Cell identification method, device and system

Abstract

In order to solve the problem of cell identification, the present invention provides a method for cell identification, which includes the steps of acquiring cell information including cell traction force information at a specific point within a cell acquired based on a cell force sensor (the cell traction force information includes the magnitude of the cell traction force at the relevant point), preprocessing the acquired cell information to form structured cell information, the structured cell information including the number of cells, the number of cell features, and feature information of each cell feature, and using the structured cell information as input data to establish a cell feature model using supervised learning, unsupervised learning, or semi-supervised learning, and applying the cell feature model to classification or clustering of cells of an unknown type or unknown state. The present invention also provides a cell identification device and a cell identification system for implementing the cell identification method, which have high throughput, high resolution, and are capable of measuring and analyzing living single cells in real time.

Description

[Technical Field]

[0001] The present invention relates to the field of cell analysis and identification, and in particular to methods, devices and systems for cell identification. [Background technology]

[0002] Currently, single-cell RNA sequencing (scRNA-seq) is the mainstream quantitative analysis technology at the single-cell level. Its unique feature is that it quantitatively analyzes the transcripts of a single cell. However, a drawback is that it is essentially a snapshot of a single cell, and is an invasive and cell-damaging method, making it impossible to detect the same cell in real time. Other methods, such as immunofluorescence, require cell staining, which inevitably affects or damages the cells, and the process is expensive and complicated. Summary of the Invention

[0003] Therefore, there is a need to provide a technology that has high-throughput, high-resolution characteristics and can measure and analyze living single cells in real time, solve the problem of cell identification, and further facilitate scientific research, drug development, and clinical applications.

[0004] The inventors provide a method for cell identification comprising the following steps. acquiring cell information, the cell information including cell traction force information at a specific point within the cell acquired based on a cell force sensor, the cell traction force information including the magnitude of the cell traction force at the point; a step of preprocessing the cell information to form structured cell information, the structured cell information including the number of cells, the number of cell features, and feature information of each cell feature; A method for cell identification, comprising the steps of using the structured cell information as input data, establishing a cell feature model using supervised learning, unsupervised learning, or semi-supervised learning, and applying the cell feature model to classify or cluster cells of unknown type or unknown state.

[0005] Furthermore, in the cell identification method, the cell traction force information also includes the direction of the cell traction force at that point.

[0006] Furthermore, in the cell identification method, the cell traction force information also includes a change in the magnitude or direction of the cell traction force at that point within a predetermined time interval.

[0007] Furthermore, in the cell identification method, the cell information also includes cell morphology information.

[0008] Furthermore, in the cell identification method, the cell information is obtained after performing a cell control operation on the cell.

[0009] The inventor also provides an apparatus for cell identification, which includes an information acquisition unit, a pre-processing unit, a learning unit, and an identification unit. the information acquisition unit is used to acquire cell information, the cell information including cell traction force information at a specific point within the cell acquired based on a cell force sensor, and the cell traction force information including the magnitude of the cell traction force at the point; The pre-processing unit is used to pre-process the cell information and form structured cell information, and the structured cell information includes the number of cells, the number of cell features, and feature information of each cell feature. the learning unit is used to use the structured cell information as input data to establish a cell feature model using supervised learning, unsupervised learning, or semi-supervised learning; The identification unit is a cell identification device used to apply the cell feature model to classify or cluster cells of an unknown type or state.

[0010] Furthermore, the cell traction force information also includes the direction of the cell traction force at that point.

[0011] Furthermore, the cell traction force information also includes a change in the magnitude or direction of the cell traction force at that point within a predetermined time interval.

[0012] Furthermore, the cell information is characterized by including cell morphology information.

[0013] Furthermore, the cell information is characterized in that it is acquired after performing a cell control operation on the cell.

[0014] The inventor also provides a system for cell identification, which includes a cell force sensor and an apparatus for cell identification according to the above technical solution.

[0015] Furthermore, the cell force sensor is characterized by including a cell traction force detection device in which a light reflecting layer is provided on a micropillar array or a micropillar.

[0016] Furthermore, a cell traction force detection device in which a light reflective layer is provided on the micropillars, The base and The micropillar array is characterized by including a micropillar array consisting of a plurality of micropillars that are placed on the base and can be deformed under the action of cell traction forces, and in which a light-reflecting layer is provided on the tip or upper part of the cylindrical surface of the micropillars.

[0017] Furthermore, the cell identification system is characterized by further including a cell morphology information acquisition device for acquiring cell morphology information.

[0018] Furthermore, in the cell identification system, the cell morphology information acquisition device is characterized by including a microscope camera or a microscope video camera.

[0019] Furthermore, the cell identification system is characterized by further including a cell control device for performing cell control operations on the cells.

[0020] The inventors also provide a method for detecting a state of a cell, the method comprising: acquiring cell traction force information through any of the cell identification method described in any of the above embodiments, the cell identification device described in any of the above embodiments, or the cell identification system described in any of the above embodiments; and analyzing and determining a state of the cell based on the cell traction force information; The state of the cells includes cell adhesion, cell vitality, cell differentiation / activation, cell proliferation and / or cell migration.

[0021] Furthermore, in any of the above embodiments, the cell may be a single cell or any form of multicellular polymer formed by two or more cells, and the present invention is not limited to a single cell or various forms formed by two or more multicellular cells.

[0022] The above technical solution offers the following advantages over conventional technologies. The present invention uses a cell force sensor to acquire cell mechanical information and use it for cell identification. This cell identification includes not only cell type but also cell state. Furthermore, this technology is non-invasive and non-invasive to living cells, and has the distinct advantages of real-time, high-throughput, and high-resolution capabilities. Based on the measurement of traction forces of individual cells, different cell types can be identified, and this can be used to detect the effects of chemotherapy drugs on cell force and even cell sorting, making it highly valuable for biomedical and medical applications. Furthermore, it can detect cell traction forces, including those of multi-layered cells and tumor aggregates, making it applicable to scenarios requiring the characterization of multicellular polymers in drug screening, regenerative medicine, gene editing, precision medicine, organ development, and pathological modeling. [Brief explanation of the drawings]

