Particle detection device and method based on photoelectric sensor data fusion

Abstract

The invention provides a particle detection device and method based on photoelectric sensor data fusion, and the device comprises a micro-channel (7) which is formed by a bottom plate (5) and a cover plate (9) in an up-and-down encircling manner, and is used for a to-be-detected cell (2) to pass through; a first electrode (1), a second electrode (3) and a third electrode (4) are sequentially and fixedly arranged at the bottom of the micro-channel (7), namely the upper side of the bottom plate (5), and are used for detecting electrical characteristics generated by the cells (2); a first optical fiber (6) and a second optical fiber (8) are arranged at corresponding positions on the bottom plate (5) and the cover plate (9) respectively, and the first optical fiber (6) and the second optical fiber (8) are used for sending or receiving optical waves respectively and used for detecting optical characteristics of the cells (2); according to the method, the device is adopted for tumor cell detection. According to the method, data of the electrical sensor and data of the optical sensor are fused, so that higher type identification precision is achieved.

Description

Technical Field

[0001] The present invention relates to a particle detection device and method, in particular to a particle detection device and method based on photoelectric sensor data fusion. Background Art

[0002] This section merely provides background information related to the present disclosure and is not necessarily prior art.

[0003] Non-labeled detection of tumor cells is a technology that does not rely on specific markers. It identifies and analyzes cells through their physical, chemical or metabolic characteristics. The significance of this method is to avoid the cell interference or limitations of marker selection that may be caused by traditional labeling methods, thereby achieving wider applicability and higher detection sensitivity. Its value is reflected in the rapid detection of rare tumor cells, real-time monitoring of tumor dynamic changes, and the promotion of personalized medicine. It provides reliable technical support for early diagnosis, precision treatment and disease course monitoring.

[0004] The label-free detection technology of tumor cells realizes species identification by analyzing the inherent physical or chemical properties of cells, avoiding the interference of traditional labeling methods on cells and the limitations of marker selection, and shows important value in the detection of rare tumor cells and real-time dynamic monitoring. At present, the detection method based on electrical impedance has been widely used in this field, which distinguishes cell types by measuring the impedance changes of cells in electric fields of different frequencies. However, the detection of a single electrical parameter is easily affected by the cell state and environmental noise, and it is difficult to fully reflect the multi-dimensional characteristics of cells, resulting in limited identification accuracy, especially when dealing with tumor cell subtypes with similar biological characteristics. There are significant bottlenecks.

[0005] In recent years, optical detection technology has attracted much attention due to its non-invasive and highly sensitive characteristics. Existing research has mostly focused on label-dependent methods such as fluorescent labeling or Raman spectroscopy. Although they can provide rich optical information, the labeling process may change cell activity, limiting subsequent in vivo analysis and clinical applications. Although label-free optical detection (such as scattered light analysis) has been explored, it is difficult to achieve high-specificity recognition in complex biological samples when used alone, and is easily interfered by factors such as cell size and morphology, making it difficult to meet the needs of precision medicine for high-precision detection.

[0006] It is worth noting that there is no public solution in the prior art that combines the label-free absorbance characteristics of cells with their electrical characteristics to achieve tumor cell detection. Electrical impedance and absorbance reflect the biological characteristics of cells from different dimensions, and the data fusion of the two is expected to break through the limitations of single parameter detection, but the relevant technical solutions and implementation methods are still blank.

[0007] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0008] Purpose of the invention: The technical problem to be solved by the present invention is to provide a particle detection device and method based on photoelectric sensor data fusion in view of the shortcomings of the prior art.

[0009] In order to solve the above technical problems, the present invention discloses a particle detection device and method based on photoelectric sensor data fusion, wherein the device comprises:

[0010] A microfluidic channel formed by the bottom plate and the cover plate, through which the cells to be detected pass;

[0011] The bottom of the microfluidic channel, i.e., the upper side of the bottom plate, is fixed with a first electrode, a second electrode and a third electrode in sequence for detecting the electrical characteristics generated by the cells;

[0012] The first optical fiber and the second optical fiber are respectively arranged at corresponding positions on the bottom plate and the cover plate, and the first optical fiber and the second optical fiber respectively send or receive light waves for detecting the optical characteristics of the cells.

[0013] Furthermore, the device also includes a controller for determining whether the cell is a tumor cell based on the electrical characteristics generated by the cell and the optical characteristics of the cell.

