A method, medium and system for adjusting voltage in flow cytometer

Abstract

The present invention provides a method, medium and system for adjusting voltage in the use of a flow cytometer, belonging to the technical field of flow cytometers, the method comprising: obtaining current experimental parameters; obtaining the initial voltage of the first time period at the beginning of the current experiment, and the obtained voltage pulse set, recording the obtained voltage pulse set as the first pulse set; matching in a pre-set flow calibration experiment database to obtain a reference data set with the highest matching degree; clustering the pulses of each channel in the reference data set to obtain a first cluster center group; clustering the first voltage pulse set to obtain a second cluster center group; calculating the difference between the second cluster center group and the first cluster center group; calculating the difference voltage corresponding to the difference, and updating the initial voltage using the obtained difference voltage. The method can solve the technical problems that the voltage adjustment of the flow cytometer in the prior art requires voltage titration in advance, and the experimenter needs to have sufficient voltage adjustment experience.

Description

Technical Field

[0001] The present invention belongs to the technical field of flow cytometers, and in particular, relates to a method, a medium and a system for adjusting voltage in the use of a flow cytometer. Background Art

[0002] Flow cytometer is a very important experimental instrument. Its detection principle is:

[0003] 1. There is a liquid flow system inside the instrument, which allows the cells in the sample to pass through individually in sequence.

[0004] 2. As cells pass through the optical system, different lasers excite the cells and their fluorescent markers.

[0005] 3. The excited light signal is first sorted by the filter and then enters the corresponding detection channel.

[0006] 4. The instrument's electronic system converts the optical signal into an electrical signal for amplification.

[0007] 5. The photomultiplier tube (PMT) amplifies the electrical signal.

[0008] 6. The software calculates the electrical signal parameters and generates the final flow cytometry spectrum.

[0009] The intensity of each channel signal in the flow cytometer can be changed by adjusting the voltage of the photomultiplier tube. This is because fluorescently labeled cells will generate voltage pulses when passing through the laser light path. The size of the voltage pulse also depends on the voltage of the PMT tube. By increasing the voltage of the PMT tube, a stronger current can be produced, thereby producing the effect of amplification gain and amplification signal. The electrical signal output from the PMT needs to be amplified before it can be input into the instrument for analysis. Therefore, in order to obtain more accurate experimental results, the voltage of the flow cytometer needs to be adjusted. In other words, the lower the voltage setting, the smaller the multiple of the electronic signal enhancement during the photoelectric signal conversion, and the weaker the electronic signal obtained. Conversely, the higher the voltage setting, the greater the multiple of enhancement, and the stronger the electronic signal obtained. When adjusting the voltage, if the target cell volume to be detected is small, the voltage value can be appropriately increased so that the target cell and cell fragments can be completely separated in the flow graph; if the target cell volume to be detected is large, the voltage value can be appropriately reduced so that all target cell groups are fully displayed in the flow graph to prevent the cell group from being deformed and distorted when approaching the boundary of the flow graph or being completely outside the boundary. In the prior art, the voltage is first adjusted using a voltage titration method, and then the voltage is manually adjusted again during the experiment. This process often requires a voltage titration to be performed in advance, which is time-consuming and requires the experimenter to have sufficient experience in voltage adjustment. Summary of the invention

[0010] In view of this, the present invention provides a method, medium and system for adjusting voltage in the use of a flow cytometer, which can solve the technical problems in the prior art of adjusting the voltage of a flow cytometer, that is, voltage titration needs to be performed in advance, which is time-consuming and requires the experimenter to have sufficient voltage adjustment experience.

[0011] The present invention is achieved in that:

[0012] A first aspect of the present invention provides a method for adjusting voltage during use of a flow cytometer, comprising the following steps:

[0013] S10, obtaining current experimental parameters, including fluorescence intensity of fluorescent materials, amount of fluorescent materials, sample volume, and sample cell concentration;

[0014] S20, obtaining the initial voltage at the beginning of the first time period of the current experiment and the obtained voltage pulse set, wherein the initial voltage is the voltage of the photomultiplier tube of each channel; the voltage pulse set is a set of voltage pulse signals obtained by each channel, each voltage pulse has two parameters, a pulse width and a pulse height, and the obtained voltage pulse set is recorded as a first pulse set;