[0023] FIG. 1 is a schematic diagram of a process for quantifying displacement information of a specific point in the fourth embodiment of the present invention. FIG. 2 is a diagram A showing the results of using the established cell feature model in the extended embodiment of the fifth embodiment of the present invention to identify unknown cells or unknown cell phenotypes. FIG. 3 is a diagram B showing the results of using the established cell feature model in the extended embodiment of the fifth embodiment of the present invention to identify unknown cells or unknown cell phenotypes. FIG. 4 is a structural schematic diagram of a cell identification device according to the twelfth embodiment of the present invention. FIG. 5 is a structural schematic diagram of a cell identification system according to an eighteenth embodiment of the present invention. FIG. 6 is a structural schematic diagram of a cell traction force detection device according to a twentieth embodiment of the present invention. FIG. 7 is a structural schematic diagram of a cell identification system according to the twentieth embodiment of the present invention. FIG. 8a is a structural schematic diagram of a cell identification system according to the twentieth embodiment of the present invention. FIG. 8b is an image of the optical reflection signal of the cell traction force detection device acquired by the information acquisition unit of the cell identification system according to the twentieth embodiment of the present invention. FIG. 8c is a visualization effect diagram of the magnitude and distribution of dynamics after processing by the pre-processing unit of the cell identification system described in the twentieth embodiment of the present invention. Figure 9a is a diagram of a cell traction force detection device that employs a silicon thin film as a cell control device. Figure 9b is a fluorescence microscope image of the cell traction force detection device employing a silicon thin film as a cell control device under light reflection. FIG. 9c is an enlarged view of FIG. 9b. FIG. 10a is a fluorescent image of a mixture of healthy cells and non-small cell lung cancer cells. FIG. 10b is a distribution diagram of the light reflection signal of the cell traction force detection device acquired by the information acquisition unit. Figure 10c is a visualization effect diagram of the magnitude and distribution of the dynamics after processing by the pre-treatment unit. Figure 10d is a magnified view of the cell force distribution of a representative single cell of a healthy cell and a non-small cell lung cancer cell in Figure 10c. Figure 10e is a comparison of cell morphology between healthy cells and non-small cell lung cancer cells. Figure 10f is a comparison of the reflected signal intensity of healthy cells, non-small cell lung cancer cells, and after mixing these two types of cells in different ratios. FIG. 10g is a clustering analysis diagram obtained by structuring FIG. 10c and then processing it based on the structured cell information. FIG. 11a is a schematic diagram of the operation flow of the cell vitality detection method. FIG. 11b shows a comparison of cell vitality measured by the MTT method and cell vitality reflected by cell traction force after treating A549 cells with different doses of 5FU for 24 hours. FIG. 11c shows a comparison of cell vitality measured by the MTT method and cell vitality reflected by cell traction force after treating A549 cells with different doses of 5FU for different periods of time. FIG. 12a is a diagram showing the operation process of the cell state detection method. FIG. 12b is a fluorescent micrograph of M0 macrophages differentiated to the M1 state. FIG. 12c is a fluorescent micrograph of M0 macrophages differentiated to the M2 state. FIG. 12d is a comparison of the cell adhesion areas of M0 macrophages, M1 and M2 states. FIG. 12e is a comparison of cell circularity in M0 macrophages, M1 and M2 states. Figure 12f is a comparison of traction forces of M0 macrophages, M1 and M2 states. Figure 13a shows the first morphology of tumor cell polymers characterized with and without the action of 5-Fu. From left to right, the image shows a mixed image of cell membrane fluorescence and reflectance signals (1), optical reflectance signals (2), cell nuclei (3), cell membranes (4), and a cell force visualization image (5) after processing with optical image analysis software (Image J). Figure 13b shows the second morphology of tumor cell polymers characterized with and without the action of 5-Fu. From left to right, the image shows a mixed image of cell membrane fluorescence and reflectance signals (1), optical reflectance signals (2), cell nuclei (3), cell membranes (4), and a cell force visualization image (5) after processing with optical image analysis software (Image J). [Explanation of symbols]

[0024] 1-Information Acquisition Unit 2-Pretreatment Unit 3-Study Unit 4-Identification Unit 10-cell force sensor (cell traction force detection device) 101-Basic 102-Optical signal generator 103-Micropillar 104-Beam Splitter 1031-Light reflective layer 20-Cell Identification Device 201-Image / Data Processing Device 30-Cell morphology information acquisition device 40-Cell Control Device DETAILED DESCRIPTION OF THE INVENTION

[0025] In order to describe the technical content, structural features, objectives and effects of the technical solution in more detail, specific examples are given below and will be described in detail with reference to the drawings. First Example

[0026] A method for cell identification (cell traction force magnitude only, no label) comprising the following steps: S1: acquiring cell information, the cell information being the magnitude of cell traction force at a specific point within a cell acquired using a cell mechanical force sensor. Specifically, using the cell mechanical force sensor to collect cell information for multiple cells, including collecting cell traction force magnitude information at multiple points within each cell, thereby obtaining cell traction force magnitude data at multiple points within the multiple cells. S2: Preprocessing the acquired cell traction force magnitude information to form structured cell information, which includes the number of cells, the number of cell features, and feature information of each cell feature. At this point, the structured cell information is It can be regarded as a two-dimensional feature matrix of TIFF0007730493000001.tif36, where N is the number of cells and P is the number of cell features. Here, P=1, meaning that the cell feature is the magnitude of the cell traction force. S3: Using the above structured cell information as input data, establishing a cell feature model using unsupervised learning, and applying the cell feature model to cluster cells of unknown type or unknown state. In some other embodiments similar to this embodiment, optimization or improvement can be achieved in the following manner. For a single cell, the acquired cell traction force magnitude information for multiple points is further processed to calculate, for example, the following: dimensional information such as the average value of the cell traction force magnitude per unit area and the distribution of the cell traction force magnitude within the cell. This is used as a new cell feature to expand the two-dimensional feature matrix (P) described in step S2. Subsequent machine learning can determine which features can better distinguish between different types or states of cells. Second Example

[0027] A method for cell identification (cell traction force magnitude only, labeled), comprising the following steps: S1: acquiring cell information of cells of several known cell types or known cell states, the cell information being the magnitude of cell traction force at a specific point within the cell acquired using a cell mechanical force sensor. Specifically, using a cell mechanical force sensor to collect information on the several types of cells, including collecting cell traction force magnitude information at multiple points within each cell, thereby obtaining cell traction force magnitude data at multiple points within the multiple cells. S2: Preprocessing the acquired cell traction force magnitude information to form structured cell information, which includes the number of cells, the number of cell features, and feature information of each cell feature. At this point, the structured cell information is TIFF0007730493000002.tif36It can be regarded as a two-dimensional matrix of feature matrix, where N is the number of cells and P is the number of cell features. Here, P=1, meaning that the cell feature is the magnitude of the cell traction force. S3: Using the structured cell information as input data, establish a cell feature model using supervised learning, and train the cell feature model using the structured cell information of a large number of cells. For example, in this embodiment, a random forest (RF) algorithm is used to extract salient features and estimate model parameters, which are then applied to new cells to estimate corresponding labels for the new cells. That is, the cell feature model is used to classify cells of unknown type or state (i.e., identification; in this invention, "identification" should be understood in a broad sense, including both "classification," i.e., determining the type or state of a cell, and "clustering," i.e., clustering cells of the same or similar type or state that may have the same or similar properties, even though the specific state or type of the cell is unknown). In other embodiments, machine learning algorithms / concepts such as support vector machines (SVMs) or deep learning can be used to establish and train the corresponding model.

[0028] In some other embodiments similar to this embodiment, optimization or improvement can be achieved in the following manner. For a single cell, the acquired cell traction force magnitude information for multiple points is further processed to calculate, for example, the following: dimensional information such as the average value of the cell traction force magnitude per unit area and the distribution of the cell traction force magnitude within the cell. This is used as a new cell feature to expand the two-dimensional feature matrix (P) described in step S2. Subsequent machine learning can determine which features can better distinguish between different types or states of cells. Third embodiment

[0029] A method for cell identification (only the magnitude of cell traction force, some data labeled, some data unlabeled) includes the following steps: S1: A step of acquiring cell information of a plurality of cells, some of which are cells of known cell types or known cell states, and other cells of unknown cell types or unknown cell states. The cell information is the magnitude of cell traction force at a specific point within the cell, acquired based on a cell mechanical force sensor. This step specifically includes: using a cell mechanical force sensor to collect information for the several types of cells, including collecting cell traction force magnitude information at multiple points within each cell, thereby obtaining cell traction force magnitude data at multiple points within the plurality of cells. S2: Preprocessing the acquired cell traction force magnitude information to form structured cell information, which includes the number of cells, the number of cell features, and feature information of each cell feature. At this point, the structured cell information is TIFF0007730493000003.tif36It can be regarded as a two-dimensional matrix of feature matrix, where N is the number of cells and P is the number of cell features. Here, P=1, meaning that the cell feature is the magnitude of the cell traction force. S3: Using the above structured cell information as input data, establish a cell feature model using semi-supervised learning, train the cell feature model using a large amount of structured cell information of cells (including both labeled and unlabeled cells), and then apply the cell feature model to classify / cluster cells of unknown types or unknown states.

