culture flask. Culture in a constant temperature incubator, change the medium every 3 days, and passage when the cell confluence reaches 75%-85%;

The third step, direct co-cultivation: the third-generation mesenchymal stem cells were cultured at (2×10 ⁵ /ml, 4×10 ⁵ /ml, 6×10 ⁵ /ml, 8×10 ⁵ /ml and 10×10 The density of ⁵ /ml) was evenly inoculated into a 10cm transparent sterile petri dish (1), and then the second generation neural stem cells in the logarithmic growth phase were inoculated (the inoculation ratios were 5:1, 2:1, 1:1, 1:2, 1:5), after that, the medium was changed every 3 days, co-cultured for 10 days, and 5 replicate co-culture dishes were made for each group of density and inoculation ratio.

FIG. 4 is a schematic schematic diagram of a specific flow of step S1 in this embodiment of the present application. As shown in FIG. 4 , the cell images during the co-culture process under the preset cell seeding density and cell seeding ratio are collected in real time, and the cell characteristic parameters of each cell at different time nodes in the cell image are extracted, including:

S41, using a plurality of microscope heads to collect in real time the cell images during the co-cultivation of cells under the preset cell seeding density and the preset cell seeding ratio, and send the collected cell images to the projection device;

S42, using the projection device to deduplicate and summarize all the cell images at the same acquisition time node, and process the received cell images into three-dimensional cell images for display;

S43, using a plurality of visual capture devices to respectively extract cell characteristic parameters from the displayed three-dimensional cell images under different acquisition time nodes.

Referring to Fig. 5, the device (5) for cell culture and capturing images for modeling shows the connection relationship between multiple microscopes, projection devices and multiple visual capture devices. Further, the device as shown in Fig. 5 is used to observe Co-culture cells and capture cell images:

Observation: From day 0, open the opening (6) of the 37°C constant temperature incubator, and place the co-culture dish (1) on the placing platform (2). In a fixed position, cells A (a) and B (b) are cultured in a co-culture dish, the multiple microscope heads (3) of the adjustment platform are focused, and the microscope objective lens is placed on an adjustable and movable plate (4), which is built into the plate The eyepiece of the microscope (3) is connected with the projection device (8), and the images observed by a plurality of microscopes are integrated in real time and projected in the projection device;

Image capture: The projection device deduplicates and summarizes all cell images at the same time point, and the cell images recorded by multiple microscope heads are reprogrammed into an enlarged three-dimensional cell image in the actual co-culture dish. The device (9) of a plurality of visual capture cameras (10a, 10b, 10c, 10d, 10e) collects the three-dimensional cell image in real time, and extracts characteristic parameters in the cell image respectively, and the characteristic parameters include cell membrane, cell nucleus, cell relative common The coordinates of the center point of the culture dish, the size and area of the cells, the average growth rate of the cells, etc., and all the mesenchymal stem cells and neural stem cells are digitally labeled respectively, with consecutive numbers A1-An for mesenchymal stem cells, B1-Bn for neural stem cells , the characteristic parameters of the two types of cells are recorded and stored every 1 hour and will be entered into the computer (11) through the sensor, a total of 240 cell characteristic parameters are obtained, and a machine identification information data set is established.

FIG. 6 is a schematic diagram of a specific flow of step S2 in this embodiment of the present application. As shown in Figure 6, the co-cultured cell inoculation density, cell inoculation ratio, cultured cell type and cell co-culture time node are used as inputs, and the cell characteristic parameters of each cell are used as outputs to train a neural network to obtain a cell co-culture model. ,include:

S61 , summarizing the type of each cell and the cell characteristics under each cell co-culture time node under the preset co-culture cell seeding density and cell seeding ratio, to obtain a cell feature data set under each cell co-culture time node;

S62, randomly dividing the cell feature data set under each cell co-culture time node into two parts, a cell feature training set and a cell feature test set;

S63, using the co-cultured cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time node in the cell characteristic training set as input, and the cell characteristic parameters of each cell as output to train a neural network to obtain an initial cell co-culture model;

S64, the co-cultured cell seeding density, cell seeding ratio, cultured cell type, and cell co-culture time node in the cell characteristic test set are used as input, and the cell characteristic parameters of each cell are used as output to test the initial cell co-culture model, which is used to adjust the initial cell co-culture model. The weight value of the cell co-culture model to obtain a cell co-culture model whose prediction accuracy is greater than the preset threshold. In practical applications, the co-cultured cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time node in the cell characteristic test set are used as input, and the cell characteristic parameters of each cell are used as output to test the initial cell co-culture model. It is realized by dynamic time node warping algorithm.