[0015] S30, in a pre-set flow calibration experiment database, using the current experimental parameters for matching, and obtaining M flow calibration experiment data with the highest matching degree as a reference data set; the flow calibration experiment database is a set of experimental parameters, experimental voltages, and voltage pulses of multiple sets of flow calibration experiments throughout the experiment, the flow calibration experiment is a flow experiment using detection microspheres instead of cells, the multiple sets of flow calibration experiments have different experimental parameters, and the experimental voltage is the voltage of the photomultiplier tube of each channel;

[0016] S40, clustering the pulses of each channel in the voltage pulse set of each reference data in the reference data set respectively, and obtaining N cluster centers, which are recorded as a first cluster center group;

[0017] S50, clustering the first voltage pulse set to obtain N cluster centers, which are recorded as a second cluster center group;

[0018] S60, using a decision tree regression model to establish a relationship model between the experimental voltage in the voltage pulse set of each reference data in the reference data set and the first cluster center group;

[0019] S70, calculating the difference between the second cluster center group and the first cluster center group;

[0020] S80, using the obtained relationship model to calculate the difference voltage corresponding to the difference, and using the obtained difference voltage to update the initial voltage to obtain a corrected voltage, which is used as the adjusted voltage to adjust the voltages of the photomultiplier tubes of the three channels of FSC, SSC, and fluorescence of the flow cytometer.

[0021] Generally, the first time period is 20 to 60 seconds, preferably 30 seconds.

[0022] Fluorescent materials usually refer to fluorescent antibodies used in flow cytometry. These antibodies can specifically bind to the cell surface or internal components and are labeled with fluorescent tags. Commonly used fluorescent tags include FITC, PE, APC, etc., which can emit fluorescent signals of different wavelengths. Fluorescence intensity refers to the intensity or photon count of the fluorescent signal emitted by the fluorescent label at a specific excitation wavelength. It can be expressed as the average fluorescence intensity or median. Fluorescence intensity is affected by factors such as antibody binding affinity, labeling degree, and excitation power. The fluorescence intensity of antibodies from different batches and brands will also vary.

[0023] On the basis of the above technical solution, the voltage adjustment method in use of a flow cytometer of the present invention can also be improved as follows:

[0024] The step of obtaining the initial voltage at the beginning of the first time period of the current experiment and the obtained voltage pulse set also includes the step of filtering and smoothing the obtained voltage pulse set.

[0025] Wherein, in the step of using the current experimental parameters to perform matching in the pre-set flow calibration experiment database, the matching algorithm adopts a matching degree calculation method of parameter value difference.

[0026] Wherein, in the step of clustering the pulses of each channel in the voltage pulse set of each reference data in the reference data set, the clustering method used is the K-Means clustering algorithm.

[0027] Wherein, the decision tree regression model adopts the CART decision tree.

[0028] The step of calculating the difference voltage corresponding to the difference value by using the obtained relationship model, and updating the initial voltage by using the obtained difference voltage to obtain the corrected voltage specifically includes:

[0029] Calculate the cluster center difference of each channel;

[0030] For each channel, the difference of the channel is input into the relational model to calculate the corresponding voltage adjustment amount;

[0031] For each channel, the calculated voltage adjustment is applied to the current voltage to obtain the updated new voltage;

[0032] Set the boundary threshold of the voltage adjustment amount to avoid exceeding the reasonable range.

[0033] Furthermore, the channels include three channels: FSC, SSC, and fluorescence.

[0034] Among them, 5≤M≤10; 5≤N≤10.

[0035] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are executed, they are used to execute the above-mentioned method for adjusting voltage in use of a flow cytometer.

[0036] A third aspect of the present invention provides a voltage adjustment system for a flow cytometer, comprising the above-mentioned computer-readable storage medium.

[0037] Compared with the prior art, the voltage adjustment method, medium and system provided by the present invention in the use of a flow cytometer have the following beneficial effects:

[0038] 1. The closed-loop automatic optimization adjustment of the photomultiplier tube voltage in the flow cytometer is realized.

[0039] -By extracting the voltage pulse characteristics and establishing a corresponding relationship model between voltage and signal pattern, the ideal voltage parameters can be predicted in real time.

[0040] -Calculate the voltage adjustment amount based on the signal deviation in a closed loop and apply the adjustment to the current voltage to complete automatic optimization.

[0041] -Compared with traditional manual experience adjustment, it realizes automatic and intelligent voltage optimization, and does not require the experimenter to have sufficient voltage adjustment experience.

[0042] 2. Improved the signal quality and stability of flow cytometry.

[0043] -The optimized voltage can maximize the dynamic range and resolution of the detection channel.