[0030] In some other embodiments similar to this embodiment, optimization or improvement can be achieved in the following manner. For a single cell, the acquired cell traction force magnitude information for multiple points is further processed to calculate, for example, the following: dimensional information such as the average value of the cell traction force magnitude per unit area and the distribution of the cell traction force magnitude within the cell. This is used as a new cell feature to expand the two-dimensional feature matrix (P) described in step S2. Subsequent machine learning can determine which features can better distinguish between different types or states of cells. Fourth embodiment

[0031] A method for cell identification (magnitude and direction of cell traction force, unlabeled), comprising the following steps: S1: Acquiring cell information, where the cell information is the magnitude and direction of cell traction force at a specific point within a cell, acquired using a cell mechanical force sensor. Specifically, using a cell mechanical force sensor (in this embodiment, a micropillar sensor) to collect cell information for multiple cells, including collecting information on the magnitude and direction of cell traction force at multiple points within each cell, thereby acquiring data on the magnitude and direction of cell traction force at multiple points within multiple cells. S2: Preprocessing the acquired information on the magnitude and direction of the cell traction force to form structured cell information, which includes the number of cells, the number of cell features, and the feature information of each cell feature. At this point, the structured cell information is TIFF0007730493000004.tif36It can be regarded as a two-dimensional matrix of feature matrix, where N is the number of cells and P is the number of cell features. In this case, P = 2, that is, the cell features are the magnitude and direction of the cell traction force. S3: Using the above structured cell information as input data, establishing a cell feature model using unsupervised learning, and applying the cell feature model to cluster cells of unknown type or unknown state.

[0032] Specifically, in step S2 of this embodiment, the following process is performed on the cell information. Assume that cell traction force vector data (magnitude and direction) of a total of n points within one cell has been acquired. The positions of these points are calculated as follows: TIFF0007730493000005.tif317), each of which has two 2D coordinates TIFF0007730493000006.tif413 and Corresponding to TIFF0007730493000007.tif412, among which TIFF0007730493000008.tif33 and TIFF0007730493000009.tif33 correspond to the initial position of the micropillar and the position of the point after displacement, respectively. The direction of TIFF0007730493000010.tif426 is the direction of the force at each coordinate point. In addition, the position of each coordinate point contains scalar information, i.e., the magnitude of the force. There should also be a file called TIFF0007730493000011.tif32. In this way, we can estimate the direction of the cell axis and the coordinates of the center point based on the existing data, and use this as a basis to organize each cell into a vector of the same length. For example, based on the position of each point in each cell, the center point is TIFF0007730493000012.tif512. The cell axis is calculated by taking the two furthest points. TIFF0007730493000013.tif38 and TIFF0007730493000014.tif38 Find the cell axis with the following formula: The result is TIFF0007730493000015.tif643.

[0033] Please refer to Figure 1. Figure 1 is a schematic diagram of the process of quantifying displacement information at the position of a specific point in the fourth embodiment of the present invention. Each point in the figure represents the position of one point, and the lighter the color at the position of each point, the smaller the force, and the darker the color, the larger the force. After obtaining the cell axis of each cell, the position of each point can be further quantitatively processed. The angle between the displacement vector of each point and the cell axis Calculate TIFF0007730493000016.tif22. The position of each point can be further combined with the magnitude of the force (a scalar), e.g., TIFF0007730493000017.tif32 is considered as a weight, which determines the position of each point as a scalar TIFF0007730493000018.tif313. In this way, all the cell information for each cell is stored in one vector. It can be organized into TIFF0007730493000019.tif315.

[0034] In some other embodiments similar to this embodiment, optimization or improvement can be achieved in the following manner: For a single cell, the acquired cell traction force magnitude or cell traction force direction information for multiple points is further processed to calculate, for example, the following: dimensional information such as the average magnitude of cell traction force per unit area, the distribution of the cell traction force magnitude within the cell, and the distribution of cell traction force vectors within the cell. This is used as a new cell feature to expand the two-dimensional feature matrix (P) described in step S2, and subsequent machine learning can determine which features can better distinguish between different types or states of cells. Fifth Example

[0035] A method for cell identification (cell traction force magnitude and direction, labeled), comprising the steps of: S1: acquiring cell information of cells of several known cell types or known cell states, the cell information being the magnitude and direction of cell traction forces at specific points within the cells acquired using a cell mechanical force sensor. Specifically, using a cell mechanical force sensor to collect information on the several types of cells, including collecting cell traction force magnitude information at multiple points within each cell, thereby obtaining cell traction force magnitude data at multiple points within the multiple cells. S2: Preprocessing the acquired cell traction force magnitude information to form structured cell information, which includes the number of cells, the number of cell features, and feature information of each cell feature. At this point, the structured cell information is TIFF0007730493000020.tif36It can be regarded as a two-dimensional matrix of feature matrix, where N is the number of cells and P is the number of cell features. Here, P = 2, that is, the cell features are the magnitude and direction of the cell traction force. S3: Using the structured cell information as input data, establish a cell feature model using supervised learning, train the cell feature model using the structured cell information of a large number of cells, and then apply the cell feature model to classify cells of unknown type or unknown state. For example, in this embodiment, a random forest (RF) algorithm is used to extract salient features and estimate model parameters, which are then applied to new cells to estimate labels corresponding to the new cells. That is, the cell feature model is applied to classify cells of unknown type or unknown state. In other embodiments, machine learning algorithms / concepts such as support vector machines (SVMs) or deep learning can be used to establish and train the corresponding model.

[0036] Please refer to Figures 2 and 3. Figures 2 and 3 show the results of using the established cell feature model in the extended embodiment of the fifth embodiment of the present invention to identify unknown cells or unknown cell phenotypes. Different rows in Figure A represent different cell types, and different columns represent different samples. The black dots in the figure can be referred to as the top 50 salient features, or salient point locations, extracted by the random forest algorithm. The point location here refers to a location within a single cell, and the cell traction force information obtained from different locations differs. Meanwhile, Figure B shows the effect of using the top 50 salient features (locations of salient points) to distinguish three different cell types. Based on the salient features learned from the labeled data, the data can be reduced in dimension and visualized. A clustering algorithm can then be used to classify and identify cells based on the reduced dimension data.

[0037] In some other embodiments similar to this embodiment, optimization or improvement can be achieved in the following manner: For a single cell, the acquired cell traction force magnitude or cell traction force direction information for multiple points is further processed to calculate, for example, the following: dimensional information such as the average magnitude of cell traction force per unit area, the distribution of the cell traction force magnitude within the cell, and the distribution of cell traction force vectors within the cell. This is used as a new cell feature to expand the two-dimensional feature matrix (P) described in step S2, and subsequent machine learning can determine which features can better distinguish between different types or states of cells. Sixth embodiment

[0038] A method for cell identification (magnitude and direction of cell traction forces, some data labeled, some data unlabeled) includes the following steps: S1: A step of acquiring cell information of a plurality of cells, some of which are cells of known cell types or known cell states, and other cells of unknown cell types or unknown cell states. The cell information is the magnitude of cell traction force at a specific point within the cells, acquired using a cell mechanical force sensor. Specifically, information is collected for the several types of cells using a cell mechanical force sensor, including collecting information on the magnitude of cell traction force at multiple points within each cell, thereby acquiring data on the magnitude of cell traction force at multiple points within the plurality of cells. S2: Preprocessing the acquired cell traction force magnitude information to form structured cell information, which includes the number of cells, the number of cell features, and feature information of each cell feature. At this point, the structured cell information is TIFF0007730493000021.tif36It can be regarded as a two-dimensional matrix of feature matrix, where N is the number of cells and P is the number of cell features. Here, P = 2, that is, the cell features are the magnitude and direction of the cell traction force. S3: Using the above structured cell information as input data, establish a cell feature model using semi-supervised learning, train the cell feature model using the structured cell information of a large number of cells, and then apply the cell feature model to classify cells of an unknown type or state.

