A flow cytometer signal background dynamic acquisition method based on window sliding screening

Abstract

The application belongs to the technical field of flow cytometry signal processing, and particularly relates to a flow cytometer signal background dynamic acquisition method based on window sliding screening, which comprises the following steps: S1. selecting a signal; S2. setting a screening parameter: setting an initial screening window size as N units, and setting an initial reasonable standard deviation of the initial screening condition signal as σ0; S3. finding a first reasonable window: slidingly finding with the screening window with a size of N units, calculating a signal value mean μ1 and a signal value standard deviation σ1 of the signal Vt in the screening window; when the standard deviation σ1 meets the condition of being less than the reasonable standard deviation σ0, the window is the first reasonable window; S4. dynamic adjustment of the screening condition: after the nth reasonable window is found, the screening condition is updated as μ n - 3σ n ≤ the signal value of Vt ≤ μ n + 3σ n , and the (n+1)th reasonable window is found with the condition, the signal value of the signal Vt in the (n+1)th reasonable window meets the condition of μ n - 3σ n ≤ the signal value of Vt ≤ μ n + 3σ n ; and S6. real-time extraction of a background value.

Description

Technical Field

[0001] This invention belongs to the field of flow cytometry signal processing technology, specifically relating to a method for dynamically acquiring the background signal of flow cytometers based on window sliding screening. Background Technology

[0002] Flow cytometry, a crucial and indispensable technology platform in modern biomedical research and clinical diagnostics, possesses the capability for rapid, multi-parameter, high-throughput analysis of single cells (or other particles) suspended in liquids. This technology can accurately detect and analyze the physical properties (such as size, shape, and internal structure) and chemical properties (such as protein expression levels and nucleic acid content) of single cells. Its core working principle can be summarized into three main steps: First, hydrodynamic focusing, which precisely controls the flow of liquid to align cells one by one and allow them to pass through the detection area sequentially; second, illumination and light scattering / emission, which uses a laser of a specific wavelength to irradiate the cells, exciting fluorescence or generating scattered light signals; and finally, signal detection and conversion, which converts the captured optical signals into digital data for further analysis.

[0003] In actual flow cytometry testing, triggering operations and related data processing are based on the signal background. The background refers to the circuit characteristics exhibited by the flow cytometer after removing background signals caused by electronic noise, non-specific fluorescence, and interference factors such as cell debris. Accurately acquiring the background signal is crucial for ensuring the reliability of experimental results. However, most flow cytometers currently employ fixed threshold methods or static averaging methods for acquiring the background signal. While these methods are simple and easy to implement, they have significant drawbacks: First, there is a lag problem; because these methods cannot respond to signal fluctuations in real time, dynamic experimental data may be distorted, affecting the accuracy of the results. Second, there is a fixed window problem; the preset window size is difficult to flexibly adapt to the possible differences in noise characteristics between different samples, thus limiting its applicability. Third, there is anomaly interference; during the acquisition of the background signal, the presence of cell debris or voltage fluctuations and other external factors can easily lead to biases in the background estimation, further reducing the reliability of data analysis.

[0004] To effectively address the aforementioned problems, this invention proposes a method for dynamically acquiring background signals in flow cytometry based on window sliding screening. This method, by introducing a dynamic adjustment mechanism, adaptively adjusts the window size and position based on real-time monitoring of signal changes, thereby more accurately reflecting the changing trends of the actual background signal. This method overcomes the limitations of traditional fixed threshold and static averaging methods, providing more precise data support for biomedical research and clinical diagnosis. Summary of the Invention

[0005] The purpose of this invention is to provide a method for dynamically acquiring the background signal of flow cytometer based on window sliding screening, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for dynamically acquiring background signals from flow cytometry based on window sliding screening includes the following steps:

[0008] S1. Signal selection: Under a fixed sampling frequency, a signal Vt is selected, which is a superimposed signal of an event Gaussian peak and an electrical noise baseline according to a Poisson distribution;

[0009] S2. Set the filtering parameters: Set the initial filtering window size to N units, and the initial reasonable standard deviation of the initial filtering condition signal to σ0;

[0010] S3. Finding the first reasonable window: After the signal data is input, mark the starting point of the signal data as 0; starting from the starting point of the signal data, slide the filter window of size N units step by step, with each slide being 1 unit, and calculate the mean value μ1 and standard deviation σ1 of the signal value of the signal Vt within the filter window in real time; when the standard deviation σ1 is less than the reasonable standard deviation σ0, the window is the first reasonable window;