[0044] -Reduces signal drift caused by human error and ensures consistency of signal quality.

[0045] - Improved the stability and repeatability of different batches of tests and long-term testing.

[0046] 3. Reduce the workload of operators and reduce testing costs.

[0047] -You only need to pre-set the flow calibration experiment database at the factory, or gradually update the flow calibration experiment database in daily use. Most of the voltage adjustment process is completed automatically without long-term manual operation.

[0048] -Reduce repeated testing and sample consumption, and reduce test costs.

[0049] -Simplifies the voltage setting process and lowers the operator's skill threshold.

[0050] In summary, the present invention designs a set of intelligent and automated flow cytometer voltage optimization solutions, which can solve the technical problems of the prior art that the voltage adjustment of flow cytometers requires voltage titration in advance, which consumes time and the experimenter has sufficient experience in voltage adjustment. It has significant technical effects and application prospects in improving detection stability, applicability, information output, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0052] Figure 1 The present invention provides a flow chart of a method for adjusting voltage during use of a flow cytometer. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0054] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0055] like Figure 1 FIG. 1 is a flowchart of a method for adjusting voltage in a flow cytometer according to a first aspect of the present invention. The method comprises the following steps:

[0056] S10, obtaining current experimental parameters, including fluorescence intensity of fluorescent materials, amount of fluorescent materials, sample volume, and sample cell concentration;

[0057] S20, obtaining the initial voltage at the beginning of the first time period of the current experiment and the obtained voltage pulse set, wherein the initial voltage is the voltage of the photomultiplier tube of each channel; the voltage pulse set is the voltage pulse signal set obtained by each channel, each voltage pulse has two parameters, a pulse width and a pulse height, and the obtained voltage pulse set is recorded as the first pulse set;

[0058] S30, in a pre-set flow calibration experiment database, use the current experimental parameters to perform matching, and obtain M flow calibration experiment data with the highest matching degree as a reference data set; the flow calibration experiment database is a set of experimental parameters, experimental voltages, and voltage pulses of multiple sets of flow calibration experiments throughout the experiment, the flow calibration experiment is a flow experiment using detection microspheres instead of cells, the multiple sets of flow calibration experiments have different experimental parameters, and the experimental voltage is the voltage of the photomultiplier tube of each channel;

[0059] S40, clustering the pulses of each channel in the voltage pulse set of each reference data in the reference data set respectively, and obtaining N cluster centers, which are recorded as the first cluster center group;

[0060] S50, clustering the first voltage pulse set to obtain N cluster centers, which are recorded as a second cluster center group;

[0061] S60, using a decision tree regression model to establish a relationship model between the experimental voltage in the voltage pulse set of each reference data in the reference data set and the first cluster center group;

[0062] S70, calculating the difference between the second cluster center group and the first cluster center group;

[0063] S80, using the obtained relationship model to calculate the difference voltage corresponding to the difference, and using the obtained difference voltage to update the initial voltage to obtain a corrected voltage, which is used as the adjusted voltage to adjust the voltages of the photomultiplier tubes of the three channels of FSC, SSC, and fluorescence of the flow cytometer.

[0064] Among them, in the above technical solution, the step of obtaining the initial voltage at the beginning of the first time period of the current experiment and the obtained voltage pulse set also includes the step of filtering and smoothing the obtained voltage pulse set.

[0065] The specific implementation of step S20 is described as follows:

[0066] First, define some parameters:

[0067] -t0: Experiment start time;

[0068] -T: length of the first time period, here T = 30s;

[0069] -t1=t0+T: end time of the first time period;

[0070] -V0: initial voltage;

[0071] -V i (t): voltage of channel i at time T, where i∈{FSC,SSC,FL} represents different channels;

[0072] -P i : voltage pulse set of channel i;

[0073] -p ij : Voltage pulse set P i The jth voltage pulse in ;

[0074] -W ij : Voltage pulse p ij Pulse width;

[0075] -h ij : Voltage pulse p ij The pulse height;

[0076] Then in the first time period [t0, t1]:

[0077] For each channel i, measure its initial voltage V0 in the time period [t0, t1], that is:

[0078] V0(i)=V i (t0), i∈{FSC, SSC, FL1};

[0079] At the same time, for each channel i, all voltage pulses occurring in the time period [t0, t1] are extracted to form a voltage pulse set P i :

[0080]

[0081] Among them, n i Indicates the number of voltage pulses of channel i.