[0014] The present invention also proposes a particle detection method based on photoelectric sensor data fusion, which uses the above-mentioned detection device for detection, including the following steps:

[0015] Step 1, pre-treating a single type of cell sample, and obtaining the electrical and optical characteristic data of the cell through the detection device; repeating the above steps using more than two types of cell samples to form a training set;

[0016] Step 2, constructing a cell recognition model for cell recognition based on the electrical and optical characteristics of the cells;

[0017] Step 3, using the training set constructed in step 1, training and optimizing the cell recognition model constructed in step 2;

[0018] Step 4: Use a detection device to detect the cells to be detected to obtain their electrical and optical characteristics, obtain the identification results through the trained cell recognition model, and complete the tumor cell detection based on photoelectric sensor data fusion.

[0019] Further, the pre-processing described in step 1 includes:

[0020] The cells to be detected are pre-focused so that they are in a relatively fixed position in the detection device and queue up to enter the detection area of ​​the detection device in turn.

[0021] Furthermore, the step of obtaining the electrical characteristic data of the cell in step 1 includes:

[0022] On the middle electrode of the electrode group consisting of three electrodes, n frequencies of f are applied. 1 ,f 2 ,...,f n The mixed AC excitation electrical signal and the feedback current on the two electrodes on both sides are differentially and phase-lockedly amplified to obtain the impedance signal of the impedance detection area;

[0023] When only fluid flows at a constant speed and no cells pass through the electrode group, the obtained electrical impedance signal is a stable signal;

[0024] When a single cell passes through the electrode group, it disturbs the electrical impedance signal, and two electrical impedance signal peaks, one positive and one negative, are obtained in the time domain. The amplitude of the electrical impedance signal fluctuation value at each frequency is extracted and recorded as R 1 ,R 2 ,...,R n , and the phase, denoted as As the electrical characteristic data.

[0025] Furthermore, the step 1 of obtaining the optical characteristic data of the cell includes:

[0026] A stable light source with m preset wavelengths is emitted from one of the two optical fibers in the optical fiber group. After the light passes through the cells to be detected, it is received and returned by the other optical fiber for data analysis and recording to obtain the absorbance of the cells to be detected.

[0027] When no cells passed through the optical fiber group, stable absorbance data were obtained;

[0028] When the cells pass through the optical fiber group, the absorbance data fluctuates. Suppose the selected wavelengths are λ 1 ,λ 2 ,...,λ m The corresponding absorbance is recorded as A 1 ,A 2 ,...,A m , as the light characteristic data.

[0029] Furthermore, forming a training set as described in step 1 includes:

[0030] The electrical characteristic data and optical characteristic data of each cell in a single type of cell sample are fused as follows:

[0031]

[0032] Where P represents the photoelectric characteristic vector of one cell in the cell sample of this unit type;

[0033] Assume that the single type of cell sample contains N cells, and obtain the training set of this type of cells {P 1 ,P 2 ,...,P N};

[0034] Using another type of cell sample, detecting electrical characteristic data and optical characteristic data, and fusing them to obtain an optoelectronic characteristic vector of the other type of cell sample;

[0035] The above steps are repeated until the preset quantity requirement is met, and all cell sample types and corresponding photoelectric feature vectors are used as the training set.

[0036] Furthermore, the construction of the cell recognition model described in step 2 includes:

[0037] Step 2-1, use support vector machine to calculate the separation hyperplane equation between different types of cells. Assuming there are k types of cells to be identified, the hyperplane Γ ij A total of where i and j are cell type numbers, 1≤i≤k, 1≤j≤k, i≠j;

[0038] Step 2-2, for any photoelectric feature vector q of a cell to be identified, assume that the initial voting result is:

[0039] v=[v 1 ,v 2 ,…,v k ]=0

[0040] Among them, v k Indicates the number of votes obtained by the k-th posture;

[0041] Step 2-3, let the photoelectric eigenvector q and the hyperplane Γ ij The distance is d ij , voting is done based on distance, as follows:

[0042] When ij >0, the cell type is judged as the i-th type, v i =v i +1;

[0043] When ij <0, the cell type is judged as the jth type, v j =v j +1;

[0044] When ij =0, no judgment is made;

[0045] For the photoelectric characteristic vector q, make After voting, the updated voting result v is obtained;

[0046] Step 2-4, calculate As the type of cell represented by the photoelectric feature vector q, identification is completed; wherein, Indicates the type number x that makes its value the largest in v;

[0047] If there are the same type numbers in v that meet the above conditions, set them to g and h, that is, v g =v h , then make a judgment, if it satisfies:

[0048]

[0049] Then the cell is judged as type g, otherwise it is judged as type h; where d gr and d hs They represent the photoelectric eigenvector q and the hyperplane Γ respectively. gr and Γ hs The distance between them, r and s are the species numbers, Z + Represents a positive integer.