[0011] S4. Dynamic adjustment of filtering criteria: After finding the first reasonable window, update the filtering criteria to the signal value of μ1-3σ1≤Vt≤μ1+3σ1, and use this as the filtering criteria to find the second reasonable window; the signal value of Vt in the second reasonable window meets the condition μ1-3σ1≤Vt≤μ1+3σ1; and so on, after finding the nth reasonable window, update the filtering criteria to μ n -3σ n Signal value ≤ Vt ≤ μ n +3σ n And using this condition, find the (n+1)th reasonable window, where the signal value of signal Vt within the (n+1)th reasonable window conforms to μ. n -3σ n Signal value ≤ Vt ≤ μ n +3σ n; where μ n Let σ be the mean signal value of signal Vt within the nth reasonable window. n Let Vt be the standard deviation of the signal value within the nth reasonable window;

[0012] S6. Real-time extraction of background value: The average value of the Vt signal within all reasonable windows is calculated, and the average value is the background value of the corresponding reasonable window; as the reasonable windows are continuously updated, the background value is updated in real time.

[0013] Preferably, step S1 includes denoising: performing Gaussian filtering on the signal Vt to denoise it based on the signal characteristics.

[0014] Preferably, S5 is included, wherein S5 is located between S4 and S6;

[0015] S5 refers to the distance control from the peak: when the signal Vt is close to the event or at the bottom of the peak, it will slowly rise or fall, and this segment of signal Vt may meet the screening condition μ. n -3σ n Signal value ≤ Vt ≤ μ n +3σ n However, the actual signal Vt is unreasonable; when calculating the background value, a small window of M units width in the middle of the window is taken for calculation. M < N, so the signal before and after the small window is discarded. At this time, the mean value μ obtained is... n It is the average of all signals within the small window, which is used to control the distance from the peak and keep it away from the Gaussian peak.

[0016] Preferably, S7 is included, which is located after S6;

[0017] S7 refers to the automatic reloading of the algorithm: when the relevant hardware parameters are adjusted, the signal Vt will increase or decrease in a stepwise manner, and the signal will no longer meet the screening condition μ. n -3σ n Signal value ≤ Vt ≤ μ n +3σ n The background value can no longer be updated in real time; at this point, the interval time of the signal Vt is judged, and signals that exceed the specified interval time no longer meet the screening condition μ. n -3σ n Signal value ≤ Vt ≤ μ n +3σ n Then, follow the steps of S3-S6 again to find a reasonable window and continue to update the new background value in real time.

[0018] Preferably, the sampling frequency in S1 is 25MHz.

[0019] Preferably, in step S2, setting the filtering parameters is as follows: the initial filtering window size is set to N=300, and the initial reasonable standard deviation of the initial filtering condition signal is σ0=200.

[0020] Preferably, in step S7, the specified interval time is 10ms.

[0021] Compared with the prior art, the beneficial effects of the present invention are:

[0022] (1) The present invention provides a method for dynamic acquisition of flow cytometry signal background based on window sliding screening. The screening parameters have an automatic iteration function and can adjust the screening conditions according to the characteristics of the signal. In actual operation, reasonable standard deviation, μ n σ n These parameters are not fixed but can be continuously optimized and updated during operation. When new signals or data are received, the system automatically analyzes the characteristics of the signal and then adjusts the original screening conditions accordingly based on these characteristics.

[0023] (2) The present invention provides a method for dynamic acquisition of flow cytometry signal background based on window sliding screening. The parameter configuration process is highly simple and operable. In the entire detection process, the operator only needs to set the size of the screening window and the initial screening conditions. During the detection process, the algorithm can continuously update and adjust the screening conditions according to the dynamic changes in the signal trend, thereby finding a reasonable window and continuously obtaining the real-time calculated background value by calculating the average of all reasonable windows.

[0024] (3) The present invention provides a method for dynamic acquisition of background signal of flow cytometer based on window sliding screening, which has the function of abnormal signal filtering. When the screening window encounters the event Gaussian peak, some signal points on the front or back of the window can be discarded to prevent them from affecting the detection results. It can also smooth out obviously unreasonable signal points. Attached Figure Description

[0025] Figure 1 To obtain a signal Vt with an event Gaussian peak superimposed on an electrical noise baseline according to a Poisson distribution at a fixed sampling frequency;

[0026] Figure 2 This is a schematic diagram of the baseline of signal Vt when there are no events.