In one embodiment, the training set is used for machine learning training, and the co-cultured cell type, cell seeding density, cell seeding ratio and cell co-cultivation time are input, and the output characteristic parameters are cell size, cell image, cell location, cell growth curve, etc. , established a mesenchymal stem cell-neural stem cell co-culture model through machine learning. For each characteristic parameter, if the error between the model output value and the actual value is less than 5%, the model prediction is accurate, and the DTW method (Dynamic TimeWarpping, dynamic time node warping algorithm) is effective to evaluate the model.

Hereinafter, the method for constructing a cell co-culture model will be described by taking the co-cultured cells as umbilical cord mesenchymal stem cells and brain tumor stem cells as examples.

First, we record the naming in the preset system, such as Experiment 1.

Next, complete the initial recording including: entering two cell names (e.g. umbilical cord mesenchymal stem cells and brain tumor stem cells), selecting the type of medium (SFM medium containing growth factors), selecting the seeding of co-cultured mesenchymal stem cells Density (respectively added to the culture dish at 2×10 ⁵ /ml, 4×10 ⁵ /ml, 6×10 ⁵ /ml, 8×10 ⁵ /ml, 10×10 ⁵ /ml), target cell ratio (that is, the ratio of the seeded number of brain tumor stem cells to mesenchymal stem cells = 1:5, 1:3, 1:1, 3:1 or 5:1), continuous culture.

After that, data acquisition during the culture was performed.

Among them, prior to data acquisition, the following steps are used to identify features related to the developmental prognosis of the target cells from various culture parameters.

Step 1: A strategy based on a combination of lasso regression and Cox proportional hazards regression was used. On the one hand, all co-culture parameters were combined and 15 of them were identified as factors with a significant independent effect on the size of cell development using a univariate Cox proportional hazards model, namely: cell culture time, cell type, culture Base type (both cell type and medium type are categorical variables, coded as 1, 2, 3...N according to known cell and medium types, respectively), medium volume (in ml), medium pH, culture Incubator real-time temperature, incubator CO ₂ concentration percentage, cell density (number of cells per unit volume, unit "cell/ml"), average cell spacing, cell viability (sampled to detect the percentage of viable cells per 100 cells), cells Growth rate (the number of cells increased in 1 hour per unit time), average cell diameter (average value, calculated using (long diameter + short diameter)/2 for irregular cells), cell center point coordinates (x _i , y _i , for The position of the center point of the cell relative to the center point of the culture device), the thickness of the cell membrane, the proportion of the nucleus (the diameter of the nucleus/the average diameter of the cell, the volume of the nucleus used by the irregular cells/the average volume), the expression content of the cell markers (the markers of different cells are different For example, Oct-4 is a marker for embryonic stem cells, CD34 is a marker for hematopoietic stem cells, CD133 is a marker for neural stem cells, etc.).

其中，(x_i，x_j，y_i，y_j)分别是观察细胞相对培养装置中心点的坐标，N为待观察细胞总数，D为细胞平均直径。A lasso regression model was used to exclude variables that contributed less. Finally, 9 variables were retained, namely: cell type, medium type (both are categorical variables, coded as 1, 2, ... N according to known cell and medium types), medium volume, medium pH Value, incubator real-time temperature, incubator _CO2 concentration percentage, cell density (number of cells per unit volume), average cell spacing, cell viability (proportion of viable cells per 100 cells sampled for detection). E.g,

Among them, ( _xi , x _j , _yi , y _j ) are the coordinates of the observed cells relative to the center point of the culture device, N is the total number of cells to be observed, and D is the average cell diameter.