[0039] In some other embodiments similar to this embodiment, optimization or improvement can be achieved in the following manner: For a single cell, the acquired cell traction force magnitude or cell traction force direction information for multiple points is further processed to calculate, for example, the following: dimensional information such as the average magnitude of cell traction force per unit area, the distribution of the cell traction force magnitude within the cell, and the distribution of cell traction force vectors within the cell. This is used as a new cell feature to expand the two-dimensional feature matrix (P) described in step S2, and subsequent machine learning can determine which features can better distinguish between different types or states of cells. Seventh embodiment

[0040] A method for cell identification (instantaneous value of cell traction force vector, change status of cell traction force vector within a predetermined time interval, unlabeled), comprising the following steps: S1: Acquiring cell information, where the cell information is the instantaneous value of the cell traction force vector at a specific point within the cell acquired using a cell mechanical force sensor and the change in the cell traction force vector at that point within a predetermined time interval. Specifically, using the cell mechanical force sensor to collect cell information for multiple cells, including collecting information on the magnitude and direction of the cell traction force at multiple points within each cell, thereby obtaining data on the magnitude and direction of the cell traction force at multiple points within the multiple cells. S2: Preprocessing the acquired information on the magnitude and direction of the cell traction force to form structured cell information, which includes the number of cells, the number of cell features, and the feature information of each cell feature. At this point, the structured cell information is TIFF0007730493000022.tif36It can be considered as a two-dimensional matrix of feature matrix, where N is the number of cells and P is the number of cell features. S3: Using the above structured cell information as input data, establishing a cell feature model using unsupervised learning, and applying the cell feature model to cluster cells of unknown type or unknown state.

[0041] In this example, the acquired single-cell mechanical data is not only the instantaneous value of the cell traction force vector at a specific point within the cell, but also its interval status within a specified time interval. Therefore, the acquired single-cell mechanical data is actually similar to an image (instantaneous value) or video (changes in the time dimension). In this way, if the array of points under the current cell can be likened to pixels in an image, and the information recorded at each point (multiple cellular features such as force magnitude and direction) can be likened to the color corresponding to the pixel, machine learning algorithms from the field of image and video data processing can be used in subsequent machine learning. For example, machine learning, which is widely applied to image recognition, or more specifically, deep learning convolutional neural networks (CNNs), can be used to model and analyze the data. Eighth embodiment

[0042] A method for cell identification (instantaneous value of cell traction force vector, change status of cell traction force vector within a predetermined time interval, labeled), comprising the following steps: S1: A step of acquiring cell information of cells of several known cell types or known cell states, the cell information being the instantaneous value of a cell traction force vector at a specific point in the cell acquired using a cell mechanical force sensor and the change in the cell traction force vector at that point within a predetermined time interval. Specifically, the cell mechanical force sensor is used to collect cell information for multiple cells, including collecting information on the magnitude and direction of the cell traction force at multiple points in each cell, thereby acquiring data on the magnitude and direction of the cell traction force at multiple points in the multiple cells. S2: Preprocessing the acquired information on the magnitude and direction of the cell traction force to form structured cell information, which includes the number of cells, the number of cell features, and the feature information of each cell feature. At this point, the structured cell information is TIFF0007730493000023.tif36It can be considered as a two-dimensional matrix of feature matrix, where N is the number of cells and P is the number of cell features. S3: Using the structured cell information as input data, establish a cell feature model using supervised learning, train the cell feature model using the structured cell information of a large number of cells, and then apply the cell feature model to classify cells of unknown type or state (i.e., identification; in this invention, "identification" should be understood in a broad sense and includes both "classification," i.e., determining the type or state of a cell, and "clustering," i.e., clustering cells of the same or similar type or state, where the specific state or type of the cell is unknown but the cells may have the same or similar properties). For example, in this embodiment, a random forest (RF) algorithm is used to extract salient features and estimate model parameters, which are then applied to new cells to estimate labels corresponding to the new cells. In other embodiments, machine learning algorithms / concepts such as support vector machines (SVMs) or deep learning can be used to establish and train the corresponding model.

[0043] In this example, the acquired single-cell mechanical data is not only the instantaneous value of the cell traction force vector at a specific point within the cell, but also its interval status within a specified time interval. Therefore, the acquired single-cell mechanical data is actually similar to an image (instantaneous value) or video (multiple instantaneous values ​​within a specified time dimension). In this way, if the array of points under the current cell can be likened to pixels in an image, and the information recorded at each point (multiple cellular features such as force magnitude and direction) can be likened to the color corresponding to the pixel, machine learning algorithms from the field of image and video data processing can be used in subsequent machine learning. For example, machine learning, which is widely applied to image recognition, or more specifically, deep learning convolutional neural networks (CNNs), can be employed to model and analyze the data. Ninth Example

[0044] A method for cell identification (instantaneous values ​​of cell traction force vectors, changes in cell traction force vectors within a predetermined time interval, some data labeled, some data unlabeled) includes the following steps: S1: A step of acquiring cell information of cells of several known cell types or known cell states, the cell information being the instantaneous values ​​of cell traction force vectors at specific points within the cells acquired using a cell mechanical force sensor, and the changes in the cell traction force vectors at those points within a predetermined time interval. Specifically, the cell mechanical force sensor is used to collect cell information for multiple cells, including information on the magnitude and direction of cell traction forces at multiple points within each cell, and the information is collected continuously within a predetermined time range, thereby obtaining data on the magnitude and direction of cell traction forces at multiple points within the multiple cells, as well as information on the changes in the cell traction force vectors at those points within the predetermined time interval. S2: Preprocessing the acquired information on the magnitude and direction of the cell traction force to form structured cell information, which includes the number of cells, the number of cell features, and the feature information of each cell feature. At this point, the structured cell information is TIFF0007730493000024.tif36It can be considered as a two-dimensional matrix of feature matrix, where N is the number of cells and P is the number of cell features. S3: Using the above structured cell information as input data, establish a cell feature model using semi-supervised learning, train the cell feature model using the structured cell information of a large number of cells, and then apply the cell feature model to classify / cluster cells of unknown types or unknown states.

[0045] In this example, the acquired single-cell mechanical data is not only the instantaneous value of the cell traction force vector at a specific point within the cell, but also its interval status within a specified time interval. Therefore, the acquired single-cell mechanical data is actually similar to an image (instantaneous value) or video (multiple instantaneous values ​​within a specified time dimension). In this way, if the array of points under the current cell can be likened to pixels in an image, and the information recorded at each point (multiple cellular features such as force magnitude and direction) can be likened to the color corresponding to the pixel, machine learning algorithms from the field of image and video data processing can be used in subsequent machine learning. For example, machine learning, which is widely applied to image recognition, or more specifically, deep learning convolutional neural networks (CNNs), can be employed to model and analyze the data. Tenth embodiment

[0046] This cell identification method differs from the first to ninth embodiments in that the cell information also includes cell morphology information, which is acquired by a microscope camera or a microscope video camera (when it is necessary to continuously collect information within a predetermined time interval). In this case, the cell features in the data pre-processing step include one or more of the cell morphology information or further information obtained by further processing or analyzing the cell morphology information, namely, cell size, cell shape, cell nucleus size, cell nucleus shape, and cell color. Eleventh embodiment

[0047] This cell identification method differs from the first to tenth embodiments in that the cell information is collected under the premise of restricting the cells. Methods for restricting cell morphology include physical methods. For example, a restricting wall that encloses a certain area can be set around the force sensors of some cells. Because the height of such a restricting wall is higher than the height of the force sensors of the cells inside it, it can be considered a kind of cell control device. By setting a certain spatial area, only cells of a corresponding size can fall into it and contact the force sensors of the cells, and it can also be thought of as only being able to accommodate a corresponding number of cells (in most cases, the restricting wall can only accommodate one cell falling into it). In special circumstances, the restricting wall can further compress the cells that enter it to some extent, conforming to the cross-sectional shape surrounded by the restricting wall, thereby achieving a certain cell morphology restriction effect.

