Inertial microfluidic-based device and method for detecting mechanical properties of tumor cells

Abstract

The present application relates to the technical field of tumor cell detection chip, in particular to a kind of based on inertial microfluidic tumor cell mechanical property detection device and method, the present application adopts sine microfluid channel inertial focusing cell, and the mechanical property detection of cell multiple deformation in channel is carried out by high-speed camera.The cell suspended in PBS is subjected to the combined action of inertial force and dean force in sine microfluid channel, and the single-column focusing of cell can be realized in a wide flow rate range, and the recognition accuracy of optical detection is improved.Cell is subjected to shear force, tensile force and wall collision in microfluid channel and generates three deformations, and different types and sizes of force make cell produce different degrees and shapes of deformation.The size, roundness, deformation and circumscribed rectangle size of cell in different deformations are extracted, and the classification model of species is trained using machine learning method, to identify cell species.The present application is suitable for the detection of various cells in blood samples and biological particles in other biological samples.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of tumor cell detection chip technology, in particular to a device and method for detecting mechanical properties of tumor cells based on inertial microfluidics. BACKGROUND

[0002] Cancer is a major global public health problem, and the situation in China is even more severe, with gradually increasing incidence and mortality rates. In 2022, there were about 4.82 million new cancer cases in China, and 3.21 million cancer deaths, with a mortality rate of more than 65% for cancer patients. Therefore, research on the diagnosis and treatment of cancer has very important scientific and social significance.

[0003] Clinical data show that the main cause of clinical death in cancer patients is the invasion and metastasis of tumor cells derived from the primary tumor. Therefore, by detecting tumor cells in human blood or pleural effusion, early diagnosis, real-time monitoring, drug screening, and effective prognosis evaluation can be achieved, which helps to control the condition of cancer patients and reduce the mortality rate due to illness, and is expected to become a new means for early diagnosis of cancer, with important scientific and social value.

[0004] Biophysical properties are an inherent indicator of cell structure, function, and pathological state, and different origins and functional structures result in differences in biophysical properties such as morphology, size, mechanical properties, and electrical properties between blood cells, normal cells, and tumor cells. Combined with microfluidic technology, image analysis technology, or impedance analysis technology, it is a non-labeled method that can characterize single-cell mechanical properties. The detection of tumor cells using the differences in cell mechanical properties can successfully avoid the difficulties of biochemical labeling methods that cause cells to lose activity. In recent years, preliminary progress has been made in the use of microfluidic deformation cell instruments to distinguish between blood cells, normal cells, and tumor cells. However, single deformation detection technology still has the problem of limited detection accuracy. SUMMARY

[0005] The present application aims to address the problems in the background art by providing a device and method for detecting mechanical properties of tumor cells based on inertial microfluidics, which improves the detection accuracy and throughput of tumor cells.

[0006] The technical solution of the present application is a device for detecting mechanical properties of tumor cells based on inertial microfluidics, which includes a sample pretreatment device, a sample inlet device, an inverted microscope, a mechanical property detection chip, a high-speed camera, a computer, and a signal processing and analysis system.

[0007] After the cell sample is treated by the sample pretreatment device, it is sent into the mechanical property detection chip placed on the workbench through the sample inlet device;

[0008] The inverted microscope is installed on a workbench and faces the mechanical property detection chip;

[0009] The inverted microscope and the high-speed camera detect the mechanical properties of the cells on the mechanical property detection chip, and analyze and learn through the signal processing analysis system loaded on the computer to obtain a cell identification model; and the cell sample is identified according to the obtained cell identification model.

[0010] Preferably, the mechanical property detection chip comprises a sample inlet tube, a liquid inlet pool, a chip cover plate, a glass substrate, a sinusoidal flow channel, a deformation flow channel, a waste liquid pool and a sample outlet tube.

[0011] The glass substrate is located at the bottom, and the liquid inlet pool, the sinusoidal flow channel, the deformation flow channel and the waste liquid pool are all arranged at the bottom of the chip cover plate; and the chip cover plate covers the glass substrate.

[0012] The sample inlet tube is connected with the inlet of the liquid inlet pool; the sinusoidal flow channel and the deformation flow channel are sequentially arranged between the liquid inlet pool and the waste liquid pool; and the sample outlet tube is connected with the outlet of the waste liquid pool.

[0013] Preferably, the deformation flow channel comprises a contraction structure, a cross structure and a T-shaped structure; two groups of flow channels are symmetrically arranged at the two sides of the contraction structure, the flow channels converge with the main flow channel to form the cross structure; and the outlet of the cross structure is connected with the T-shaped structure.

