Deep Learning-Based Method and Device for Cell Mechanical Phenotype Analysis

Abstract

This application is applicable to the field of cell analysis technology, and particularly relates to a method and device for cell mechanical phenotype analysis based on deep learning. The method includes: acquiring temporal images, mechanical response data, and culture environment parameters of target cells; based on the mechanical response data and culture environment parameters, performing regional segmentation on the temporal images to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of subcellular structures, thereby achieving refined and targeted feature recognition; generating a mechanical phenotype descriptor of the target cells according to the mechanically sensitive regions, local deformation characteristics, and culture environment parameters to realize multi-modal data fusion; using a mechanical phenotype analysis model pre-trained with sample mechanical phenotype descriptors of sample cells to process the mechanical phenotype descriptor, and obtaining a heterogeneity prediction result of the target cells, automatically identifying the heterogeneity of cell mechanical properties under complex culture environments, and significantly improving the dynamics and generalization ability of heterogeneity classification.

Description

Technical Field

[0001] This application belongs to the field of cell analysis technology, and particularly relates to a method and device for cell mechanical phenotype analysis based on deep learning. Background Art

[0002] Cell mechanical phenotype is the dynamic behavioral characteristics exhibited by cells under external mechanical stimuli, which is closely related to cell function, pathological state, and drug response. In recent years, with the in-depth research of biomechanics, the analysis of cell mechanical properties has become an important research direction in the fields of disease diagnosis (such as cancer, cardiovascular diseases), drug screening, and regenerative medicine.

[0003] In the prior art, the analysis of cell mechanical phenotype heterogeneity relies on single-point mechanical measurements (such as AFM) and artificial feature design, which requires pre-setting static mechanical parameters and is difficult to dynamically track local deformation, and is prone to misjudgment and missed detection in complex culture environments or at the subcellular scale, lacking effective mechanical feature analysis tools, resulting in low accuracy of heterogeneity classification and weak generalization ability. Summary of the Invention

[0004] The embodiments of this application provide a method and device for cell mechanical phenotype analysis based on deep learning, which can solve the problem of low accuracy of heterogeneity classification and weak generalization ability due to the lack of effective mechanical feature analysis tools during the analysis of cell mechanical phenotype heterogeneity.

[0005] In a first aspect, the embodiments of this application provide a method for cell mechanical phenotype analysis based on deep learning, including:

[0006] Obtaining the time-series images, mechanical response data, and culture environment parameters of the target cells;

[0007] Based on the mechanical response data and culture environment parameters, performing region segmentation on the time-series images to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of subcellular structures;

[0008] Generating a mechanical phenotype descriptor for the target cells according to the mechanically sensitive regions, the local deformation characteristics, and the culture environment parameters;

[0009] Processing the mechanical phenotype descriptor using a mechanical phenotype analysis model to obtain the heterogeneity prediction result of the target cells; wherein, the mechanical phenotype analysis model is a machine learning model pre-trained using the sample mechanical phenotype descriptors of sample cells.

[0010] The above technical solutions in the embodiments of this application have at least the following technical effects:

[0011] The method for analyzing cell mechanical phenotypes based on deep learning provided by the embodiments of the present application provides fundamental data basis for subsequent analysis by acquiring the time-series images, mechanical response data, and culture environment parameters of target cells. Based on the mechanical response data and culture environment parameters, the time-series images are segmented regionally to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of subcellular structures, realizing refined and targeted feature recognition. According to the mechanically sensitive regions, local deformation characteristics, and culture environment parameters, mechanical phenotype descriptors of the target cells are generated to achieve multimodal data fusion. The mechanical phenotype descriptors are processed using a mechanical phenotype analysis model that is pre-trained with sample mechanical phenotype descriptors of sample cells to obtain the heterogeneity prediction results of the target cells, automatically identifying the heterogeneity of cell mechanical properties in complex culture environments, and significantly improving the dynamics and generalization ability of heterogeneity classification.