[0048] In the cell identification method of the present invention, compared to when cells are not restricted, restricting cells (by restricting the number or shape) first reduces contact between cells, ensuring that most cells remain in a single-cell state. Second, the dimensions of cell size and shape can be reduced, i.e., dimension reduction can be performed on the number of cell features, thereby lowering the difficulty of the analysis process to some extent. Furthermore, specific cell shapes can be used to control the differentiation state of cells, for example, morphologically restricting immune cells to the M1 state, which can produce more valuable results when used in specific scenarios.

[0049] Of course, if no shape constraints are imposed on the cells, the cells can be left in their natural state, and the impact of external interference on the cells can be reduced. Furthermore, only when no shape constraints are imposed can edge detection be performed on the cells, thereby obtaining characteristics such as cell shape and orientation, which can assist in cell identification and prediction. In short, the cell constraint solution of this embodiment has unique advantages for operation in specific technical scenarios. Twelfth embodiment

[0050] See Figure 4. This embodiment provides a cell identification device, which includes an information acquisition unit 1, a pre-processing unit 2, a learning unit 3 and an identification unit 4. The information acquisition unit 1 is used to acquire cell information, and the cell information includes cell traction force information at a specific point within the cell acquired based on a cell force sensor, and the cell traction force information includes the magnitude of the cell traction force at that point. The pre-processing unit 2 is used to pre-process the cell information and form structured cell information, which includes the number of cells, the number of cell features, and feature information of each cell feature. The learning unit 3 is used to use the structured cell information as input data to establish a cell feature model using supervised learning, unsupervised learning, or semi-supervised learning. The identification unit 4 is used to apply the cell feature model to classify or cluster cells of unknown type or state.

[0051] The cell identification device of this embodiment can be used to realize the technical solutions of the cell identification methods described in the first to third embodiments. Thirteenth embodiment

[0052] The cell identification device of this embodiment differs from the twelfth embodiment in that the cell traction force information acquired by the information acquisition unit 1 also includes the direction of the cell traction force at that point. The cell identification device of this embodiment can be used to realize the technical solutions of the cell identification methods described in the fourth to sixth embodiments. Fourteenth embodiment

[0053] The cell identification device of this embodiment differs from the twelfth and thirteenth embodiments in that the cell traction force information acquired by the information acquisition unit 1 also includes changes in the magnitude or direction of the cell traction force at that point within a predetermined time interval. The cell identification device of this embodiment can be used to realize the technical solutions of the cell identification methods described in the seventh to ninth embodiments. Fifteenth Example

[0054] The cell identification device of this embodiment differs from the twelfth to fourteenth embodiments in that the cell traction force information acquired by the information acquisition unit 1 also includes cell morphology information. The cell identification device of this embodiment can be used to realize the technical solution of the cell identification method described in the tenth embodiment. Sixteenth embodiment

[0055] This cell identification device differs from the twelfth to fifteenth embodiments in that the cell information is acquired after performing a cell control operation on the cells. Methods for restricting cell morphology include physical methods. For example, a restricting wall that encloses a certain area can be set around the force sensors of some cells. Because the height of such a restricting wall is higher than the height of the force sensors of the cells inside it, it can be considered a kind of cell control device. By setting a certain spatial area, only cells of a corresponding size can fall into it and contact the force sensors of the cells, and it can also be considered that only a corresponding number of cells (in most cases, the restricting wall can only accommodate one cell) can be accommodated within it. In special circumstances, the restricting wall can further compress the cells that enter it to some extent to conform to the cross-sectional shape surrounded by the restricting wall, thereby achieving a certain cell morphology restriction effect.

[0056] In the cell identification method of the present invention, compared to when cells are not restricted, restricting cells (by restricting the number or shape) first reduces contact between cells, ensuring that most cells remain in a single-cell state. Second, the dimensions of cell size and shape can be reduced, i.e., dimension reduction can be performed on the number of cell features, thereby lowering the difficulty of the analysis process to some extent. Furthermore, specific cell shapes can be used to control the differentiation state of cells, for example, morphologically restricting immune cells to the M1 state, which can produce more valuable results when used in specific scenarios. Of course, if no morphological constraints are imposed on the cells, the cells will be in their natural state, reducing the impact of external interference on the cells. Furthermore, only when no morphological constraints are imposed can cell edge detection be performed, thereby obtaining characteristics such as cell morphology and tropism, which can aid in cell identification and prediction. In short, the cell constraint method described in this embodiment has unique advantages in certain technical scenarios.

[0057] In the cell identification method of the present invention, compared to when cells are not restricted, restricting cells (by restricting the number or shape) first reduces contact between cells, ensuring that most cells remain in a single-cell state. Second, the dimensions of cell size and shape can be reduced, i.e., dimension reduction can be performed on the number of cell features, thereby lowering the difficulty of the analysis process to some extent. Furthermore, specific cell shapes can be used to control the differentiation state of cells, for example, morphologically restricting immune cells to the M1 state, which can produce more valuable results when used in specific scenarios.

[0058] Of course, if cells are not restricted, the cells can be left in their natural state and the impact of external interference on the cells can be reduced. Furthermore, only when no morphological restrictions are imposed can edge detection be performed on the cells, thereby obtaining characteristics such as cell morphology and orientation, which can assist in cell identification and prediction. In short, the cell restriction method of this embodiment has unique advantages for use in specific technical scenarios. The cell identification device of this embodiment can be used to realize the technical solution of the cell identification method described in the eleventh embodiment. Seventeenth Example

[0059] A cell identification system includes a cell force sensor 10 and a cell identification device 20. The cell identification device is the cell identification device described in the twelfth to sixteenth embodiments above, and is used to realize the technical solutions of the cell identification methods described in the first to ninth embodiments above.

[0060] The cell force sensor 10 in this embodiment is a micropillar array, but in other embodiments, any device capable of acquiring cell traction force information can also be used as the cell force sensor. Eighteenth Example

[0061] See Figure 5. This embodiment provides a cell identification system, which differs from the seventeenth embodiment in that it further includes a cell morphology information acquisition device 30 for acquiring cell morphology information. The cell morphology information acquisition device 30 in this embodiment is a microscope camera; in other embodiments, the cell morphology information acquisition device 30 may be a microscope video camera or other device capable of acquiring cell morphology information. The cell identification system of this embodiment can be used to realize the technical solution of the cell identification method described in the tenth embodiment. Nineteenth Example

[0062] This cell identification system differs from the seventeenth and eighteenth embodiments in that it further includes a cell control device 40 for performing cell control operations on cells. The cell control device 40 in this embodiment specifically has the following structure: A limiting wall is provided around the force sensors of several cells, enclosing a certain area. Because the height of this limiting wall is higher than the height of the force sensors of the cells inside it, it can be considered a type of cell control device. Because a certain spatial area is set, only cells of a corresponding size can fall into it and contact the force sensors of the cells, and it can also be considered that only a corresponding number of cells (in most cases, the limiting wall can only accommodate one cell) can be accommodated within it. In special circumstances, the limiting wall can further compress the cells that enter it to a certain extent and adapt to the cross-sectional shape surrounded by the limiting wall, thereby achieving a certain cell shape restriction effect. The cell identification system in this embodiment can be used to realize the technical solution of the cell identification method described in the eleventh embodiment above. Twentieth Example

[0063] This cell identification system differs from the seventeenth, eighteenth, and nineteenth embodiments in that the cell force sensor 10 of this embodiment is a cell traction force detection device with a light-reflecting layer provided on a micropillar. In other embodiments, any device capable of acquiring cell traction force information can be used as the cell force sensor.