[0050] Furthermore, the mixed AC excitation electrical signal in step 1 is a multi-frequency mixed excitation electrical signal, including a multi-frequency AC electrical signal from a low frequency of 500kHz to a high frequency of 50MHz, and the peak-to-peak value of the mixed frequency electrical signal is less than 10V.

[0051] Furthermore, the first optical fiber and the second optical fiber are single-mode optical fibers.

[0052] Beneficial effects:

[0053] 1. The present invention uses both the electrical and optical properties of cells to identify cell types, which can reflect the characteristic differences of cells in multiple dimensions and has higher type identification accuracy;

[0054] 2. The present invention adopts non-labeling and non-destructive optical detection technology to replace flow cytometry that requires immunofluorescence labeling, and can still obtain living tumor cells after the detection is completed, so as to perform further biochemical detection or primary culture of the tumor cells. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments, and the above and / or other advantages of the present invention will become more clear.

[0056] Figure 1 This is the overall schematic diagram of the cell photoelectric fusion detection device.

[0057] Figure 2 This is a top view schematic diagram of the cell photoelectric fusion detection device.

[0058] Figure 3 This is a schematic diagram of the left side of the cell photoelectric fusion detection device.

[0059] Figure 4 It is a schematic diagram of the front side view of the cell photoelectric fusion detection device.

[0060] Figure 5a Schematic diagram of the confusion matrix for different cell types identified using photoelectric fusion data.

[0061] Figure 5b Schematic diagram of the confusion matrix for different cell types when using only electrical impedance data.

[0062] In the figure, 1 is the first electrode; 2 is the cell to be detected; 3 is the second electrode; 4 is the third electrode; 5 is the bottom plate; 6 is the first optical fiber; 7 is the microfluidic channel; 8 is the second optical fiber; and 9 is the cover plate. DETAILED DESCRIPTION

[0063] The general idea of ​​the present invention is as follows:

[0064] The electrical impedance detection technology of tumor cells is one of the important means in the field of non-label detection of tumor cells. Different types of tumor cells have certain differences in their electrical properties due to differences in their internal cell structure and molecular composition. The malignant phenotype of tumor cells leads to systematic changes in biophysical properties (membrane structure, ion homeostasis) and biochemical composition (macromolecule concentration), so they will show differences in optical and electrical properties from white blood cells in the blood. This difference can provide a technical basis for the detection of rare circulating tumor cells in the blood. Because tumor cells have increased cell membrane permeability and abnormal intracellular ion concentrations (such as K + / Na + The ratio of white blood cells to tumor cells is unbalanced), and the amplitude is 20% to 40% higher than that of white blood cells at low-frequency electrical impedance (<5MHz), and the high-frequency phase shift (>20MHz) is more obvious, reflecting the abnormal nuclear-cytoplasmic ratio and the difference in the dielectric properties of organelles; in addition, tumor cells have a higher nucleic acid / protein content (such as abnormal DNA copy number), resulting in an absorbance of 15% to 30% stronger than that of white blood cells in the near-infrared band (800-1200nm), and a characteristic absorption peak appears at a specific wavelength (such as 1000nm), which is significantly different from the spectral fingerprint of white blood cells. Therefore, the use of the optical and electrical properties of cells to detect and count rare circulating tumor cells can effectively improve the identification accuracy compared to the detection of tumor cells using a single parameter or characteristic.

[0065] The present invention combines electrical impedance detection with cell non-label absorbance detection to perform non-destructive and high-precision detection of cell types. In order to further improve the recognition accuracy of non-label non-destructive detection of tumor cells, it is necessary to combine the multi-dimensional physical property data of cells for type identification. The present invention proposes a method for tumor cell type identification based on photoelectric sensor data fusion. The specific technical solution is as follows: a particle detection device and method based on photoelectric sensor data fusion. The focus is on the extraction of multi-dimensional data and the training of machine learning models.