[0082] For each voltage pulse p ij , extract its pulse width w ij and pulse height h ij :

[0083] p ij =(w ij ,h ij ), j = 1, 2, ..., n i ;

[0084] In summary, we can get the initial voltage V0 and voltage pulse set P at the beginning of the first time period [t0, t1] of the current experiment: i ,in:

[0085] V0(i)=V i (t0), i∈{FSC, SSC, FL1};

[0086]

[0087] p ij=(w ij ,h ij ), j = 1, 2, ..., n i ;

[0088] It should be noted that there may be noise interference in the above process, resulting in an unstable voltage signal. Therefore, in actual implementation, it may be necessary to filter and smooth the signal to improve the accuracy of subsequent analysis. Here, the median filter method is used for filtering.

[0089] In short, the S20 step provides the basic data source for the entire voltage adjustment method, and its quality directly affects the effect of subsequent steps. It is necessary to comprehensively consider the performance of experimental equipment, signal processing algorithms, and database construction, so that the obtained initial voltage and voltage pulse set can accurately reflect the current experimental state and provide a stable and reliable basis for voltage adjustment.

[0090] Among them, in the above technical solution, in the step of matching using the current experimental parameters in the pre-set flow calibration experiment database, the matching algorithm adopts a matching degree calculation method of parameter value difference.

[0091] In step S30, we need to use the current experimental parameters to match in the pre-established flow calibration experiment database to obtain M flow calibration experiment data with the highest matching degree as a reference data set.

[0092] First, define some parameters:

[0093] -P: current experimental parameters, including fluorescence intensity, dosage, sample volume, cell concentration, etc.;

[0094] -D = {d1, d2, ..., d N}: Flow calibration experiment database, containing multiple sets of experimental data;

[0095] -d i : The i-th group of experimental data in the database, including parameter p i 、Voltage i , Pulse Sets i ;

[0096] -p i : Parameters of the i-th group of experiments;

[0097] -v i : The voltage of the i-th group of experiments;

[0098] -s i : The voltage pulse set of the i-th group of experiments;

[0099] -M: the number of the best matching experimental data selected;

[0100] The specific process of step S30 is:

[0101] (1) Define the experimental parameter matching function:

[0102]

[0103] where p j and p ij are the current experimental parameters and the jth item in the i-th group of experimental parameters respectively, and n is the total number of parameters.

[0104] (2) For each experimental data d i , calculate its parameter matching degree m(P, p i ).

[0105] (3) Arrange all experimental data from high to low according to the matching degree:

[0106] d′1,d′2,…,d′ N ;

[0107] where m(P, p′1)≥m(P, p′2)≥…≥m(P, p′ N );

[0108] (4) Select the first M groups of data with the highest matching degree as the reference data set:

[0109] D′={d′1, d′2,…,d′ M};

[0110] At this point, we have obtained M groups of reference data that best match the current experimental parameters. These data come from the flow calibration experiment database, and their parameters are closest to the current experiment. The voltage and pulse data can provide a reference for the current experiment.

[0111] Optionally, there are many designs for the matching function of the experimental parameters. Here, a simple matching method based on parameter value differences is adopted. To improve the matching effect, you can consider:

[0112] (1) Give higher weight to key parameters, such as fluorescence intensity and cell concentration.

[0113] (2) Use parameter normalization to eliminate the dimension effect.

[0114] (3) Add sample library relevance and select reference data that is more relevant to the current sample.

[0115] (4) Using fuzzy matching theory, design the fuzzy matching function.

[0116] (5) Increase human experience and perform customized parameter screening.

[0117] By continuously optimizing the matching function and matching process, the reference data closest to the current experimental state can be selected, providing a reliable basis for subsequent voltage adjustment.

[0118] Preferably, in step S30, a matching degree calculation method based on parameter value difference is used. The calculation formula of this method is:

[0119]

[0120] This method of calculating the matching degree is mainly based on the following considerations:

[0121] 1. Intuitive and simple: This method intuitively compares the current experimental parameters p j With reference data parameter p ij The numerical difference between them is simple to calculate and easy to implement.

[0122] 2. Parameter difference: The formula calculates the difference ratio between the values ​​of each parameter, which can reflect the overall difference between the two sets of parameters. The smaller the difference, the higher the match.

[0123] 3. Balance: The denominator uses the sum of parameter values, which can balance parameters of different dimensions.

[0124] 4. Standardization: The calculation result is in the range of 0-1 and can be regarded as the standardized matching degree.

[0125] 5. Easy to expand: You can easily introduce extensions such as parameter weights to enhance expressiveness.