[0064] See Figure 6 for details. This is a schematic diagram of the structure of a cell traction force detection device. As shown in Figure 6, the cell traction force detection device of this embodiment includes a light-transmitting base 101 and micropillars 103 mounted on the base 101 and capable of deforming under the action of cell traction force. A light-reflecting layer 1031 is coated on the tip of the micropillar 103, with a thickness of 5 nm. (In some other embodiments, the thickness of the light-reflecting layer 1031 can be between 5 nm and 20 nm. The thickness of the coating layer is related to the coating material. Assuming the same coating material is used, the thickness of the coating layer should be limited to ensure light transmittance, ensure the stability of the micropillars, and prevent peeling of the connection with the micropillars.) The micropillars 103 are capable of transmitting light, and the bundles of arrows in opposite directions in the figure represent the incident and reflected light. (Note: Although the word "coating" is used in this embodiment, it merely indicates that the light reflecting layer 1031 of this embodiment can be manufactured by a coating process, and does not necessarily mean that the light reflecting layer 1031 must be manufactured by a coating process.)

[0065] When the cell traction force detection device 1 of this embodiment is put into use, the number of micropillars 103 is not limited to one. See FIG. 7, which is a structural schematic diagram of a cell identification system related to a twenty-first embodiment of the present invention. The system shown in FIG. 7 includes not only the cell traction force detection device 1 and cell identification device (including an information acquisition unit, a preprocessing unit, a learning unit, and an identification unit) described in this embodiment, but also the following: an optical signal generator 102 having a light source, installed below the base 101. Light emitted from the light source travels through the incident optical path and is irradiated from the translucent base 101 of the cell traction force detection device 10 to the light reflecting layer of the micropillar 103. The information acquisition unit 1 detects light reflected from the light reflecting layer 1031 at the tip of the micropillar 103. The light reflected from the light reflecting layer 1031 travels through the reflected optical path and enters the cell identification device 20 via the beam splitter 104. After the information acquisition unit 1 (optical signal detection device) of the cell identification device 20 acquires the reflected light signal, the information is preprocessed by the preprocessing unit 2 to form structured cell information. The structured cell information includes the number of cells, the number of cell features, and feature information for each cell feature. At this point, the structured cell information can be viewed as a two-dimensional feature matrix, where N is the number of cells and P is the number of cell features, where P = 1 or 2. That is, the cell features are the magnitude of cell traction force and / or the distribution of cell traction force within the cell. Next, using the structured cell information as input data, a cell feature model is established using the learning unit 3, which employs supervised, unsupervised, or semi-supervised learning, and the structured cell information of a large number of cells is used to train the cell feature model. Finally, the cell feature model is applied to classify or cluster cells of unknown type or state through the identification unit 4. When the micropillars 12 are not subjected to force, they should maintain an upright state, thereby maximizing the reflection of the detection light. On the other hand, when the micropillars 103 come into contact with cells, under the action of cell traction force, the micropillars 103 bend and the light reflection level decreases. Therefore, the larger the cell traction force, the smaller the obtained light reflection signal should be.In this way, the magnitude of the cell traction force at that point can be easily calculated by observing the intensity of the light reflection signal.

[0066] In some embodiments of the present invention, beam splitter 104 may be a semi-transmissive, semi-reflective or other equivalent optical element, the main purpose of which is to simplify the design of the optical path.

[0067] In some embodiments of the present invention, the optical signal generating device 102 may be an LED, a halogen lamp, a laser (e.g., an infrared laser), or other light source, or other device having such a light source, and the present invention is not specifically limited thereto.

[0068] In some embodiments of the present invention, the information acquisition unit 1 of the cell identification device can be a microscope, a charge-coupled device (CCD), a complementary metal-oxide semiconductor (CMOS), a photomultiplier tube (PMT) and a photoelectric converter (PT), a film, or other optical signal detection elements having the same function, and the present invention is not specifically limited thereto.

[0069] In some embodiments of the present invention, the pre-processing unit, learning unit, and identification unit of the cell identification device are integrated into an image / data processing device 201. For example, it may be optical image analysis software such as ImageJ, Matlab, Fluoview, Python, or other optical image / data analysis elements having the same functions, or a combination of these analysis software, and the present invention is not specifically limited thereto.

[0070] The traction force detection and cell identification analysis processes of the cell identification system related to this embodiment will be specifically introduced below. Please refer to Figures 8a to 8c. Figure 8a is a structural schematic diagram of the cell identification system, Figure 8b is an image of the optical reflection signal of the cell traction force detection device acquired by the information acquisition unit, and Figure 8c is a visualization diagram of the force magnitude and distribution. As shown in Figure 8a, on the cell traction force detection device, each micropillar has a metal reflective layer on its tip and an anti-reflection layer on its side. When there are no cells, the light beam illuminates the micropillar from below, is completely reflected, and is completely received by the information acquisition unit (e.g., a CCD camera) of the cell identification device. However, when a cell adheres to the micropillar, the cellular force generated by the cell's movement tilts the micropillar, thereby reducing the reflected signal. The strength of the cellular force can be calculated by analyzing the optical reflected signal.

[0071] Furthermore, an information acquisition unit (e.g., a CCD camera) collects an image of the optical reflection signal of the cell traction force detection device and a magnified image of the local cell adhesion area (shown in Figure 8b).Then, a preprocessing unit further processes the image in Figure 8b and converts it into a more intuitive visualization of the force magnitude and distribution (Figure 8c).

[0072] The specific processing process is as follows: First, based on Figure 8b, a bright-field reflected signal image (I, focused on the cell) is obtained. Next, the image is Fourier transformed, followed by filtering the high-frequency signal and performing an inverse Fourier transform to calculate and obtain the micropillar reflected signal image (I0) under unshifted conditions. Next, the I and I0 images are further processed to convert the reflected signal image into a more intuitive cell dynamics diagram (by subtracting the I signal value from the I0 signal value), and then normalized to obtain a more intuitive cell traction force intensity diagram j. Twenty-first embodiment

[0073] This embodiment differs from the twentieth embodiment in that the cell control device 40 of this embodiment is a silicon thin film.

[0074] Specifically, see Figures 9a to 9c. Figure 9a is an actual diagram of a cell traction force detection device employing a silicon membrane as a cell control device. In Figure 9a, the silicon membrane is laser-punched and then attached to a base, with micropillars in each hole. The silicon membrane restricts cell morphology and migration while simultaneously controlling contact or adhesion between cells. Figure 9b is a fluorescent microscopic image of a cell traction force detection device employing a silicon membrane as a cell restriction mechanism under light reflection, and Figure 9c is an enlarged view of Figure 9b. In some embodiments, the size of each hole in the silicon membrane can be tailored to fit the size of a single cell, making it suitable for single-cell adhesion and thereby restricting cell contact, cell morphology, and the range of cell migration. Twenty-second embodiment

[0075] This embodiment specifically provides the application of cell traction force information obtained by the cell identification system described in the twentieth embodiment to a cell identification method.

[0076] See Figures 10a to 10g. Figure 10a shows a fluorescent image of a mixture of healthy cells and non-small cell lung cancer cells. Figure 10b shows the distribution of light reflection signals from the cell traction force detection device acquired by the information acquisition unit. Figure 10c shows a visualization of the magnitude and distribution of force. Figure 10d shows a close-up of the cell force distribution of a representative single cell of healthy cells and non-small cell lung cancer cells in Figure 10c. Figure 10e shows a comparison of the cell morphology of healthy cells and non-small cell lung cancer cells. Figure 10f shows a comparison of the reflection signal intensity of healthy cells, non-small cell lung cancer cells, and cells mixed at different ratios. Figure 10g shows a clustering analysis obtained by structuring Figure 10c and then processing it based on the structured cell information.

