A poisson event distribution-based flow cytometer waveform signal simulation modeling method

Abstract

The application belongs to the technical field of biomedical detection and computer simulation, and particularly relates to a flow cytometer waveform signal simulation modeling method based on Poisson event distribution. The flow cytometer waveform signal simulation modeling method based on Poisson event distribution comprises the following steps: S1. random event sequence generation; S2. pulse waveform modeling; S3. pulse waveform convolution superposition; S4. noise addition and signal output.

Description

Technical Field

[0001] This invention belongs to the field of biomedical detection and computational simulation technology, specifically relating to a method for simulating and modeling flow cytometer waveform signals based on Poisson event distribution. Background Technology

[0002] The waveform signal of a flow cytometer is essentially a discrete pulse sequence. The waveform of the pulses is a Gaussian-shaped waveform generated by cells or particles flowing through the laser light wave, and the interval between each pulse depends on the random time interval of the cell passing through the detection zone. Traditional simulation methods usually use uniform time intervals or simplified random models to generate cell events, but the actual cell flow principle follows a Poisson process (independent events with a constant occurrence rate). Especially at high speeds or high concentrations, the cell arrival time intervals exhibit an exponential distribution, leading to a significant increase in the probability of signal overlap.

[0003] During flow cytometry, cells flow one by one past a laser detection point, generating electrical signal pulses. Traditional simulation models have two major drawbacks: 1. Inaccurate arrival times: They typically assume cells arrive at equal time intervals (e.g., every 100 microseconds) or using simple random numbers. However, in reality, cell arrival in fluids is completely random and independent, a randomness that, in principle, conforms to the statistical principle of a Poisson process. 2. Inability to simulate high-concentration scenarios: When cell concentrations are high and flow rates are fast, two cells may pass through the detection point almost simultaneously or one after the other, causing their signal pulses to overlap, forming distorted composite waveforms. Traditional models cannot effectively simulate this crucial "signal overlap" interference phenomenon.

[0004] To address the aforementioned shortcomings, this invention aims to provide a method for simulating and modeling flow cytometer waveform signals based on Poisson event distribution. Summary of the Invention

[0005] The purpose of this invention is to provide a method for simulating and modeling waveform signals of flow cytometers based on Poisson event distribution, 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: A method for simulating and modeling flow cytometry waveform signals based on Poisson event distribution includes the following steps: S1. Random Event Sequence Generation: Generate a discrete time series at fixed time intervals, where the number of cells in each time interval follows a Poisson distribution. x is the time interval between two events; the cell concentration is set as λ, where λ refers to the number of events (cell count) within a fixed time interval; based on the influence of cell diameter on waveform height, multiple height signals are used, the intensity of the highest signal peak is taken as logarithm a, the intensity of the lowest signal peak is taken as logarithm b, and a to b are divided into eight equal parts. The value corresponding to each part is raised to a power and corresponds to a height value. Taking the highest signal peak as 60000 and the lowest as 1, the highest signal intensity is taken as log10(60000) = 4.77, log10(1) = 0. The range from 0 to 4.77 is divided into eight equal parts, corresponding to 0.59625, 1.1925, 1.78875, 2.385, 2.98125, 3.5775, 4.17375, and 4.77. After removing the exponents, these correspond to heights of 4, 16, 61, 243, 958, 3780, 14919, and 58884, respectively. These heights are used as the cell event height features. The cell counts within the interval are multiplied by the height signal. The height signal is randomly obtained from the eight heights, and a Gaussian random value is superimposed on each height to characterize the differences between different cells. S2. Pulse Waveform Modeling: The light spot shape adopts a standard Gaussian shape with energy density, and the light spot width is proportional to 1 / e. 2 =10um as the standard, the highest energy density is normalized to 1, the sheath flow velocity is 5m / s, assuming that the cell is considered as a point and normalized to 1, the sheath flow velocity passes through the light spot, the cell flows through the light spot and generates a corresponding light intensity signal, and a standard pulse template is generated with a sampling rate of 25MHz; the standard pulse template refers to the pulse signal waveform generated when a single cell passes through. S3. Pulse Waveform Convolution Superposition: The sequence data containing eight types of height information is convolved with the standard template data. The resulting signal will contain various independent pulses, partially overlapping pulses, or completely overlapping pulses generated due to different time intervals of cell events. At the same time, due to the randomness of the signal height, the resulting overlapping signal has various waveform shapes. S4. Add noise and output signal: The convolution signal obtained in step S3 is added with random basis noise; after adding basis noise, the final waveform signal is output.

[0007] Preferably, the background noise includes the background noise generated by the electronic and optical systems of the analog instrument itself.

[0008] Preferably, the cell concentration λ, cell size distribution, and floor noise can be adjusted as needed.

[0009] Preferably, the intensity of the high-intensity signal includes the intensity of the cellular fluorescence characteristic signal and the fluorescence intensity of various high-intensity microspheres.

[0010] Compared with the prior art, the beneficial effects of the present invention are: (1) The present invention provides a simulation modeling method for flow cytometer waveform signals based on Poisson event distribution, which combines the Poisson statistical model (mathematics) with the pulse waveform physical model (physics) to construct the simulation basis; it realistically simulates the arrival time, in which the arrival of cells in the fluid is completely random and independent, and this randomness is in principle consistent with the statistical Poisson process; (2) This invention provides a flow cytometry waveform signal simulation modeling method based on Poisson event distribution. This method simulates high-concentration scenarios where, when cell concentration is high and flow rate is fast, two cells are likely to pass through the detection point almost simultaneously or one after the other, causing their signal pulses to overlap and form a distorted composite waveform. This method can predict and reproduce the key phenomenon of signal overlap, providing a crucial test benchmark and data source for developing signal processing algorithms such as "de-overlap". Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the structure of the standard pulse template in this invention; Figure 2 This refers to a portion of the waveform signal after convolution in S3 of the present invention. Figure 3 This is a schematic diagram of the waveform of the independent pulse signal in S3 of the present invention; Figure 4 This is a schematic diagram of the waveform of the partially overlapping pulse signal in S3 of the present invention; Figure 5 This is a schematic diagram of the waveform of the completely overlapping pulse signal in S3 of the present invention. Detailed Implementation

[0012] 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.