The second step is to establish a multivariate Cox proportional hazards regression model. The collected cell sample data were randomly divided into training and independent test sets. One-third of the samples (n=156) were alternately used as the independent test set, and the other two-thirds were used as the training set, thus constructing two pairs of sample sets. The trained multivariate Cox proportional hazards regression model performed satisfactorily, with a mean concordance index (C-index) equal to 0.836. Next, a risk score for each sample was calculated according to the established multivariate Cox proportional hazards regression model, which is highly discriminative for cell growth and development status. The mean AUC values for cell size prediction in the training set at day 2, 4, 6, 8, and 10 of co-culture reached 0.861, 0.859, 0.867, 0.871, and 0.918, respectively. Regarding the predictions on the test set, the performance dropped slightly, with mean AUC values of 0.786, 0.796, 0.756, 0.761, and 0.767 for cell size prediction on days 2, 4, 6, 8, and 10, respectively. Cox proportional hazards regression models integrated cell types and other processing-related prognostic models to create comprehensive nomograms. The performance of the integrated nomogram was evaluated by RMS curve, time-dependent ROC analysis, calibration curve and decision curve analysis (DCA), and the prediction ability was verified by the internal and external validation sets (HMU and GEO), and the results showed that the integrated nomogram has reliability and stability.

Step 3: Use the neural network toolbox of the existing software (such as MATLAB 7.0) to build the BP neural network model of the training group, verify it with the test set, and combine it with the COX proportional hazards regression model of the first step, for example, The average AUC value predicted by the COX proportional hazards regression model for the average AUC value of the cell size prediction on the 2nd, 4th, 6th, 8th and 10th days in the test set is not less than the preset AUC value, and the neural network model predicts the test set. The accuracy of the BP neural network model is not less than the preset threshold to comprehensively determine the prediction accuracy of the BP neural network model. BP (back propagation) neural network is a multi-layer feedforward neural network trained according to the error back propagation algorithm, and it is one of the most widely used neural network models.

Among them, neural network models include but are not limited to the following three:

The first is to train the neural network with the characteristic size of the cell as the output parameter, and the input value is the cell name, the medium name, the seeding density and proportion of co-cultured cells, and the number of days of culture;

The second is to train the neural network with the reproduction number of cells as the output parameter, and the input values are the cell name, the medium name, the seeding density and proportion of co-cultured cells, and the number of days of culture;

The third is to train the neural network with the cell growth curve as the output parameter, and the input values are the cell name, the medium name, the seeding density and proportion of co-cultured cells, and the number of days of culture.

The construction of the cell co-culture model in the embodiment of the present invention can provide accurate prediction of cell growth and development state for multi-omics characteristics, so as to improve the experimental efficiency of clinical staff. By completing the real-time observation of the cell co-culture, the embodiment of the present invention records the process of the cell co-culture in more detail, and provides a better data acquisition basis for future cell research.

The embodiment of the present invention provides a good data analysis basis for obtaining characteristic data in cell culture by performing machine learning on a large amount of research data obtained by real-time observation of cell co-culture.

The embodiment of the present invention associates time variables with cell characteristic parameters, and can analyze the state of cells in the co-culture process at different time nodes, which is beneficial to observe the differences in different time periods in the cell culture process, and summarize and analyze.

The embodiment of the present invention not only pays attention to the physiological and biochemical parameters of the cells themselves, but also marks the cell surface markers and metabolites through selective markers, so as to facilitate the analysis of the interaction between cells.

FIG. 7 is a schematic flowchart of a method for constructing a cell model according to an embodiment of the present application. As shown in Figure 7, the cell model construction method of the present application includes the following steps:

S71, acquiring the growth curve and morphological characteristic data of the cells at different time nodes, and determining the degree of cell differentiation according to the cell morphological characteristic data;

S72, constructing a cell culture characteristic parameter database according to the growth curves of the cells at different time nodes, the morphological characteristic data, the degree of cell differentiation, and the collected cell physical and biochemical characteristic data;

S73, using three-dimensional modeling technology to construct a virtual culture environment including culture container, culture medium, and culture condition parameters, and based on the cell culture characteristic parameter database, use motion planning technology to generate virtual target cells and add them to the virtual culture environment, so as to The rendering engine is used to generate frame-by-frame cell information and culture environment information, and a cell model under the simulated cell culture system is constructed according to the frame-by-frame cell information and culture environment information.