[0066] like Figure 1 As shown, first, the cell should undergo a pre-focusing process before entering the photoelectric detection area, so that the cell 2 occupies a relatively fixed position in the cross section of the microchannel 7 to improve the accuracy of the photoelectric signal detection. In addition, the pre-focusing process should make the cells queue up and enter the photoelectric detection area in turn to avoid the situation where multiple cells enter the photoelectric detection area at the same time. The present invention does not limit the specific cell pre-focusing method. Passive inertial or viscoelastic flow field forces can be used to focus the cells in the microchannel, or sheath fluid flow in the width and height directions of the channel can be used to focus the cells. Active focusing means can also be used, such as using ultrasound to form standing waves in the channel, generating acoustic radiation force, pushing the cells to the standing wave nodes or anti-nodes to achieve focusing, or using dielectrophoretic forces to manipulate the cells so that the cells form a focus in the channel.

[0067] like Figure 3 The microfluidic chip for detecting cell photoelectric characteristics is composed of a bottom plate 5 and an upper cover plate 9 combined to form a microchannel 7, such as Figure 2 As shown, there is a group of electrodes 1 / 3 / 4 made by micro-nano processing technology between the bottom plate 5 and the microchannel 7.

[0068] The present invention does not limit the material of the microelectrode, and preferably uses Cr as the substrate and Au as the upper electrode layer to improve the chemical stability of the electrode.

[0069] like Figure 4 As shown, the bottom plate 5 and the upper cover plate 9 have a group of aligned optical fiber insertion holes for installing the optical fiber 6 and the optical fiber 8 for optical detection.

[0070] The pre-focused cells then pass through the photodetection area.

[0071] The present invention does not limit the order of electrical impedance detection and optical detection.

[0072] The frequency f is applied to the middle electrode 3 1 ,f 2 ,...,f nThe mixed AC excitation electrical signal, the feedback current on electrode 1 and electrode 4 is differentially and phase-lockedly amplified to obtain the impedance signal of the impedance detection area. When only the fluid flows at a uniform speed and no cells pass through this area, the obtained impedance signal is a stable signal. After a single cell passes through the cell impedance detection area composed of the microelectrode group and the microchannel 7, the impedance signal of this area will be disturbed, and two impedance signal peaks, one positive and one negative, will be obtained in the time domain. The amplitude R of the impedance signal fluctuation value at each frequency is extracted 1 ,R 2 ,...,R n and phase

[0073] The biological differences between different cell types not only lead to significant differences in their electrical impedance characteristics, but also lead to differences in their absorbance under different bands. Specifically, there are significant differences in the elemental composition, protein macromolecular structure, etc. of different types of tumor cells, resulting in different populations of cells to different wavelengths of light absorption. Since multimode optical fiber is usually only suitable for transmitting light with a wavelength of 850nm and 1300nm, the loss is large when transmitting light of other wavelengths, and its core generally has a larger diameter, usually 50μm or 62.5μm. The wavelength range applicable to single-mode optical fiber is wider, and has a smaller core diameter, which is easier to integrate in the microfluidic channel, so optical fiber 6 and optical fiber 8 are preferably single-mode optical fiber. A stable light source containing a specific wavelength range is emitted by a spectrophotometer, and microfluidic channel 7 is irradiated by optical fiber 6, and optical fiber 8 receives light and returns to the spectrophotometer for data analysis and recording. When no cells pass through the optical detection area, a stable absorbance of the cell suspension is obtained. When cells pass between optical fiber 6 and optical fiber 8, the absorbance of the flow channel fluctuates, and this fluctuation serves as one of the important bases for determining the type of cells.

[0074] A spectrophotometer is an analytical instrument based on the principle of light absorption. It is used to measure the degree of absorption of light of a specific wavelength by a substance. Its core principle is the Beer-Lambert Law:

[0075]

[0076] Where A is absorbance, I 0 is the incident light intensity, I is the transmitted light intensity, ε is the molar absorption coefficient (related to the substance and wavelength), c is the solution concentration, and l is the optical path. When light passes through the sample to be tested, light of a specific wavelength will be absorbed by the substance, and the concentration or composition of the sample can be qualitatively or quantitatively inferred based on the absorbance.

[0077] Selected wavelength λ 1 ,λ 2 ,...,λ m , the absorbance A corresponding to the cell was measured1 ,A 2 ,...,A m .