[0027] Figure 3 A schematic diagram of the data after Gaussian filtering and denoising of signal Vt;

[0028] Figure 4 A diagram illustrating the search for the first reasonable window;

[0029] Figure 5 A diagram illustrating the search for the first and second reasonable windows;

[0030] Figure 6 A schematic diagram illustrating the handling of the filtering window to cope with event peaks;

[0031] Figure 7 Detailed illustration of the filtering window handling to cope with event peaks. Figure 1 ;

[0032] Figure 8 Detailed illustration of the filtering window handling to cope with event peaks. Figure 2 ;

[0033] Figure 9 This is a schematic diagram of the background data results for this segment. The signal values ​​are all around 1000, and the influence of the event peak on the background results was successfully removed when there was a peak. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] At a sampling frequency of 25MHz, the signal Vt, which follows a Poisson distribution with Gaussian peaks superimposed on an electrical noise baseline, is used. Its operating data is as follows: Figure 1 As shown. In the absence of events, the baseline of signal Vt is around 1000 AD, as... Figure 2 As shown. Further, an FIR filter is used to perform Gaussian filtering on the signal Vt for noise reduction. The denoised data is shown below. Figure 3 As shown.

[0036] Set the filtering data, with an initial filter window size of 300 units and an initial reasonable standard deviation of 200 for the initial filter condition signal. After setting the initial filter condition, begin searching for the first reasonable window. Mark the starting point of the signal data as 0, and starting from the 0 mark, the filter window gradually slides along the signal data to find a reasonable window, with each slide being 1 unit.

[0037] For each unit scrolled, the mean and standard deviation of the signal value of signal Vt within the filter window are automatically calculated. When searching for the first reasonable window, if the standard deviation of the signal value in the filter window is less than the initial reasonable standard deviation of 200, the filter window is considered the first reasonable window. Figure 4 A diagram illustrating the search for the first reasonable window.

[0038] The filter window starts at scale 0 and slides two units to find the first reasonable window. The mean signal value of this reasonable window is 999, and the standard deviation of the signal value is 21. Figure 4 After finding the first reasonable window, the filtering criteria are automatically updated. According to the condition that the signal value of Vt is ≤ μ1-3σ1≤Vt≤μ1+3σ1, if the signal value of Vt is in the range of 936~1062, it can be identified as the second reasonable window. Figure 5 This is a diagram illustrating the search for the first and second reasonable windows.

[0039] Based on the filtering criteria of the second reasonable window, a second reasonable window for signal Vt is found. The mean signal value of the second reasonable window is 998, and the standard deviation of the signal value of the second reasonable window is 19. Figure 5 After finding the second reasonable window, the filtering criteria are automatically updated. According to the condition μ2-3σ2≤Vt signal value≤μ2+3σ2, when the Vt signal value meets the range of 941~1055, it can be identified as the third reasonable window.

[0040] The dynamic adjustment rule for the filtering criteria can be summarized as follows: after finding the nth reasonable window, update the filtering criteria to μ. n -3σ n Signal value ≤ Vt ≤ μ n +3σ n And using this condition, find the (n+1)th reasonable window, where the signal value of signal Vt within the (n+1)th reasonable window conforms to μ. n -3σ n Signal value ≤ Vt ≤ μ n +3σ n ; where μ n Let σ be the mean signal value of signal Vt within the nth reasonable window. n Let Vt be the standard deviation of the signal value within the nth reasonable window. Based on this pattern, reasonable windows are continuously sought. Within a reasonable window, the mean signal value of Vt approaches the background value of signal Vt infinitely as the reasonable window is continuously updated.

[0041] When the relevant hardware parameters are adjusted, the signal Vt will increase or decrease in a step-like manner, causing the Vt signal to no longer meet the screening condition μ. n -3σ n Signal value ≤ Vt ≤ μ n +3σ nAt this point, the background value cannot be updated in real time. In such cases, a specified interval of 10ms is set. Within 10ms, the filtering window continues to repeatedly search for a suitable window along the signal data. When the specified interval is exceeded, the relevant data of the signal no longer meets the filtering criteria. n -3σ n Signal value ≤ Vt ≤ μ n +3σ n When the algorithm is reloaded, it searches for the first reasonable window. After finding the first reasonable window, it continues to update the background value of the reasonable window in real time.

[0042] In the process of finding a reasonable window, it is necessary to control the distance from the peak. When the signal Vt is close to the event or at the bottom of the peak, it will slowly rise or fall, and this segment of signal Vt may meet the screening condition μ. n -3σ n Signal value ≤ Vt ≤ μ n +3σ n However, the actual signal Vt is unreasonable; when calculating the background value, a small window of M units width in the middle of the window is taken for calculation. M < N, so the signal before and after the small window is discarded. At this time, the mean value μ obtained is... n It is the average of all signals within the small window, which is used to control the distance from the peak and keep it away from the Gaussian peak.