Preferably, when the virtual culture environment is a co-culture virtual environment, the growth curves and morphological characteristic data of the cells at different time nodes are obtained based on a pre-built cell co-culture model.

FIG. 8 is a schematic flowchart of the construction of a cell co-culture model and a method for acquiring cell state information, taking the virtual culture environment as the co-culture virtual environment and the co-culture cells including two types of cells as an example. As shown in FIG. 8 , the cell co-culture of the present application Model building methods, including:

Co-culture of target cells A and B to be studied;

Combining an observation system and a projection system to obtain cell culture parameters with a visual capture device, wherein the observation system includes a plurality of microscopes, and the projection system at least includes a projection device;

Collect data sets of machine identification information including cell shape, size, location, growth, etc.;

Build co-cultivation machine learning programs;

Build a training library and a test library according to the information data set, and input the co-cultivation machine learning program to obtain a co-cultivation model;

The co-culture model is used to identify and confirm the target cells to be studied, and to obtain the co-cultured cell status at different time points, so as to obtain the growth curve and morphological characteristic data of the cells at different time points.

Taking the virtual culture environment as the co-cultivation virtual environment as an example, the construction of a simulated cell culture system includes the following steps:

Based on the co-culture model of mesenchymal stem cells, the growth curve and morphological characteristic data of mesenchymal stem cells at different time points are obtained, the differentiation degree of mesenchymal stem cells is determined according to the identification of cell morphological characteristics, and the mesenchymal stem cells are obtained by using the data set definition of cell physical and biochemical characteristics. Plasma cell co-culture characteristic parameter database, output cell characteristic data and follow-up statistical analysis according to different needs.

(1) Establishment of cell co-culture model;

(2) Construction of a simulated cell co-culture model: Based on the data model and the co-culture characteristic parameter database, first, a virtual co-culture environment is constructed by using 3D modeling technology, including culture container, culture medium, culture condition parameters, etc. The planning technology generates target cells (including target cells to be studied and conditional observation cells, etc.) and adds them to the co-culture environment. Finally, the rendering engine is used to generate frame-by-frame sensor information and build a simulated mesenchymal stem cell co-culture system.

Taking the virtual culture environment as an irregular cell virtual culture environment as an example, the construction of a simulated cell culture system includes the following steps:

Using the method of taking the co-culture of bone marrow mesenchymal stem cells and neural stem cells as an example, only one type of cell was added to the culture dish. In this example, HepG2 cells were used as an example, and HepG2 cells were seeded at a density of 1 × 104/ ^cm2 . It was inoculated into a petri dish containing 84.5% DMEM+15% fetal bovine serum+0.5% double antibody, and the medium was changed every other day. After 5 days of culture, it was passaged at a ratio of 1:3; the digestion step was washed three times with PBS with pH 7.2. , and then digested with 0.25% trypsin solution for 0.5 minutes; the culture environment, temperature, pH, etc. remained unchanged throughout the culture process, and the cell morphology was displayed and recorded by the projection system, and the cell characteristic proteins were detected by PCR and Western blotting with specific primers. expression.

Culture results: Direct observation of cells showed accumulation growth, the shape of single cells was not obvious, the cell adhesion was not firm, and it was not easy to distinguish the intercellular space. PCR results showed that HepG2 cells could express alpha-fetoprotein, albumin, α-2-macroglobulin, α-1-antitrypsin, transferrin, α-1-antichymotrypsin, haptoglobin, cerulo blue protein, plasminogen, etc., but do not express HBV.

According to the culture results and the obtained cell characteristic parameter database, the C++ programming language is used to digitally simulate the morphology of HepG2 cells, so that the biological signals of cells can be transformed into digital and electronic signals that can be recognized and processed, and a 3D model of HepG2 cells is constructed. The signals are automatically stored, processed, analyzed, integrated and applied to become virtual cells. After storing the virtual cell information and the real culture process, the cell morphology and culture process can be simulated and reproduced, providing the possibility for further applications of technologies such as virtual twin cells in the future.