[0126] 6. Computational efficiency: The matching degree calculation process is simple and efficient, which is conducive to quickly finding the best matching data.

[0127] Of course, this method also has some shortcomings, such as the lack of consideration of the parameter distribution shape. On this basis, we can obtain more accurate matching results by designing more complex matching functions, such as introducing correlation coefficients, machine learning models, etc. However, as a preliminary screening, this simple difference ratio method can meet the requirements and quickly and effectively obtain the reference data that best matches the current experiment.

[0128] Wherein, in the above technical solution, in the step of clustering the pulses of each channel in the voltage pulse set of each reference data in the reference data set, the clustering method used is the K-Means clustering algorithm.

[0129] Step S40 is to cluster the voltage pulse set of each reference data in the reference data set to obtain a first cluster center group. The main idea is:

[0130] 1. Input Data

[0131] -D′: the top M reference data obtained by screening in step S30;

[0132] -d′ i ∈D′: single reference data;

[0133] -s′ i : Data d′ i The voltage pulse set in;

[0134] -s′ ij : Pulse set s′ i The jth pulse in ;

[0135] 2. Clustering method selection

[0136] Here we can choose to use K-Means clustering algorithm to cluster pulses. The advantages are simplicity, efficiency and easy implementation.

[0137] 3. Feature extraction

[0138] For each pulse s′ ij , extract its pulse width w′ ij and pulse amplitude h′ ij as clustering features.

[0139] Then the pulse s′ ij The eigenvectors of are:

[0140] x ij =[w′ ij , h′ ij ];

[0141] 4. Data Standardization

[0142] All pulse features are Z-Score standardized to eliminate the impact of data dimension.

[0143] 5. Determination of the number of cluster centers

[0144] Here, the number of cluster centers M is set, which can be set by empirical values ​​or determined by determining performance indicators.

[0145] 6. K-Means Clustering

[0146] Clustering of all pulse features of each channel, algorithm flow:

[0147] (1) Initialize cluster centers μ1, μ2, …, μ N ;

[0148] (2) Assign each sample to the nearest cluster center according to distance;

[0149] (3) Calculate the cluster center as the new round of cluster center;

[0150] (4) Repeat (2) and (3) until convergence;

[0151] 7. Output cluster centers

[0152] After K-Means clustering, each channel will get N cluster centers The first cluster center group is formed.

[0153] 8. Optimization process

[0154] The K-Means process can be optimized by adjusting parameters and selecting features to make the clustering results more stable and accurate.

[0155] Through the above process, we can get the pulse clustering results of each channel in the reference data set, which constitutes the first cluster center group and lays the foundation for the subsequent steps. We can also consider using other clustering methods to obtain better clustering results.

[0156] Step S50 is to cluster the first pulse set to obtain a second cluster center group. The main idea is similar to step S40:

[0157] 1. Input Data

[0158] -P: current experimental parameters;

[0159] -V0: initial voltage of the current experiment;

[0160] -P i : The voltage pulse set of the current experimental channel i;

[0161] -p ij :Pulse set P i The jth pulse in ;

[0162] 2. Feature extraction

[0163] Same as S40, for each pulse p ij Extraction width W ij and amplitude h ij Construct the feature vector:

[0164] x ij =[w ij ,h ij ];

[0165] 3. Data Standardization

[0166] Perform Z-Score normalization on the feature vector.

[0167] 4. Determination of the number of cluster centers

[0168] The number of cluster centers is also set to N here, the same as S40.

[0169] 5. K-Means Clustering

[0170] The K-Means algorithm is used to cluster the pulses of the current experiment.

[0171] 6. Output cluster centers

[0172] Get the N cluster centers μ1, μ2, ..., μ of each channel in the current experiment N , forming the second cluster center group.

[0173] 7. Optimization process

[0174] Similar to S40, more stable and accurate clustering results can be obtained by adjusting parameters and optimizing them.

[0175] So far, we have clustered the pulses of the current experiment and obtained the second cluster center group. The difference relationship with the first center group can reflect the deviation between the current experiment and the reference experiment, providing a basis for subsequent modeling analysis.

[0176] You can also consider using other clustering methods or introducing preprocessing steps such as dimensionality reduction. You can also study the patterns at different clustering granularities by analyzing the selection of the number of clusters N. These can be used as optimization space to obtain better clustering results and provide solid data support for voltage adjustment.

[0177] Among them, in the above technical solution, the decision tree regression model adopts the CART decision tree.