[0078] When the cell passes through the electrical impedance detection area and the optical detection area, the photoelectric characteristic vector of the single cell can be obtained.

[0079]

[0080] When constructing the training set of the cell recognition model, a single type of cell sample is centrifuged and resuspended to obtain a cell sample, and the resuspension solution is a phosphate buffer solution that dissolves hyaluronic acid; then the cell sample is injected into the photoelectric detection device at a fixed flow rate. The cells are focused in a single column in the flow channel through specific technical means; the electrical impedance detection instrument is used to perform electrical impedance detection on the cells after single column focusing, and the signal peak caused by the cells is extracted using a signal processing and analysis system to obtain the amplitude R 1 ,R 2 ,...,R n and phase Use a spectrophotometer to obtain the absorbance A corresponding to the cell at a specific wavelength 1 ,A 2 ,...,A m . Assume that the cell sample of this type contains N cells, then the training set of this type of cell {P 1 ,P 2 ,...,P N}. Repeat the above process to obtain a training set of two or more cell samples. Use the support vector machine (SVM) algorithm to calculate the separation hyperplane equation between different cells (reference: Ding S, Hua X, Yu J. An overview on nonparallel hyperplane support vector machine algorithms [J]. Neural computing and applications, 2014, 25: 975-982. and: Hao PY. Support vector classification with fuzzy hyperplane [J]. Journal of Intelligent & Fuzzy Systems, 2016, 30 (3): 1431-1443.). Assuming that there are k types of cells to be identified, the hyperplane equation Γ ij (1≤i≤k,1≤j≤k,i≠j) There are a total of For a feature vector q of a cell parameter to be identified, the initial voting result is v = [v 1 ,v 2 ,…,vk ]=0, assuming that it is consistent with the hyperplane Γ ij The distance is d ij (When d ij >0, the cell type is judged as the i-th type, v i =v i +1; when d ij <0, the cell type is judged as the jth type, v j =v j +1; when d ij = 0, no judgment is made), then the process is called a vote. vote. In the end, That is, the type of cell represented by the characteristic vector q. Make v g =v h , then calculate like If the cell is positive, it is judged to be of type g, otherwise it is judged to be of type h.

[0081] Embodiment 1:

[0082] A particle detection device and method based on photoelectric sensor data fusion, comprising: integrating a microelectrode group (including microelectrodes 1, 3, 4) and a micron-level optical fiber group (optical fiber 6, optical fiber 8) in a microchannel to detect the electrical and optical properties of a cell 2.

[0083] After the pre-focusing process, the cells 2 in the microchannel 7 are in a relatively fixed position in the cross section of the channel 7, which can improve the accuracy of electrical detection and optical detection.

[0084] A multi-frequency mixed excitation electrical signal is applied to electrode 3, and the feedback currents on electrode 1 and electrode 4 are differentiated and then passed through a phase-locked amplifier to obtain the electrical impedance spectrum of the cell. The multi-frequency mixed excitation electrical signal applied to electrode 3 includes a multi-frequency AC signal from a low frequency of 500kHz to a high frequency of 50MHz, and the peak-to-peak value of the mixed frequency electrical signal is less than 10V to avoid damage to the cells.

[0085] A near-infrared spectrophotometer is used to detect the spectral disturbance generated when cells pass through the optical detection area through the transmitting optical fiber 8 and the receiving optical fiber 6.

[0086] The multi-frequency electrical impedance data and spectral perturbation data of cells are fused together as the basis for identifying the type of tumor cells. The machine learning method is used to train the identification model using the collected training set data to make judgments on the types of cells to be identified.

[0087] Embodiment 2:

[0088] Polyethylene oxide (PEO) with a relative molecular mass of 1000 and phosphate buffered saline (PBS) were used to prepare a cell suspension with a PEO mass concentration of 0.1%. PEO solution is a non-Newtonian fluid that exerts viscoelastic forces on cells when flowing. Human white blood cells WBCs, human breast cancer cells MCF7, human lung adenocarcinoma cells A549, human lung squamous cell carcinoma cells H226, and human lung large cell carcinoma cells H460 were centrifuged and resuspended in 0.1% PEO solution. Different types of cell suspensions were respectively passed into a direct current channel with a width of 50 microns, a height of 40 microns, and a length of 3000 mm using a precision syringe pump at a flow rate of 30 microliters per minute. Under the coupling of inertial force and viscoelastic force, the cells will focus on the center of the flow channel cross section to improve the accuracy of the electrical impedance detection signal and the optical detection signal, and eliminate the uncertain influence of the relative position of the cells in the flow channel on the signal.