[0012] In a second aspect, the embodiments of the present application provide a system for analyzing cell mechanical phenotypes based on deep learning, including:

[0013] An acquisition unit for acquiring the time-series images, mechanical response data, and culture environment parameters of target cells;

[0014] A segmentation unit for regionally segmenting the time-series images based on the mechanical response data and culture environment parameters to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of subcellular structures;

[0015] A generation unit for generating mechanical phenotype descriptors of the target cells according to the mechanically sensitive regions, the local deformation characteristics, and the culture environment parameters;

[0016] A result unit for processing the mechanical phenotype descriptors using a mechanical phenotype analysis model to obtain the heterogeneity prediction results of the target cells; wherein, the mechanical phenotype analysis model is a machine learning model pre-trained with sample mechanical phenotype descriptors of sample cells.

[0017] In a third aspect, the embodiments of the present application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the method described in any one of the above aspects is implemented.

[0018] In a fourth aspect, the embodiments of the present application provide a computer program product, which causes an electronic device to execute the method described in any one of the above aspects when the computer program product runs on the electronic device.

[0019] It can be understood that the beneficial effects of the above second aspect to the fourth aspect can be referred to the relevant descriptions in the above aspects and will not be elaborated here. Brief Description of the Drawings

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0021] Figure 1 It is a schematic flowchart of a method for analyzing cell mechanical phenotypes based on deep learning provided by an embodiment of the present application;

[0022] Figure 2 It is an operating schematic diagram of a method for analyzing cell mechanical phenotypes based on deep learning provided by an embodiment of the present application;

[0023] Figure 3 It is a schematic structural diagram of a system for analyzing cell mechanical phenotypes based on deep learning provided by an embodiment of the present application;

[0024] Figure 4 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed Embodiments

[0025] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0026] It should be understood that when used in the specification of the present application and the appended claims, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0027] It should also be understood that the term " / and / " used in the specification of the present application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0028] As used in the specification and appended claims of the present application, the term "if" may be construed as "when", "once", "in response to determining", or "in response to detecting" according to the context. Similarly, the phrase "if determined" or "if the described condition or event is detected" may be construed as meaning "once determined", "in response to determining", "once the described condition or event is detected", or "in response to detecting the described condition or event" according to the context.

[0029] In addition, in the description of the specification and appended claims of the present application, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be construed as indicating or implying relative importance.

[0030] The reference to "one embodiment" or "some embodiments" or the like described in the specification of the present application means that a specific feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the present application. Thus, the statements "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized.

[0031] In the prior art, the analysis of the heterogeneity of cell mechanical phenotypes relies on single-point mechanical measurements (such as AFM) and artificial feature design, requires pre-setting static mechanical parameters and is difficult to dynamically track local deformations, is prone to misjudgment in complex culture environments or at the subcellular scale, lacks effective mechanical feature analysis tools, resulting in low accuracy of heterogeneity classification and weak generalization ability.

[0032] To solve the above problems, the embodiments of the present application provide a method and device for analyzing cell mechanical phenotypes based on deep learning. In this method, by obtaining the time-series images, mechanical response data, and culture environment parameters of the target cells, fundamental data basis for subsequent analysis is provided. Based on the mechanical response data and culture environment parameters, region segmentation is performed on the time-series images to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of subcellular structures, realizing refined and targeted feature recognition. According to the mechanically sensitive regions, local deformation characteristics, and culture environment parameters, a mechanical phenotype descriptor of the target cells is generated, realizing multi-modal data fusion. The mechanical phenotype descriptor is processed using a mechanical phenotype analysis model pre-trained with the sample mechanical phenotype descriptors of sample cells to obtain the heterogeneity prediction result of the target cells, automatically identifying the heterogeneity of the mechanical properties of cells in complex culture environments, and significantly improving the dynamics and generalization ability of heterogeneity classification.