In the embodiments of the present invention, by simulating digital cell modeling for solid cells, the obtained digital model can lay a foundation for simulating large-scale cell populations in the future.

As shown in FIG. 9 , an embodiment of the present invention provides a cell co-culture model construction and cell state information acquisition device, the device includes: a cell characteristic data receiving and preprocessing module, a cell characteristic data reading and input module, a time Feature input module, machine learning training module, machine learning testing module, virtual digital cell co-culture model and cell feature parameter output module.

In this embodiment, the cell characteristic data receiving and preprocessing module is used for receiving and preprocessing the cell characteristic data for training the virtual digital cell co-culture model.

In this embodiment, the cell characteristic data reading and inputting module is used to read and input the preprocessed cell characteristic data to train a virtual digital cell co-culture model.

In this embodiment, the temporal feature input module is used for inputting temporal features.

In this embodiment, the machine learning training module is used to train a virtual digital cell co-culture model with temporal features and corresponding cell feature data.

A machine learning test module for testing virtual digital cell co-culture models.

A virtual digital cell co-culture model, which is used to obtain cell characteristic parameters corresponding to the time characteristic based on the input time characteristic;

The cell feature parameter output module is used for outputting cell feature parameters corresponding to the input temporal features.

For details of the implementation process of the functions and functions of each unit in the above device, please refer to the implementation process of the corresponding steps in the above method, which will not be repeated here.

For the apparatus embodiments, since they basically correspond to the method embodiments, reference may be made to the partial descriptions of the method embodiments for related parts. The device embodiments described above are only illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed over multiple network elements. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of the present invention. Those of ordinary skill in the art can understand and implement it without creative effort.

Based on the same inventive concept, as shown in FIG. 10 , an embodiment of the present invention provides a computer device including a processor 1110 , a communication interface 1120 , a memory 1130 and a communication bus 1140 , wherein the processor 1110 , the communication interface 1120 , and the memory 1130 Complete mutual communication through the communication bus 1140;

memory 1130 for storing computer programs;

The processor 1110 is configured to implement the following method for constructing a cell co-culture model when executing the program stored in the memory 1130:

Collect real-time cell images in the process of cell co-culture under the preset cell seeding density and cell seeding ratio, extract the cell characteristic parameters of each cell at different time nodes in the cell image, and mark the type and number of cells in the cell image; Taking the co-cultured cell seeding density, cell seeding ratio, cultured cell type and cell co-culture time node as input, and using the cell characteristic parameters of each cell as the output to train the neural network, the cell co-culture model is obtained.

The above-mentioned communication bus 1140 may be a Peripheral Component Interconnect (PCI for short) bus or an Extended Industry Standard Architecture (EISA for short) bus or the like. The communication bus 1140 can be divided into an address bus, a data bus, a control bus, and the like. For ease of presentation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

The communication interface 1120 is used for communication between the above electronic device and other devices.

The memory 1130 may include random access memory (Random Access Memory, RAM for short), and may also include non-volatile memory (non-volatile memory), such as at least one disk storage. Optionally, the memory 1130 may also be at least one storage device located away from the aforementioned processor 1110 .

The above-mentioned processor 1110 may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU for short), a network processor (Network Processor, NP for short), etc.; it may also be a digital signal processor (Digital Signal Processing, DSP for short) , Application Specific Integrated Circuit (ASIC for short), Field-Programmable Gate Array (FPGA for short) or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components.

An embodiment of the present invention provides a computer-readable storage medium, where the computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement any of the above Steps of a method for constructing a cell co-culture model in a possible implementation.

Alternatively, the storage medium may be a non-transitory computer-readable storage medium, for example, the non-transitory computer-readable storage medium may be ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage equipment, etc.

Embodiments of the present invention also provide a computer program product, including a computer program, which, when executed by a processor, implements the steps of the method for constructing a cell co-culture model in any possible implementation manner described above.