[0089] Mixed excitation AC signals with frequencies of 0.5 MHz, 1 MHz, 2 MHz, 4 MHz, 8 MHz, 16 MHz, 32 MHz, and 64 MHz were applied to the central electrode (3). The absorbance of the cells at wavelengths of 600 nm, 800 nm, 1000 nm, and 1200 nm was measured using a UV-visible-near infrared spectrophotometer. 50% of the data sets generated by the detection of human white blood cells WBCs, human breast cancer cells MCF7, human lung adenocarcinoma cells A549, human lung squamous cell carcinoma cells H226, and human lung large cell carcinoma cells H460 were randomly selected as training sets Ω 1 ,Ω 2 ,Ω 3 ,Ω 4 ,Ω 5 The remaining data is used as the test set. The training set is trained using the SVM algorithm, and the recognition results of the training model on the remaining test set are as follows: Figure 5a As shown in Figure 2. For comparison, the same training process was performed on cells using only electrical impedance data, and the recognition results for the test set are shown in Figure 2. Figure 5b As shown. It can be seen that the use of optoelectronic data fusion can significantly improve the accuracy of cell type identification.

[0090] In a specific implementation, the present application provides a computer storage medium and a corresponding data processing unit, wherein the computer storage medium can store a computer program, and when the computer program is executed by the data processing unit, the invention content of a particle detection device and method based on photoelectric sensor data fusion provided by the present invention and some or all steps in each embodiment can be executed. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0091] Those skilled in the art can clearly understand that the technical solutions in the embodiments of the present invention can be implemented by means of computer programs and their corresponding general hardware platforms. Based on such an understanding, the technical solutions in the embodiments of the present invention can be essentially or partly contributed to the prior art in the form of computer programs, i.e., software products, which can be stored in a storage medium and include several instructions for enabling a device including a data processing unit (which can be a personal computer, a server, a single-chip microcomputer, an MCU or a network device, etc.) to execute the methods described in various embodiments of the present invention or certain parts of the embodiments.

[0092] The present invention provides a particle detection device and method based on photoelectric sensor data fusion. There are many methods and ways to implement the technical solution. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the protection scope of the present invention. All components not specified in this embodiment can be implemented by existing technologies.

Claims

1. A particle detection device based on photoelectric sensor data fusion, characterized in that: include: A microfluidic channel (7) formed by the bottom plate (5) and the cover plate (9) and used for the cells (2) to be detected to pass through; The bottom of the microchannel (7), i.e., the upper side of the bottom plate (5), is fixedly provided with a first electrode (1), a second electrode (3), and a third electrode (4) in sequence, for detecting the electrical characteristics generated by the cell (2); A first optical fiber (6) and a second optical fiber (8) are respectively arranged at corresponding positions on the base plate (5) and the cover plate (9), and the first optical fiber (6) and the second optical fiber (8) respectively send or receive light waves for detecting the optical characteristics of the cells (2).

2. A particle detection device based on photoelectric sensor data fusion according to claim 1, characterized in that: The device also includes a controller for determining whether the cell (2) is a tumor cell based on the electrical characteristics generated by the cell (2) and the optical characteristics of the cell (2).

3. A particle detection method based on photoelectric sensor data fusion, characterized in that: The detection device according to claim 1 or 2 is used for detection, comprising the following steps: Step 1, pre-treating a single type of cell sample, and obtaining the electrical and optical characteristic data of the cell through the detection device; repeating the above steps using more than two types of cell samples to form a training set; Step 2, constructing a cell recognition model for cell recognition based on the electrical and optical characteristics of the cells; Step 3, using the training set constructed in step 1, training and optimizing the cell recognition model constructed in step 2; Step 4: Use a detection device to detect the cells to be detected to obtain their electrical and optical characteristics, obtain the identification results through the trained cell recognition model, and complete the tumor cell detection based on photoelectric sensor data fusion.

4. The particle detection method based on photoelectric sensor data fusion according to claim 3 is characterized in that: Perform the pre-processing described in step 1, including: The cells to be detected are pre-focused so that they are in a relatively fixed position in the detection device and queue up to enter the detection area of ​​the detection device in turn.