[0033] The method for analyzing cell mechanical phenotypes based on deep learning provided by the embodiments of this application can be applied to an electronic device. In this case, the electronic device is the execution subject of the method for analyzing cell mechanical phenotypes based on deep learning provided by the embodiments of this application. The embodiments of this application do not impose any restrictions on the specific type of the electronic device.

[0034] For example, the electronic device can be an ultra-mobile personal computer (UMPC), a netbook, a desktop computer, a computer, a laptop computer, a communication device, a computing device, a satellite wireless device, etc.

[0035] To better understand the method for analyzing cell mechanical phenotypes based on deep learning provided by the embodiments of this application, the following provides an exemplary introduction to the specific implementation process of the method for analyzing cell mechanical phenotypes based on deep learning provided by the embodiments of this application.

[0036] Figure 1 The schematic flowchart of the method for analyzing cell mechanical phenotypes based on deep learning provided by the embodiments of this application is shown. The method for analyzing cell mechanical phenotypes based on deep learning includes:

[0037] S100, obtaining the time-series images, mechanical response data, and culture environment parameters of the target cells.

[0038] It can be understood that the time-series images reflect the morphological changes of cells at different time points and can be obtained through a high-resolution microscope or a live-cell imaging system; the mechanical response data can be collected by devices such as a microfluidic chip, an atomic force microscope, or an optical tweezer, recording the stress-strain relationship of cells under mechanical stimulation; the culture environment parameters include temperature, pH value, nutrient concentration, etc., and can be monitored in real time through sensors. The time-series images, mechanical response data, and culture environment parameters together constitute the basic data set for cell mechanical behavior analysis, providing multi-dimensional data support for subsequent analysis, avoiding the limitations of a single data source, and improving the comprehensiveness and reliability of the analysis results.

[0039] S200, based on the mechanical response data and the culture environment parameters, performing region segmentation on the time-series images to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of subcellular structures.

[0040] It can be understood that by fusing mechanical response data (such as stress distribution) with environmental parameters (such as the hardness of the culture medium), and combining image segmentation algorithms (such as U-Net, watershed algorithm), the time-series images can be divided into different functional regions to determine the mechanically sensitive regions on the cell surface. The mechanically sensitive regions refer to specific regions on the cell surface that highly respond to stress changes (such as cell membrane wrinkles, focal adhesions), and the deformation characteristics of subcellular structures include the morphological changes or movement trajectories of organelles such as mitochondria and endoplasmic reticulum.

[0041] In a possible implementation, in S200, based on the mechanical response data and culture environment parameters, perform region segmentation on the time-series images to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of subcellular structures, including:

[0042] S210, perform time-varying spatial registration processing on the time-series images to generate the time-varying spatial correlation relationship between the cell deformation field of the target cell and the movement trajectory of subcellular structures.

[0043] It can be understood that the influence of cell movement or imaging drift can be eliminated through spatio-temporal alignment algorithms (such as dynamic time warping DTW) to establish the spatial correspondence relationship of images at different time points. The cell deformation field describes the vector field of the overall shape change of the cell, and the movement trajectory of subcellular structures records the curve of the position of subcellular structures changing with time in three-dimensional space. The time-varying spatial correlation relationship between the cell deformation field and the movement trajectory of subcellular structures can be achieved through coordinate mapping.

[0044] Optionally, in S210, perform time-varying spatial registration processing on the time-series images to generate the time-varying spatial correlation relationship between the cell deformation field of the target cell and the movement trajectory of subcellular structures, including:

[0045] S211, perform displacement field calculation and feature matching on the time-series images to determine the cell deformation field of the target cell and the movement trajectory of subcellular structures.

[0046] It can be understood that the displacement field calculation can adopt the optical flow method or the block matching algorithm. By tracking the displacement of pixels between adjacent images, the displacement field is determined; feature matching identifies the feature points of the cell skeleton or organelles through algorithms such as SIFT and SURF to establish the cross-frame correspondence relationship. The deformation field is obtained by integrating the displacement field, and the movement trajectory of subcellular structures is generated by fitting the coordinate sequence of feature points.