[0077] Specifically, in this example, healthy cells (Normal) and a non-small cell lung cancer cell line (Cancer) were used as detection targets, and the cell membranes of the healthy cells and lung cancer cells were pre-stained using two different fluorescent dyes (Dil & DIO), mixed at a certain ratio, and then added to the same cell traction force detection device (although in some other examples, they may be added to different independent cell traction force detection devices).

[0078] Furthermore, an image of the light reflection signal from the cell traction force detection device was collected by an information acquisition unit (a microscope was used in this example) (shown in Figure 10b). The high-resolution force field distributions within the two types of cells were directly rendered by the information acquisition unit and converted into readable light intensity attenuation signals (reflecting the cell force intensity) and displayed in the image (shown in Figure 10c). Based on the difference in the degree of light attenuation between the two types of cells shown in the image in Figure 10c, the two types of cells can be intuitively distinguished by observing them with the naked eye (qualitative analysis).

[0079] The optical reflection signal in Figure 10c is further processed by an optical signal analyzer. Specifically, in this embodiment, ImageJ and Python analysis software (although other image / data analysis software can be used in other embodiments) are used to collect information on the obtained cell force field in Figure 10c, including collecting the magnitude information of the cell traction forces at multiple points on each cell, thereby obtaining the magnitude data of the cell traction forces at multiple points within multiple cells. The acquired cell traction force magnitude information is preprocessed to form structured cell information, and analysis is performed based on the structured cell information to obtain a comparison diagram of the cell morphology of healthy cells and non-small cell lung cancer cells (shown in Figure 10e).

[0080] The structured cell information includes the number of cells, the number of cell features, and feature information for each cell feature (e.g., the cell adhesion area and cell circularity in this example). At this point, the structured cell information can be viewed as a two-dimensional feature matrix, where N is the number of cells and P is the number of cell features, where P=2 in this example. That is, the cell features are the magnitude of cell traction force and the distribution of cell traction force within the cell.

[0081] Furthermore, using the above structured cell information as input data, a cell feature model is established using supervised learning in the learning unit of the image / data processing device (compared to pre-staining two types of cell lines with different cell membrane dyes (Dil and DIO)). The structured cell information of a large number of cells is used to train the cell feature model, resulting in the clustering analysis diagram shown in Figure 10g. The resulting cell feature model is then applied to classify and identify cells of unknown type or state. Thus, by using the structured cell feature data (magnitude of cell traction force, distribution of cell traction force within the cell) as input data, the classification unit of the image / data processing device can be used to cluster and classify normal healthy cells and cancer cells, thereby enabling the identification of unknown cell types.

[0082] Figure 10e shows that there are no statistically significant differences in the morphology of different cells (including cell adhesion area and cell circularity). Figure 10f shows that there are clear differences in the reflected signal intensity (reflecting cell force) of normal and tumor cells, and that there is a linear relationship between the reflected signal intensity and the mixing ratio after normal and tumor cells are mixed at a certain ratio. This demonstrates that the cell mechanical characteristics measured by the cell traction force detection device of the present invention can be used to more intuitively and accurately identify the state and type of cells (quantitative and qualitative analysis) compared with other cell characteristics (e.g., morphological information such as cell adhesion area and cell circularity in Figure 10e).

[0083] Furthermore, the data in Figures 10d and 10f show that tumor cells exhibit higher traction force magnitudes and more heterogeneous distributions than normal cells. After visualizing cell traction forces, it is possible to intuitively see clear differences in the force field characteristics of different cells. Furthermore, by using image analysis software to structure and process the force field magnitudes at each point of different cells and then comprehensively analyzing them, we obtain the cell morphology information in Figure 10e, the reflected signal intensity (reflecting cell force) in Figure 10f, and the clustering analysis diagram in Figure 10g. Through comprehensive analysis of the structured force field information at each point of cells, the present invention can perform clustering classification and quantitative analysis of different cells (e.g., healthy cells and non-small cell lung cancer cells in this example), thereby enabling accurate identification of cell types.

[0084] Overall, based on the cell identification system including the cell traction force detection device of this example, it was demonstrated that not only can qualitative analysis be performed through intuitive distinction by the naked eye, but also that the state and type of cells can be more intuitively and accurately identified based on the measured cell mechanical characteristics (quantitative and qualitative analysis), and that cell types can be better distinguished by using the cell force field as a marker. Twenty-third embodiment

[0085] This example specifically provides the application of cell traction force information obtained by the cell identification system described in Example 20 to monitoring cell vitality.

[0086] Please refer to Figures 11a to 11c. Figure 11a is a schematic diagram of the operation flow of the cell vitality detection method, Figure 11b is a comparison graph of the cell vitality measured by the MTT method and the cell traction force measured by the cell identification system of this example after treating A549 cells with different doses of 5FU for 24 hours, and Figure 11c is a comparison graph of the cell vitality measured by the MTT method and the cell traction force measured by the cell identification system of this example after treating A549 cells with different doses of 5FU for different periods of time.

[0087] Specifically, in this example, non-small cell lung cancer cells A549 were cultured on multiple cell traction force detection devices and treated with different doses of the cytostatic drug 5-fluorouracil (5-FU). Cell traction force was monitored at different time points using the cell identification system described in Example 20, and cell proliferation and cytotoxicity were monitored at different time points using a CCK-8 reagent kit. At the same time, cell vitality measured by the MTT assay served as the control group, and the data in Figures 11b and 11c were obtained.

[0088] As shown in Figures 11b and 11c, after measurement using the conventional MTT assay and the cell identification system of this embodiment, the cell vitality measured by the MTT assay and the cell vitality reflected by the cell traction force both showed a gradual decrease in a dose-dependent manner, i.e., there is a positive correlation between the cell traction force and the cell vitality.

[0089] Furthermore, as shown in Figure 11b, after 24 hours of treatment with different doses of 5FU, the traction force showed a greater decrease in cell vitality compared to the DMSO control, allowing for a more intuitive assessment of cell vitality. As shown in Figure 11c, after 12 hours of treatment with 5FU, no significant changes in cell vitality were observed as measured by the MTT method. However, by measuring traction force, we were able to observe a decrease in cell traction force at an earlier time point, before the decrease in cell metabolic activity was detected by the MTT method. Specifically, a clear decrease was observed at 6 hours with a 0.5 μM treatment dose, and at 3 hours with a 1 μM treatment dose, providing a more sensitive indication of the decrease in cell vitality.

[0090] Overall, direct detection of cell traction force through a cell discrimination system including the cell traction force detection device of this embodiment is a highly sensitive and effective method for evaluating the drug response activity of cells. Twenty-fourth embodiment

[0091] This embodiment specifically provides cell traction force information obtained by the cell identification system described in the twentieth embodiment, and analyzes and determines the state of the cell based on the cell traction force information.

[0092] See Figures 12a to 12f. Figure 12a is a diagram illustrating the operation process of the cell state detection method, Figure 12b is a fluorescent micrograph of M0 macrophages differentiated into the M1 state, Figure 12c is a fluorescent micrograph of M0 macrophages differentiated into the M2 state, Figure 12d is a comparison of the cell adhesion area between M0 macrophages, the M1 state, and the M2 state, Figure 12e is a comparison of the cell circularity between M0 macrophages, the M1 state, and the M2 state, and Figure 12f is a comparison of the traction force between M0 macrophages, the M1 state, and the M2 state.