5. The particle detection method based on photoelectric sensor data fusion according to claim 3, characterized in that: The step 1 of obtaining electrical characteristic data of cells includes: On the middle electrode of the electrode group consisting of 3 electrodes, n frequencies f1, f2, ..., f n The mixed AC excitation electrical signal and the feedback current on the two electrodes on both sides are differentially and phase-lockedly amplified to obtain the impedance signal of the impedance detection area; When only fluid flows at a constant speed and no cells pass through the electrode group, the obtained electrical impedance signal is a stable signal; When a single cell passes through the electrode group, it disturbs the impedance signal, and two impedance signal peaks, one positive and one negative, are obtained in the time domain. The amplitude of the impedance signal fluctuation value at each frequency is extracted and recorded as R1, R2, ..., R n , and the phase, denoted as As the electrical characteristic data.

6. The particle detection method based on photoelectric sensor data fusion according to claim 5, characterized in that: The step 1 of obtaining optical characteristic data of cells includes: A stable light source with m preset wavelengths is emitted from one of the two optical fibers in the optical fiber group. After the light passes through the cells to be detected, it is received and returned by the other optical fiber for data analysis and recording to obtain the absorbance of the cells to be detected. When no cells passed through the optical fiber group, stable absorbance data were obtained; When the cells pass through the optical fiber group, the absorbance data fluctuates. Suppose the selected wavelengths are λ1, λ2, ..., λ m The corresponding absorbances are recorded as A1, A2, ..., A m , as the light characteristic data.

7. The particle detection method based on photoelectric sensor data fusion according to claim 6, characterized in that: The training set described in step 1 is formed, including: The electrical characteristic data and optical characteristic data of each cell in a single type of cell sample are fused as follows: Where P represents the photoelectric characteristic vector of one cell in the cell sample of this unit type; Assume that the single cell type sample contains N cells, and obtain the training set of cells of this type {P1, P2, ..., P N }; Using another type of cell sample, detecting electrical characteristic data and optical characteristic data, and fusing them to obtain an optoelectronic characteristic vector of the other type of cell sample; The above steps are repeated until the preset quantity requirement is met, and all cell sample types and corresponding photoelectric feature vectors are used as the training set.

8. The particle detection method based on photoelectric sensor data fusion according to claim 7, characterized in that: The step 2 of constructing a cell recognition model comprises: Step 2-1, use support vector machine to calculate the separation hyperplane equation between different types of cells. Assuming there are k types of cells to be identified, the hyperplane Γ ij A total of where i and j are cell type numbers, 1≤i≤k, 1≤j≤k, i≠j; Step 2-2, for any photoelectric feature vector q of a cell to be identified, assume that the initial voting result is: v=[v1,v2,…,v k ]=0 Among them, v k Indicates the number of votes obtained by the k-th posture; Step 2-3, let the photoelectric eigenvector q and the hyperplane Γ ij The distance is d ij , voting is done based on distance, as follows: When ij >0, the cell type is judged as the i-th type, v i =v i +1; When ij <0, the cell type is judged as the jth type, v j =v j +1; When ij =0, no judgment is made; For the photoelectric characteristic vector q, make After voting, the updated voting result v is obtained; Step 2-4, calculate As the type of cell represented by the photoelectric feature vector q, identification is completed; wherein, Indicates the type number x that makes its value the largest in v; If there are the same type numbers in v that meet the above conditions, set them to g and h, that is, v g =v h , then make a judgment, if it satisfies: Then the cell is judged as type g, otherwise it is judged as type h; where d gr and d hs They represent the photoelectric eigenvector q and the hyperplane Γ respectively. gr and Γ hs The distance between them, r and s are the species numbers, Z + Represents a positive integer.

9. The particle detection method based on photoelectric sensor data fusion according to claim 5, characterized in that: The mixed AC excitation electrical signal described in step 1 is a multi-frequency mixed excitation electrical signal, including a multi-frequency AC electrical signal from a low frequency of 500kHz to a high frequency of 50MHz, and the peak-to-peak value of the mixed frequency electrical signal is less than 10V.

10. The particle detection device based on photoelectric sensor data fusion according to claim 1, characterized in that: The first optical fiber (6) and the second optical fiber (8) are single-mode optical fibers.

Images