[0178] The goal of step S60 is to establish a relationship model between the voltage vi in ​​the reference data set and the corresponding first cluster center group using a decision tree regression model.

[0179] 1. Input Data

[0180] -D′: Top M reference data;

[0181] -v i :data d i Voltage;

[0182] -C i :data d i The first cluster center group of each channel (result of step S40);

[0183] 2. Data Preprocessing

[0184] - Pulse clustering feature: Use the coordinate value C of each cluster center i As a feature;

[0185] - Nominal voltage value: Set the voltage v iDiscretized into several nominal values ​​for easy modeling;

[0186] -Data matching: ensure that the number of cluster centers of each data piece corresponds to the voltage value one by one;

[0187] 3. Modeling algorithm selection

[0188] Here we choose the decision tree regression model, mainly considering the following advantages:

[0189] -Non-parametric model, no need to assume data distribution;

[0190] -Can handle nonlinear relationships;

[0191] - Insensitive to outliers;

[0192] - The results are highly interpretable;

[0193] 4. Decision Tree Generation

[0194] Build a decision tree based on the following steps:

[0195] (1) Select segmentation attributes and segmentation points;

[0196] (2) Split the data set to obtain child nodes;

[0197] (3) Recursively generate child nodes until the stopping condition is met;

[0198] 5. Model Optimization

[0199] - Pre-pruning: Control the complexity of the tree during the generation process;

[0200] - Post-pruning: remove non-critical nodes and branches in the tree;

[0201] -Model fusion: integrate multiple decision trees to improve stability;

[0202] 6. Final Model

[0203] The voltage value v in the reference data obtained after training i and the corresponding cluster center C i Decision tree regression model.

[0204] The model establishes a mapping relationship between voltage and clustering pattern, making it possible to predict voltage based on real-time clustering conditions.

[0205] Preferably, the CART decision tree model is used for implementation:

[0206] 1. Input Data

[0207] - Training set: D′ train ={(v1, C1), (v2, C2),…, (vm , C m )}, containing m reference data;

[0208] -Test set: D′ test ={(v′1, C′1), (v′2, C′2),…, (v′ n , C′ n )}, containing n reference data;

[0209] -v i : The voltage of the i-th data in the training set;

[0210] -C i : The cluster center group of each channel of the i-th data in the training set;

[0211] 2. CART decision tree generation

[0212] -Traverse the coordinates of each cluster center and select the coordinates and threshold with the largest information gain as the split point;

[0213] - Recursively split the dataset until the sample set purity meets the requirements;

[0214] -Generate a decision tree, with leaf nodes storing the target voltage mean;

[0215] 3. Decision Tree Example

[0216] If the first layer chooses to split "Channel 1, cluster 3 center x value < 0.5", the splitting rule is:

[0217] C 1，3 (x)<0.5;

[0218] Repeat the splitting until a leaf node is reached, which stores the voltage mean.

[0219] 4. Pruning optimization

[0220] Obtaining the optimal subtree through CCP pruning

[0221] 5. Model Fusion

[0222] Integrate multiple decision trees to improve generalization ability

[0223] After the above steps, we can obtain the CART decision tree model of the mapping relationship between voltage and cluster center. This model expresses the nonlinear relationship between the two in the form of a series of splitting rules, providing a basis for subsequent voltage prediction. We can also continue to optimize the structure and hyperparameters of the model to obtain better fitting and prediction results.

[0224] In step S70, we need to calculate the difference between the second cluster center group and the first cluster center group to reflect the deviation between the current experimental state and the ideal state.

[0225] 1. Input Data

[0226] -C i : The first cluster center group comes from the reference data and has a length of N;

[0227] -C′ i : The second cluster center group, from the current experiment, also has a length of N;

[0228] Among them, i represents different channels, such as the cluster center group of the FSC channel is C FSC , C′ FSC .

[0229] 2. Difference calculation

[0230] For each channel i, calculate the difference between the corresponding cluster center groups:

[0231] For j = 1, 2, ..., N:

[0232] ΔC ij =C′ ij -C ij ;

[0233] Then we get the N cluster center differences ΔC of the channel i ={ΔC i1 , ΔC i2 ,....,ΔC iN};

[0234] 3. Difference analysis

[0235] By analyzing the cluster center differences of each channel, we can find out the differences between the current experiment and the ideal state in various signal modes.

[0236] For example, if ΔC i3 It corresponds to the center of the cell cluster, and its significant deviation may indicate that there is a deviation in the signal of this type of cell at present.