[0014] Preferably, the sinusoidal flow channel focuses the cell sample in a single column.

[0015] Preferably, the cell focused by the sinusoidal flow channel first flows through the contraction structure to generate the first shear deformation, the cell-free fluid flows out from the flow channels at the two sides of the contraction structure to the cross structure to make the cell generate the second deformation, and the focused cell of the converged fluid generates the stretching deformation in the T-shaped structure.

[0016] Preferably, the inner and outer wall radii of the sinusoidal flow channel are 140 μm and 240 μm respectively.

[0017] The length and width dimensions of the contraction structure are 500 μm and 20 μm respectively.

[0018] The flow channel widths of the cross structure in the horizontal and vertical directions are 60 μm and 27.6 μm respectively.

[0019] The flow channel widths of the T-shaped structure in the horizontal and vertical directions are 30 μm and 25 μm respectively.

[0020] The height of the mechanical property detection chip is 20 μm.

[0021] A tumor cell mechanical property detection method based on inertial microfluidics, which is detected by using the above tumor cell mechanical property detection device, and comprises the following specific steps:

[0022] S1: centrifuging, resuspending and treating normal cell strains to obtain normal cell samples;

[0023] S2: passing the normal cell strain samples into a mechanical property detection chip, and making the cell samples flow through the sine flow channel in the mechanical property detection chip for single column focusing;

[0024] S3: after the normal cell strain samples are focused through the sine flow channel, the cell samples enter the contraction structure, cross structure and T-shaped structure of the deformation flow channel, and the cell samples are deformed three times;

[0025] S4: setting the parameters of the inverted microscope and the high-speed camera, and acquiring the cell image information entering the deformation flow channel through the high-speed camera;

[0026] S5: using a signal processing analysis system to extract the profile of the cell image information acquired in S4, extract the cell profile data value, and couple the data information extracted in the three deformations;

[0027] S6: repeating S1-S5 to detect the cancer cell strains respectively;

[0028] S7: inputting the mechanical properties of the two cell samples in different deformations into a machine learning system to train a cell identification model;

[0029] S8: using the cell identification model to identify the cell types of the two cell samples.

[0030] Preferably, the resuspension solution in S1 is PBS.

[0031] Preferably, the cell type identification model is trained by using the Ansys machine learning method in S7.

[0032] Compared with the prior art, the present application has the following beneficial technical effects:

[0033] 1. The present application uses a sine flow channel to focus the measured cells in the center to improve the detection accuracy of mechanical properties, solves the problem of uneven stress on cells caused by the random distribution of cells in the detection area flow channel, and improves the detection accuracy of mechanical properties.

[0034] 2、The cells focused by the sinusoidal flow channel first flow through the contraction structure to occur the first shear deformation, the fluid without cells flows out from both sides of the contraction structure to the cross structure to make the cells occur the second deformation, the fluid after the convergence further focuses the cells and generates greater flow rate to make the cells occur greater tensile deformation in the T-shaped structure. The advantage of the deformation region design is that it does not need to add unconventional medium components to enhance cell deformation in the first deformation, and does not need to increase the sheath liquid input in the second deformation. In addition, the cells are subjected to different degrees of stress in the three deformations, so that the multi-dimensional parameters obtained in the cell deformation detection have more statistical significance.

[0035] 3、The present application utilizes different detection structures to make the cells deform at the contraction structure, the cross structure and the T-shaped structure respectively, and obtains mechanical property parameters such as cell size, roundness, deformability and circumscribed rectangle size in different deformations. The multiple parameters are used for machine learning training model, and then the cell identification model is used for identification analysis of two kinds of cell samples, so that the identification accuracy is greatly improved.

[0036] 4、The present application adopts multiple deformations for mechanical property detection of cells, and uses machine learning method to establish a cell type identification model for cell type identification in real samples, which can be used for detection of various cells in blood, and can also be used for detection of biological particles in other biological samples, and has important value and commercial prospect. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 is a mechanical property detection system diagram based on the detection method of the present application;

[0038] Figure 2 is a mechanical property detection chip structure schematic diagram;

[0039] Figure 3 is a flowchart of the detection method of the present application;

[0040] Figure 4 is a structure schematic diagram of the asymmetric sinusoidal flow channel in the specific embodiment of the present application;

[0041] Figure 5 is a superimposed diagram of the focusing condition of the microspheres suspended in the PBS solution flowing through the sinusoidal flow channel in the specific embodiment of the present application;

[0042] Figure 6 is a superimposed diagram of the focusing condition of the microspheres suspended in the PBS solution flowing out of the sinusoidal flow channel in the specific embodiment of the present application;

[0043] Figure 7 is a superimposed diagram of the deformation condition of the cells suspended in the PBS solution flowing through the deformation flow channel in the specific embodiment of the present application;

[0044] Figure 8 This is a confusion matrix diagram of the machine learning model in a specific embodiment of the present invention.