[0043] Figure 6 A schematic diagram illustrating the handling of the filtering window to cope with event peaks. Figure 7 Detailed illustration of the filtering window handling to cope with event peaks. Figure 1 . Figure 8 Detailed illustration of the filtering window handling to cope with event peaks. Figure 2 .

[0044] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for dynamically acquiring background signals from flow cytometry based on window sliding screening, characterized in that, Includes the following steps: S1. Select signal Vt; S2. Set the filtering parameters: Set the initial filtering window size to N units, and the initial reasonable standard deviation of the initial filtering condition signal to σ0; S3. Finding the first reasonable window: After the signal data is input, mark the starting point of the signal data as 0; starting from the starting point of the signal data, slide the filter window of size N units step by step, with each slide being 1 unit, and calculate the mean and standard deviation of the signal value of signal Vt within the filter window in real time; when the standard deviation is less than the reasonable standard deviation σ0, the window is the first reasonable window; S4. Dynamic adjustment of filtering criteria: After finding the first reasonable window, update the filtering criteria to the signal value of μ1-3σ1≤Vt≤μ1+3σ1, and use this as the filtering criteria to find the second reasonable window; the signal value of Vt in the second reasonable window meets the condition μ1-3σ1≤Vt≤μ1+3σ1; and so on, after finding the nth reasonable window, update the filtering criteria to μ n -3σ n Signal value ≤Vt≤μ n +3σ n And using this condition, find the (n+1)th reasonable window, where the signal value of signal Vt within the (n+1)th reasonable window conforms to μ. n -3σ n Signal value ≤Vt≤μ n +3σ n ; Where μ1 is the mean signal value of signal Vt within the first reasonable window, σ1 is the standard deviation of signal Vt within the first reasonable window, and μ n Let σ be the mean signal value of signal Vt within the nth reasonable window. n Let Vt be the standard deviation of the signal value within the nth reasonable window; S5. Peak Distance Control: When the signal Vt is close to the event peak or at the bottom edge of the peak, it will slowly rise or fall. This segment of signal Vt may meet the screening condition μ. n -3σ n Signal value ≤Vt≤μ n +3σ n However, the actual value of the signal Vt in this segment is unreasonable. When calculating the background value, a small window with a width of M units in the middle of the window is taken for calculation. M < N, and the signals before and after the small window are discarded. At this time, the mean value calculated is the mean value of all signals in the small window, thereby controlling the distance from the peak and moving away from the Gaussian peak. The window refers to a window that contains a slowly rising or slowly falling signal at the bottom edge of the peak. S6. Real-time extraction of background value: The average value of the Vt signal within all reasonable windows is calculated, and this average value is the background value of the corresponding reasonable window; as the reasonable windows are continuously updated, the background value is updated in real time.

2. The method for dynamic acquisition of flow cytometry signal background based on window sliding screening according to claim 1, characterized in that, Step S1 further includes denoising: Gaussian filtering is performed on the signal Vt to denoise it based on the signal characteristics.

3. The method for dynamic acquisition of flow cytometry signal background based on window sliding screening according to claim 1, characterized in that: It also includes step S7, which is located after step S6; Step S7 is the automatic reloading of the algorithm: when adjusting the relevant hardware parameters, the signal Vt will generally increase or decrease in a stepwise manner, and the signal will no longer meet the screening condition μ. n -3σ n Signal value ≤Vt≤μ n +3σ n The baseline value can no longer be updated in real time; at this point, the signal Vt is judged as follows: when signals exceeding the specified interval no longer meet the filtering condition μ n -3σ n Signal value ≤Vt≤μ n +3σ n At that time, follow steps S3-S6 to find a reasonable window again and continue to update the new background value in real time.

4. The method for dynamic acquisition of flow cytometry signal background based on window sliding screening according to claim 1, characterized in that: The sampling frequency in step S1 is 25MHz.

5. The method for dynamic acquisition of flow cytometry signal background based on window sliding screening according to claim 1, characterized in that: Step S2. Set the filtering parameters: Set the initial filtering window size to N=300, and the initial reasonable standard deviation of the initial filtering condition signal to σ0=200.

6. The method for dynamic acquisition of flow cytometry signal background based on window sliding screening according to claim 3, characterized in that: In step S7, the specified interval time is 10ms.

7. The method for dynamic acquisition of flow cytometry signal background based on window sliding screening according to claim 1, characterized in that: Step S1 involves taking a signal Vt, which is a superimposed electrical noise baseline of an event Gaussian peak following a Poisson distribution, at a fixed sampling frequency.

Images