[0237] 4. Optional, multivariate statistical analysis

[0238] The difference values ​​of each channel can be subjected to multivariate statistical analysis, such as principal component analysis, to extract the main difference patterns;

[0239] This can further deepen the quantitative understanding of the difference between the current experimental state and the ideal state.

[0240] 5. Optional, relationship visualization

[0241] Use visualization techniques to show the difference relationship between cluster center groups of each channel;

[0242] For example, the cluster centers can be represented on a two-dimensional graph and then connected with vectors to show their difference relationships.

[0243] The above is one way to calculate the difference of cluster center groups. We can also consider introducing weights when calculating the difference to emphasize the focus on specific patterns. This can provide more targeted difference analysis results. In short, through the calculation and analysis of the difference of cluster centers, the state deviation of the current experiment can be quantitatively evaluated, providing a basis for subsequent model prediction and voltage adjustment.

[0244] Among them, in the above technical solution, the step of calculating the difference voltage corresponding to the difference using the obtained relationship model, and updating the initial voltage using the obtained difference voltage to obtain the corrected voltage specifically includes:

[0245] Calculate the cluster center difference of each channel;

[0246] For each channel, the difference of the channel is input into the relational model to calculate the corresponding voltage adjustment amount;

[0247] For each channel, the calculated voltage adjustment is applied to the current voltage to obtain the updated new voltage;

[0248] Set the boundary threshold of the voltage adjustment amount to avoid exceeding the reasonable range.

[0249] Step S80 is to calculate the corresponding voltage adjustment amount according to the calculated cluster center difference using the established decision tree model, and update the voltage to complete the final voltage adjustment.

[0250] 1. Input Data

[0251] -ΔC i : The cluster center difference of each channel (result of step S70);

[0252] -f: The decision tree regression model established in step S60, with the input as cluster center and the output as voltage;

[0253] 2. Differential voltage calculation

[0254] For each channel i:

[0255] ΔC i Input to model f:

[0256] v i =f(ΔC i );

[0257] Then the voltage adjustment value v corresponding to the channel is obtained i .

[0258] 3. Voltage update

[0259] For each channel i, the voltage v is adjusted i Applied to the current voltage V i To get the new voltage:

[0260] V′ i =V i +v i ;

[0261] Then the updated voltage V′ of each channel is obtained i .

[0262] 4. Set boundaries

[0263] Set the boundary threshold of the voltage adjustment amount to avoid extreme updates that exceed the range.

[0264] 5. Dynamic Adjustment

[0265] Closed-loop control can be constructed to continuously adjust the voltage according to the signal to achieve dynamic optimization.

[0266] Through the above process, we can adjust the voltage parameters in a targeted manner according to the model prediction results, complete the closed-loop voltage optimization, and make the signal response reach the optimal state.

[0267] Furthermore, in the above technical solution, the channels include three channels: FSC, SSC, and fluorescence.

[0268] Among them, in the above technical solution, 5≤M≤10; 5≤N≤10.

[0269] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are executed, they are used to execute the above-mentioned method for adjusting voltage in use of a flow cytometer.

[0270] A third aspect of the present invention provides a voltage adjustment system for a flow cytometer, comprising the above-mentioned computer-readable storage medium.

[0271] Specifically, the principle of the present invention is:

[0272] The core principle of the present invention is to establish a correlation model between voltage parameters and detection signal patterns in a flow cytometer, and to guide closed-loop optimization adjustment of voltage through model prediction to ensure that the signal is stable in the optimal response state. The principles include:

[0273] 1. There is a corresponding relationship between signal mode and voltage.

[0274] The photomultiplier tube voltage setting of the flow cytometer affects the gain and dynamic range of the measurement channel, and thus affects the detected cell signal pattern. Appropriate voltage can fully distinguish the signal patterns of different cell populations.

[0275] 2. The signal pattern can be characterized by the voltage pulse characteristics.

[0276] The shape and distribution of voltage pulses reflect the statistical characteristics of different cell signaling events. Extracting pulse features can quantitatively represent different cell signaling patterns.

[0277] 3. Establish a mapping model between voltage and signal pattern.

[0278] Using the experimental data of flow calibration, a relationship model between voltage parameters and corresponding ideal signal patterns is established. The model expresses the complex mapping between the two through learning.

[0279] 4. Adjust the voltage according to the real-time signal.

[0280] For the running cell samples, the real-time pulse characteristics are extracted to represent the current signal mode. The corresponding voltage parameters are predicted by the input model to complete the closed-loop control.