[0045] Reference numerals: 1. Sample pretreatment device; 2. Sample injection device; 3. Inverted microscope; 4. Mechanical property detection chip; 5. High-speed camera; 6. Computer; 7. Signal processing and analysis system; 40. Sample injection tube; 41. Liquid injection tank; 42. Chip cover plate; 43. Glass substrate; 44. Sine flow channel; 45. Shrinkage structure; 46. Cross structure; 47. T-shaped structure; 48. Waste liquid tank; 49. Sample outlet tube. Detailed Implementation

[0046] like Figure 1 As shown in this embodiment, a tumor cell mechanical property detection device based on inertial microfluidics is characterized by comprising a sample pretreatment device 1, a sample injection device 2, an inverted microscope 3, a mechanical property detection chip 4, a high-speed camera 5, a computer 6, and a signal processing and analysis system 7.

[0047] After being processed by the sample pretreatment device 1, the cell sample is sent into the mechanical property detection chip 4 placed on the worktable via the sample injection device 2.

[0048] The inverted microscope 3 is mounted on the worktable and faces the mechanical property detection chip 4.

[0049] An inverted microscope 3 and a high-speed camera 5 detect the mechanical properties of cells on a mechanical property detection chip 4, and the signals are analyzed and learned by a signal processing and analysis system 7 mounted on a computer 6 to obtain a cell identification model; the cell sample is then identified based on the obtained cell identification model.

[0050] like Figure 2 As shown, the mechanical property detection chip 4 includes a sample inlet tube 40, a liquid inlet pool 41, a chip cover plate 42, a glass substrate 43, a sinusoidal flow channel 44, a deformation flow channel, a waste liquid pool 48, and a sample outlet tube 49.

[0051] The glass substrate 43 is located at the bottom, and the liquid inlet 41, the sinusoidal flow channel 44, the deformation flow channel and the waste liquid tank 48 are all located at the bottom of the chip cover plate 42; the chip cover plate 42 covers the glass substrate 43.

[0052] The sample inlet tube 40 is connected to the inlet of the liquid inlet pool 41; the sinusoidal flow channel 44 and the deformation flow channel are arranged sequentially between the liquid inlet pool 41 and the waste liquid pool 48; the sample outlet tube 49 is connected to the outlet of the waste liquid pool 8; the deformation flow channel includes a contraction structure 45, a cross structure 46 and a T-shaped structure 47; two sets of symmetrical flow channels are arranged on both sides of the contraction structure 45, and the flow channels merge with the main flow channel to form the cross structure 46; the outlet of the cross structure 46 is connected to the T-shaped structure 47.

[0053] The sinusoidal flow channel 44 focuses the cell sample in a single column.

[0054] The cells focused by the sinusoidal flow channel 44 first flow through the contraction structure 45 to be subjected to a first shear deformation, the cell-free fluid flows out of the flow channels on both sides of the contraction structure to the cross structure 46 to be subjected to a second deformation, and the focused fluid is subjected to a stretching deformation at the T-shaped structure 47.

[0055] A tumor cell mechanical property detection method based on inertial microfluidics, which is detected by using the tumor cell mechanical property detection device, comprises the following specific steps:

[0056] S1: centrifuging and resuspending a normal cell strain to obtain a normal cell sample;

[0057] S2: passing the normal cell strain sample into the mechanical property detection chip 4, and making the cell sample flow through the sinusoidal flow channel 44 in the mechanical property detection chip 4 to be focused in a single column;

[0058] S3: after the normal cell strain sample is focused by the sinusoidal flow channel 44, the sample enters the contraction structure 45, the cross structure 46 and the T-shaped structure 47 of the deformation flow channel, and the cell sample is subjected to three deformations;

[0059] S4: setting parameters of the inverted microscope 3 and the high-speed camera 5, and acquiring image information of the cells entering the deformation flow channel by the high-speed camera 5;

[0060] S5: using a signal processing and analysis system 7 to extract the profile of the image information of the cells acquired in S4, extracting cell profile data values, and coupling the data information extracted in the three deformations;

[0061] S6: repeating S1-S5 to detect cancer cell strains respectively;

[0062] S7: inputting the mechanical properties of the two cell samples in different deformations into a machine learning system to train a cell identification model;

[0063] S8: using the cell identification model to identify the cell types of the two cell samples.