In the above-mentioned embodiments, it may be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented in software, it can be implemented in whole or in part in the form of a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the procedures or functions according to the embodiments of the present invention result in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable device. Computer instructions may be stored on or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server, or data center over a wire (e.g. coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (eg, infrared, wireless, microwave, etc.) to another website site, computer, server, or data center. A computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, a data center, or the like that includes an integration of one or more available media. Useful media may be magnetic media (eg, floppy disk, hard disk, magnetic tape), optical media (eg, DVD), or semiconductor media (eg, Solid State Disk (SSD)), among others.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that it can still be The technical solutions described in the foregoing embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (7)

1. A cell co-culture model construction method is characterized by comprising the following steps:

collecting cell images in the cell co-culture process under the preset cell inoculation density and cell inoculation proportion in real time, extracting cell characteristic parameters of each cell in the cell images under different time nodes, and marking the types and the number of the cells in the cell images;

the inoculation density, the inoculation proportion, the category and the co-culture time node of the cells are used as input, the cell characteristic parameter of each cell is used as output to train the neural network, so as to obtain a cell co-culture model,

the method comprises the following steps of collecting cell images in a cell co-culture process under the preset cell inoculation density and the preset cell inoculation proportion in real time, and extracting cell characteristic parameters of each cell in the cell images at different time nodes, wherein the cell characteristic parameters comprise:

collecting cell images in a cell co-culture process under a preset cell inoculation density and a preset cell inoculation proportion in real time by utilizing a plurality of micro lenses, and sending the collected cell images to a projection device;

the projection device is used for de-duplicating and summarizing all cell images of the same acquisition time node, and the received cell images are processed into three-dimensional cell images to be displayed;

and respectively extracting cell characteristic parameters from the displayed three-dimensional cell images under different acquisition time nodes by using a plurality of visual capturing devices.

2. The method of claim 1, wherein prior to said acquiring in real time an image of cells during cell co-culture at a predetermined cell seeding density and a predetermined cell seeding ratio, the method further comprises:

and (3) carrying out cell co-culture on different types of cells according to the same co-culture condition and according to the preset cell inoculation density and cell inoculation ratio, and collecting cell images in the cell co-culture process in the next step.

3. The method of claim 1, wherein training the neural network with the co-culture cell seeding density, the cell seeding ratio, the cultured cell category and the cell co-culture time node as inputs and the cell characteristic parameter of each cell as an output to obtain the cell co-culture model comprises:

summarizing the types of each cell and the cell characteristics under each cell co-culture time node under the preset co-culture cell inoculation density and cell inoculation proportion to obtain a cell characteristic data set under each cell co-culture time node;

randomly dividing a cell characteristic data set under each cell co-culture time node into a cell characteristic training set and a cell characteristic testing set;

taking the co-culture cell inoculation density, the cell inoculation proportion, the culture cell category and the cell co-culture time node in the cell characteristic training set as inputs, and taking the cell characteristic parameter of each cell as an output training neural network to obtain an initial cell co-culture model;

and taking the co-culture cell inoculation density, the cell inoculation proportion, the culture cell category and the cell co-culture time node in the cell characteristic test set as inputs, taking the cell characteristic parameters of each cell as an output test initial cell co-culture model, and adjusting the weight value of the initial cell co-culture model to obtain the cell co-culture model with the prediction accuracy rate larger than a preset threshold value.

4. The method according to claim 3, wherein the co-culture cell seeding density, cell seeding ratio, cultured cell category and cell co-culture time node in the cell characteristic test set are used as input, and the cell characteristic parameter of each cell is used as output test initial cell co-culture model and is realized by a dynamic time node warping algorithm.

5. The method of claim 1, wherein the cell characteristic parameters comprise at least one of cell membrane morphology, cell nucleus morphology, coordinates of cells relative to a center point of the co-culture dish, cell size, cell area, average cell growth rate, and number of cell surface markers and metabolites calculated from images of cells labeled with different fluorescent or isotopic substances.

6. The computer equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor and the communication interface are used for realizing the communication between the processor and the memory through the communication bus;

a memory for storing a computer program;

a processor for implementing the steps of the method of constructing a cell co-culture model according to claims 1-5 when executing the program stored in the memory.

7. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for constructing a cell co-culture model according to claims 1-5.

Images