[0281] 5. The voltage gradually approaches the optimum.

[0282] The adjusted voltage can make the signal pattern close to the ideal distribution state. Through continuous optimization, the voltage parameters can gradually approach the optimal value.

[0283] 6. Improve detection stability.

[0284] Under closed-loop control, the system can autonomously adapt to different batches of cells and signal drift, ensuring that the voltage is stable at the optimal point and enhancing detection stability.

[0285] Through the above technical means and principles, the present invention realizes closed-loop, intelligent optimization and adjustment of the voltage parameters of the flow cytometer. This is sufficient to improve the detection performance, expand the scope of application, increase information output, and has important technological progress significance.

[0286] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for adjusting voltage in a flow cytometer, characterized in that: The following steps are involved: S10, obtaining current experimental parameters, including fluorescence intensity of fluorescent materials, amount of fluorescent materials, sample volume, and sample cell concentration; S20, obtaining the initial voltage at the beginning of the first time period of the current experiment and the obtained voltage pulse set, wherein the initial voltage is the voltage of the photomultiplier tube of each channel; the voltage pulse set is a set of voltage pulse signals obtained by each channel, each voltage pulse has two parameters, a pulse width and a pulse height, and the obtained voltage pulse set is recorded as a first pulse set; S30, in a pre-set flow calibration experiment database, using the current experimental parameters for matching, and obtaining M flow calibration experiment data with the highest matching degree as a reference data set; the flow calibration experiment database is a set of experimental parameters, experimental voltages, and voltage pulses of multiple sets of flow calibration experiments throughout the experiment, the flow calibration experiment is a flow experiment using detection microspheres instead of cells, the multiple sets of flow calibration experiments have different experimental parameters, and the experimental voltage is the voltage of the photomultiplier tube of each channel; S40, clustering the pulses of each channel in the voltage pulse set of each reference data in the reference data set respectively, and obtaining N cluster centers, which are recorded as a first cluster center group; S50, clustering the first voltage pulse set to obtain N cluster centers, which are recorded as a second cluster center group; S60, using a decision tree regression model to establish a relationship model between the experimental voltage in the voltage pulse set of each reference data in the reference data set and the first cluster center group; S70, calculating the difference between the second cluster center group and the first cluster center group; S80, using the obtained relationship model to calculate the difference voltage corresponding to the difference, and using the obtained difference voltage to update the initial voltage to obtain a corrected voltage, which is used as the adjusted voltage to adjust the voltages of the photomultiplier tubes of the three channels of FSC, SSC, and fluorescence of the flow cytometer.

2. A method for adjusting voltage in use of a flow cytometer according to claim 1, characterized in that: The step of obtaining the initial voltage at the beginning of the first time period of the current experiment and the obtained voltage pulse set also includes the step of filtering and smoothing the obtained voltage pulse set.

3. A method for adjusting voltage in use of a flow cytometer according to claim 1, characterized in that: In the step of using the current experimental parameters to perform matching in the pre-set flow calibration experiment database, the matching algorithm adopts a matching degree calculation method of parameter value difference.

4. A method for adjusting voltage in use of a flow cytometer according to claim 1, characterized in that: In the step of clustering the pulses of each channel in the voltage pulse set of each reference data in the reference data set, the clustering method used is the K-Means clustering algorithm.

5. The method for adjusting voltage in use of a flow cytometer according to claim 1, characterized in that: The decision tree regression model adopts the CART decision tree.

6. A method for adjusting voltage in use of a flow cytometer according to claim 1, characterized in that: The step of calculating the difference voltage corresponding to the difference value by using the obtained relationship model, and updating the initial voltage by using the obtained difference voltage to obtain the corrected voltage specifically includes: Calculate the cluster center difference of each channel; For each channel, the difference of the channel is input into the relational model to calculate the corresponding voltage adjustment amount; For each channel, the calculated voltage adjustment is applied to the current voltage to obtain the updated new voltage; Set the boundary threshold of the voltage adjustment amount to avoid exceeding the reasonable range.

7. A method for adjusting voltage in use of a flow cytometer according to claim 1, characterized in that: 5≤M≤10;5≤N≤10.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program instructions, and when the program instructions are executed, they are used to execute the method for adjusting voltage in use of a flow cytometer according to any one of claims 1 to 7.

9. A voltage adjustment system for a flow cytometer, characterized in that: Contains the computer-readable storage medium of claim 8.

Images