[0064] In an optional embodiment, the resuspension solution in S1 is PBS.

[0065] In an optional embodiment, the cell type identification model is trained by using an Ansys machine learning method in S7.

[0066] Embodiment 1

[0067] The principles and effects of the present application will be described below by taking the focusing of standard polystyrene microspheres and the mechanical property detection of cells as examples.

[0068] Polystyrene particles with diameters of 15 μm and 20 μm, and cells of human breast cancer cell line MDA-MB-231 and normal human breast cell line MCF-10A.

[0069] As a preferred mode, as shown in Figure 4 the inner and outer wall radii of the sinusoidal flow channel outlet 44 are 140 μm and 240 μm respectively; the length and width dimensions of the downstream contraction structure 45 are 500 μm and 20 μm respectively; the flow channel widths of the horizontal and vertical directions of the cross structure 46 are 60 μm and 27.6 μm respectively; the flow channel widths of the horizontal and vertical directions of the T-shaped structure 47 are 30 μm and 25 μm respectively; and the height of the entire mechanical property detection chip 4 is 20 μm.

[0070] Polystyrene microspheres with diameters of 15 μm and 20 μm were suspended in PBS solution, and the sample was passed into the mechanical property detection chip 4 at a flow rate of 200 μL / min through the liquid sampling device 2.

[0071] Figure 5 The particle focusing condition superimposed graph for the microspheres suspended in PBS solution flowing through the sinusoidal flow channel 44. Figure 6 The particle focusing condition superimposed graph for the microspheres suspended in PBS solution flowing out of the sinusoidal flow channel 44.

[0072] It can be seen that the cells suspended in PBS solution can achieve single-file focusing of cells in a wide flow rate range under the combined action of inertial lift force and Dean drag force in the sinusoidal flow channel. The inertial lift force is the resultant force of shear-induced inertial lift force and wall-induced inertial lift force.

[0073] F L = f L ρU 2 a p 4 / D h (1)

[0074] In formula (1), f L is the lift coefficient, ρ is the solution density, U is the average flow rate of the solution, a p is the particle diameter, D h is the hydraulic diameter, and D h = 2wh / (w+h), w and h are the width and height of the flow channel respectively.

[0075] The Dean drag force FD is generated by the secondary flow in the curved flow channel,

[0076] F D ≈ ρU 2 a p D h2 r (2)

[0077] In formula (2), r is the curvature radius of the flow channel.

[0078] F L The size and proportion of F D determine the focusing mode of the particles and the transverse position of the particles in the flow channel.

[0079] Under the combined action of the above two forces, the particles finally balance at a single equilibrium position on the cross section in the flow channel, thereby realizing single-column focusing of the particles.

[0080] MDA-MB-231 cells were floated in PBS solution, and the sample was passed into the mechanical property detection chip 4 at a flow rate of 200 μL / min through the liquid injection device 2.

[0081] Figure 7 The superimposed graph of the deformation of the cells suspended in the PBS solution flowing through the deformation flow channel.

[0082] It can be seen that the cells suspended in the PBS solution successively undergo hydrodynamic shear, hydrodynamic stretching and wall collision deformation in the deformation flow channel. Different stresses cause different deformations of the cells in different forms and degrees.

[0083] Human breast cancer cell line MDA-MB-231 and normal human breast cells MCF-10A were used as cell samples, and the detection method in the above embodiment was implemented. The flow rate used was 200 μL / min. The mechanical property changes caused by the deformation of the cells in different deformations were extracted, and a classification model of the cell species was trained using machine learning method, which was used for cell identification in different cell samples. Figure 8 The confusion matrix diagram of the machine learning classification model of MDA-MB-231 and MCF-10A cells. As can be seen from the diagram, the identification accuracy of MDA-MB-231 can reach 92.8%, and the identification accuracy of MCF-10A can reach 87.7%.

[0084] The embodiments of the application are described in detail above in combination with the drawings, but the application is not limited thereto, and various changes can be made within the knowledge of those skilled in the art without departing from the purpose of the application.