[0013] A method for simulating and modeling flow cytometry waveform signals based on Poisson event distribution includes the following steps: S1. Random event sequence generation: A discretized time series is generated at fixed time intervals, where the number of cells in each time interval follows a Poisson distribution. Set the time interval between two events to x, with a sampling rate of 25MHz. The time interval x is 1 / 25MHz = 40ns. The average flow rate of the cell concentration is 16000 per second. λ refers to the number of events (cell count) within a fixed time interval. The average flow rate λ within each time period is 16000 * 40 nanoseconds / 1 second = 0.00064, which means that the average number of events occurring within each time interval is 0.00064. Substitute the parameter λ into a Poisson random number generation tool such as Python to generate a random number time series of several seconds or any length. Each data point contains the number of events (cell count) within the time interval.

[0014] Based on the influence of cell diameter on waveform height, multiple height signals are used. Taking the highest signal peak value as 60000 and the lowest as 1, the highest signal strength is log10(60000) = 4.77, log10(1) = 0. The range from 0 to 4.77 is divided into eight equal parts, corresponding to 0.59625, 1.1925, 1.78875, 2.385, 2.98125, 3.5775, 4.17375, and 4.77 respectively. After removing the exponentiation, the corresponding heights are 4, 16, 61, 243, 958, 3780, 14919, and 58884 respectively. These are used as cell event height features. The cell count within the interval is multiplied by the height signal. The height signal is randomly obtained from the eight heights. At the same time, a Gaussian random value is superimposed on each height to characterize the differences between different cells.

[0015] S2. Pulse waveform modeling: The beam pattern adopts a standard Gaussian shape with energy density, and the beam width is 1 / e 2 =10um as the standard, the highest energy density is normalized to 1, the sheath flow velocity is 5m / s, assuming the cell is considered a point and normalized to 1, the sheath flow velocity passes over the light spot, the cell flowing over the light spot generates a corresponding light intensity signal, and a standard pulse template is generated with a sampling rate of 25MHz, see Figure 1 The standard pulse template refers to the pulse signal waveform generated when a single cell passes through it. S3. Pulse Waveform Convolution Stacking: The sequence data containing eight types of height information is convolved with standard template data. The resulting signal will contain various independent pulses, partially overlapping pulses, or completely overlapping pulses generated due to different time intervals of cellular events. At the same time, due to the randomness of the signal height, the resulting overlapping signal has various waveforms.

[0016] S4. Add noise and signal output: In step S3, the convolutional signal is obtained, and random floor noise is added. After adding floor noise, the final waveform signal is output. The floor noise includes the background noise generated by the electronic and optical systems of the analog instrument itself.

[0017] 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 simulating and modeling flow cytometry waveform signals based on Poisson event distribution, characterized in that, Includes the following steps: S1. Random Event Sequence Generation: Generate a set of discretized time series at fixed time intervals, where the number of cells in each time interval follows a Poisson distribution. x is the time interval between two events; the cell concentration is set as λ, where λ refers to the number of events within a fixed time interval. Based on the influence of cell diameter on waveform height, multiple height signals are used. The intensity of the highest signal peak is taken as logarithm 'a', and the intensity of the lowest signal peak is taken as logarithm 'b'. The values ​​from 'a' to 'b' are divided into eight equal parts. The value corresponding to each part is raised to a power and corresponds to a height value. This is used as the height feature of cell events. The cell count within the interval is multiplied by the height signal. The height signal is randomly obtained from the eight heights. At the same time, a Gaussian random value is superimposed on each height to characterize the differences between different cells. S2. Pulse Waveform Modeling: The light spot morphology adopts a standard Gaussian shape energy density, the light spot width is based on a fixed value, the highest energy density is normalized to 1, the sheath flow velocity is v, and the cell is considered as a point and normalized to 1. The sheath flow velocity v passes through the light spot, and the cell generates a corresponding light intensity signal as it flows through the light spot. A standard pulse template is generated according to a fixed sampling rate; the standard pulse template refers to the pulse signal waveform generated when a single cell passes through. S3. Pulse Waveform Convolution Superposition: The sequence data containing eight types of height information is convolved with the standard template data. The resulting signal will contain various independent pulses, partially overlapping pulses, or completely overlapping pulses generated due to different time intervals of cell events. At the same time, due to the randomness of the signal height, the resulting overlapping signal has various waveform shapes. S4. Add noise and output signal: The convolution signal obtained in step S3 is added with random basis noise; After adding the base noise, the final waveform signal is output.

2. The method for simulating and modeling flow cytometry waveform signals based on Poisson event distribution according to claim 1, characterized in that: The background noise includes the background noise generated by the electronic and optical systems of the analog instrument itself.

3. The method for simulating and modeling flow cytometry waveform signals based on Poisson event distribution according to claim 1, characterized in that: The cell concentration λ, cell size distribution, and floor noise can be adjusted as needed.

4. The method for simulating and modeling flow cytometry waveform signals based on Poisson event distribution according to claim 1, characterized in that: The intensity of the high-intensity signal includes the intensity of the characteristic signal of cell fluorescence and the fluorescence intensity of various high-intensity microspheres.

Images