[0093] Specifically, in this example, macrophages were used as the detection target and placed on the micropillars of different independent cell traction force detection devices. The macrophages were differentiated from M0 to M1 and M2 states using lipopolysaccharide (LPS) and interleukin (IL4), respectively, with the M0 state serving as the control. After cell differentiation, images (Figures 12b and 12c) were collected using an information acquisition unit (a microscope in this example). Images (Figures 12b and 12c) were further processed and analyzed using an image / data processing system (ImageJ and Python), which then converted the data into structured information and analyzed it to produce the data shown in Figures 12d through 12f. From the data shown in Figures 12a through 12f, clear differences can be seen between M0 macrophages and the differentiated M1 and M2 states, whether through intuitive observation (Figures 12b and 12c) or quantitative processing of the structured data (Figures 12d through 12f). 25th Example

[0094] This embodiment specifically provides cell traction force information obtained by the cell identification system described in the twentieth embodiment, and analyzes and determines the state of the cell based on the cell traction force information.

[0095] Specifically, this embodiment provides that the multicellular polymer can be attached to the cell traction force measurement device in multiple ways, and this embodiment provides two specific attachment methods. First, a culture medium is placed on the micropillars of the cell traction force detection device, and cells are transplanted into the culture medium on the micropillars and cultured to obtain a multicellular polymer. In some embodiments, this method allows real-time monitoring of the cell culture process while visualizing cell traction force information. This method can be applied to the effects of chemical, biological, and physical external stimuli such as culture medium and drugs on cell growth. Second, the cultured multicellular polymer is directly attached to the micropillars of the cell traction force detection device for detection.

[0096] More specifically, this example provides that tumor cell polymers cultured on a cell traction force detection device are applied to drug sensitivity testing, which includes the following steps: S1. Tumor cell polymer generation: The tips of the micropillars of the cell traction force detection device were coated with 50 μg / mL of FN and sterilized with UV light for 30 minutes. MCF-7 breast cancer cells (approximately 1 × 105 to 9 × 105 cells) were seeded onto the tips of the micropillars of the cell traction force detection device (the number of micropillars is not limited). The cell traction force monitoring device was then immersed in 3dGRO™ Spheroid Medium (S3077) culture medium and cultured for at least 3 days to induce tumor cell polymer generation. Step S2: The tumor cell polymers generated above were used in a 5-Fu drug sensitivity experiment. 200 μM 5-Fu was added to the tumor cell polymers and then cultured for one day. For the experiment, the cell traction force detection device (in which the tumor cell polymers were generated and may be combined with the culture medium) was used to culture the cells for one day before and after the addition of 5-Fu. The optical reflection signal was acquired through the information acquisition unit (CCD photosensitive element) of the cell identification device, and an optical image analysis software (Image J) was used to obtain a cell force distribution image (shown in Figure 13).

[0097] See Figures 13a and 13b. Figure 13a shows the first morphology of tumor cell polymers characterized with and without the action of 5-Fu, and Figure 13b shows the second morphology of tumor cell polymers characterized with and without the action of 5-Fu. From left to right, they show a mixed image of cell membrane fluorescence and reflectance signals (1), optical reflectance signals (2), cell nuclei (3), cell membranes (4), and a cell force visualization image (5) after processing with optical image analysis software (Image J). Because cells are heterogeneous, tumor cells are different, and polymers also have various morphologies. Therefore, cell polymers adhere together in different morphologies. In this example, two representative morphologies were selected for cell morphology detection. The first morphology refers to the morphology of two relatively large cells adhered together, and the second morphology refers to the morphology of a group of small cells adhered together.

[0098] Figures 13a and 13b show that the reflected signal is significantly weaker after treatment with an antitumor drug to reduce cell vitality. Furthermore, the changes in cell mechanical force before and after 5-Fu treatment reveal significant differences in the drug sensitivity of the two different tumor cell polymer forms. This demonstrates that the cell traction force detection device of the present invention can measure cell traction force on multicellular polymers (such as tumor polymers), monitor the vitality of cell polymers through cell traction force, and distinguish between different cell morphologies.

[0099] In the present definition, cellular polymers refer to: Cells are the basic structural and functional units of living organisms, and cells typically proliferate or differentiate to form groups of two or more cells together to form cell colonies, or multicellular polymers. Multicellular polymers include cell groups such as tumor polymers obtained by culturing in vitro or in vivo.

[0100] It should be noted that although the above embodiments have been described herein, they do not limit the scope of patent protection of the present invention. Therefore, based on the creative idea of ​​the present invention, any changes and modifications to the embodiments described herein, or the conversion of equivalent structures or equivalent processes made by utilizing the contents of the specification and drawings of the present invention, or the application of the above technical solutions directly or indirectly to other related technical fields, are all within the scope of patent protection of the present invention.

Claims

1. 1. A method for cell identification, comprising: acquiring cell information, the cell information including cell traction force information at a specific point within the cell acquired based on a cell force sensor, the cell traction force information including the magnitude of the cell traction force at the point; a step of preprocessing the cell information to form structured cell information, the structured cell information including the number of cells, the number of cell features, and feature information of each cell feature; using the structured cell information as input data to establish a cell feature model using supervised, unsupervised, or semi-supervised learning, and applying the cell feature model to classify or cluster cells of unknown type or state; A method for cell identification, characterized in that in the step of acquiring cell information, the cell force sensor is a cell traction force detection device having a light reflecting layer provided on a micropillar, and the cell information is acquired based on the intensity of the light reflection signal from the light reflecting layer.

2. The method of claim 1 , wherein the cell traction force information also includes the direction of the cell traction force at the point.

3. The method of claim 1 , wherein the cell traction force information also includes changes in the magnitude or direction of the cell traction force at the point within a predetermined time interval.

4. The method of claim 1 , wherein the cell information also includes cell morphology information.

5. 2. The method for cell identification according to claim 1, wherein the cell information is obtained after performing a cell control operation on the cell.

6. An apparatus for cell identification, comprising: an information acquisition unit, a pre-processing unit, a learning unit, and an identification unit; the information acquisition unit is used to acquire cell information, the cell information including cell traction force information at a specific point within the cell acquired based on a cell force sensor, the cell traction force information including a magnitude of the cell traction force at the point; the pre-processing unit is used to pre-process the cell information and form structured cell information, the structured cell information including the number of cells, the number of cell features, and feature information of each cell feature; the learning unit is used to use the structured cell information as input data to establish a cell feature model using supervised learning, unsupervised learning, or semi-supervised learning; the identification unit is used to apply the cell feature model to classify or cluster cells of unknown type or unknown state; The cell force sensor is a cell traction force detection device having a light reflecting layer provided on a micropillar, and the information acquisition unit is for acquiring a light reflection signal reflected from the light reflecting layer, and the cell information is acquired based on the intensity of the light reflection signal.

7. The cell identification device according to claim 6 , wherein the cell traction force information also includes the direction of the cell traction force at that point.

8. The cell identification device according to claim 6, wherein the cell traction force information also includes a change in the magnitude or direction of the cell traction force at the point within a predetermined time interval.

9. The cell identification device according to claim 6, wherein the cell information also includes cell morphology information.

10. 7. The cell identification device according to claim 6, wherein the cell information is obtained after performing a cell control operation on the cell.

11. A cell identification system comprising a cell force sensor and the cell identification device according to any one of claims 6 to 10.

12. The cell identification system according to claim 11, further comprising a cell morphology information acquisition device for acquiring cell morphology information.

13. The cell identification system according to claim 12, wherein the cell morphology information acquisition device includes a microscope camera or a microscope video camera.

14. The system for cell identification according to claim 11, further comprising a cell control device for performing a cell control operation on the cell.

15. A method for detecting a state of a cell, comprising the steps of acquiring cell traction force information through the cell identification method according to any one of claims 1 to 5, the cell identification device according to any one of claims 6 to 10, or the cell identification system according to any one of claims 11 to 14, and analyzing and determining the state of the cell based on the cell traction force information; The cell state is characterized as comprising cell adhesion, cell vitality, cell differentiation / activation, cell proliferation and / or cell migration.

Images