Claims

1. A device for detecting the mechanical properties of tumor cells based on inertial microfluidics, characterized in that, It includes a sample pretreatment device (1), a sample injection device (2), an inverted microscope (3), a mechanical property detection chip (4), a high-speed camera (5), a computer (6), and a signal processing and analysis system (7); After being processed by the sample pretreatment device (1), the cell sample is sent into the mechanical property detection chip (4) placed on the workbench by the sample injection device (2); An inverted microscope (3) is mounted on the worktable and faces the mechanical property detection chip (4). An inverted microscope (3) and a high-speed camera (5) are used to detect the mechanical properties of the cells on the mechanical property detection chip (4), and the signal processing and analysis system (7) mounted on the computer (6) is used to analyze and learn the data to obtain a cell identification model; the cell sample is then identified based on the obtained cell identification model. The mechanical property detection chip (4) includes an injection tube (40), a liquid inlet (41), a chip cover plate (42), a glass substrate (43), a sinusoidal flow channel (44), a deformation flow channel, a waste liquid tank (48), and an outlet tube (49). The glass substrate (43) is located at the bottom, and the liquid inlet (41), sinusoidal flow channel (44), deformation flow channel, and waste liquid tank (48) are all located at the bottom of the chip cover plate (42). The chip cover plate (42) covers the glass substrate (43). The injection tube (40) is connected to the inlet of the liquid inlet (41). The sinusoidal flow channel (44) and the deformation flow channel are arranged sequentially between the liquid inlet (41) and the waste liquid tank (48). The outlet tube (49) is connected to the outlet of the waste liquid tank (48). The deformable flow channel includes a contraction structure (45), a cross structure (46), and a T-shaped structure (47); two sets of symmetrical flow channels are set on both sides of the contraction structure (45), and the flow channels merge with the main flow channel to form a cross structure (46); the outlet of the cross structure (46) is connected to the T-shaped structure (47); the sinusoidal flow channel (44) performs single-row focusing on the cell sample.

2. The tumor cell mechanical property detection device based on inertial microfluidics according to claim 1, characterized in that, Cells focused by the sinusoidal flow channel (44) first flow through the contraction structure (45) and undergo the first shear deformation. Cellless fluid flows out from the flow channels on both sides of the contraction structure and merges with the cross structure (46) to cause the cells to undergo the second deformation. After merging, the fluid focuses on the cells and causes them to undergo stretching deformation in the T-shaped structure (47).

3. The tumor cell mechanical property detection device based on inertial microfluidics according to claim 1, characterized in that, The inner and outer wall radii of the sinusoidal flow channel (44) are 140 μm and 240 μm, respectively; The length and width dimensions of the contraction structure (45) are: 500 μm and 20 μm; The horizontal and vertical channel widths of the cross structure (46) are 60 μm and 27.6 μm, respectively; The horizontal and vertical flow channel widths of the T-shaped structure (47) are 30 μm and 25 μm, respectively; The height of the mechanical property detection chip (4) is 20 μm.

4. A method for detecting the mechanical properties of tumor cells based on inertial microfluidics, wherein the detection is performed using the tumor cell mechanical property detection device as described in any one of claims 1-3, characterized in that, The specific steps include the following: S1: Centrifuge and resuspend normal cell lines; S2: Pass the normal cell line sample into the mechanical property detection chip (4) and make the cell sample flow through the sinusoidal flow channel (44) in the mechanical property detection chip (4) for single-row focusing; S3: After being focused by the sinusoidal flow channel (44), the normal cell line sample enters the contraction structure (45), cross structure (46), and T-shaped structure (47) of the deformation flow channel, and the cell sample undergoes three deformations. S4: Set the parameters of the inverted microscope (3) and the high-speed camera (5), and obtain the image information of the cells entering the deformation channel through the high-speed camera (5); S5: Use the signal processing analysis system (7) to extract the contour of the cell image information obtained in S4, extract the cell contour data value, and couple the data information extracted by the three deformations. S6: Repeat S1-S5 to detect cancer cell lines separately; S7: Input the mechanical properties of the two cell samples under different deformations into the machine learning system to train a model for cell identification; S8: Use a cell identification model to identify the cell types of two cell samples.

5. The method for detecting the mechanical properties of tumor cells based on inertial microfluidics according to claim 4, characterized in that, The resuspension in S1 is PBS.

6. The method for detecting the mechanical properties of tumor cells based on inertial microfluidics according to claim 4, characterized in that, Ansys machine learning methods were used in S7 to train a cell species identification model.

Images