[0047] Exemplarily, in S211, perform displacement field calculation and feature matching on the time-series images to determine the cell deformation field of the target cell and the movement trajectory of subcellular structures, including:

[0048] S2111, perform per-pixel displacement calculation on adjacent time-series images to generate the initial displacement field of the target cell.

[0049] It can be understood that, based on the Lucas-Kanade optical flow algorithm, the displacement vector of each pixel between adjacent frames can be calculated to obtain an initial displacement field reflecting the local movement of cells. The initial displacement field quantifies the instantaneous deformation trend of cells under mechanical stimulation.

[0050] S2112: Identify subcellular structure feature points from the sequential images, and determine the sequence of motion trajectory coordinates of the subcellular structure according to the subcellular structure feature points.

[0051] It can be understood that the feature points of subcellular structures can be accurately located from the sequential images of cells, and the sequence of their motion trajectory coordinates can be constructed based on these feature points, so as to realize the tracking of the dynamic changes of subcellular structures.

[0052] Exemplarily, when identifying the feature points of subcellular structures, the time-series images can be preprocessed first. Gaussian filtering is used to remove the noise in the images to avoid noise interfering with the identification of feature points. A feature point detection algorithm based on grayscale, such as the Harris corner detection algorithm, is utilized. The Harris corner detection algorithm determines the corner points with significant features by calculating the grayscale changes of each pixel point in the image in different directions. For subcellular structures, these corner points often locate at the edges, corners, etc. of the structures and can effectively represent the features of subcellular structures. During the calculation process, the response value of each pixel point will be obtained. A suitable threshold can be preset in advance, and only the pixel points with response values greater than this threshold will be confirmed as feature points. In this way, the feature points of subcellular structures can be initially screened out from the time-series images. The morphological features of subcellular structures can be combined. For example, mitochondria usually present as elliptical or rod-shaped structures. According to this characteristic, further screening is carried out among the detected feature points to remove those points that do not conform to the morphological features of mitochondria. The shape factor, aspect ratio, etc. of the area around the feature points can be calculated and compared with the known morphological parameter range of mitochondria, so as to retain the feature points that truly belong to mitochondria. When determining the coordinate sequence of the movement trajectory of subcellular structures, a feature matching-based method can be adopted. For two adjacent frames of images, first, the feature points of subcellular structures are detected in the first frame of image, and then the feature points that match them are searched for in the second frame of image. A matching algorithm based on feature descriptors, such as the SIFT (Scale-Invariant Feature Transform) descriptor or the ORB (Oriented FAST and Rotated BRIEF) descriptor, is used. Taking the ORB descriptor as an example, it has a certain invariance to the rotation and scale changes of the image and has a relatively high calculation efficiency. By calculating the ORB descriptor of the feature points in the first frame of image, the most similar descriptor is searched for in the second frame of image to find the matching feature points. During the matching process, a matching threshold can be preset in advance, and only the feature point pairs with similarity higher than this threshold are considered valid matches. If in the second frame of image, the spatial position of a matching point differs too much from the corresponding point in the first frame of image, exceeding the reasonable movement range, then this match is considered incorrect and is excluded. At the same time, the tracking window technology can be adopted. In the second frame of image, a tracking window of a certain size is set centered on the position of the feature points in the first frame of image, and feature point matching is only carried out within this window. This can reduce the calculation amount, improve the matching efficiency, and can better cope with the overall movement and local deformation of cells.

[0053] When processing images of multi-subcellular structures, feature point matching conflicts may occur, that is, multiple feature points match to the same point in the second frame image. At this time, a conflict resolution strategy based on distance and similarity is applied. Calculate the distance between each conflicting feature point and the matching point as well as the similarity of the descriptors, and select the feature point with the closest distance and the highest similarity as the correct matching point, so as to ensure that the feature points of each subcellular structure can be accurately matched to the corresponding points in the next frame image, and then generate an accurate sequence of motion trajectory coordinates, identify the feature points of subcellular structures from temporal images, and determine their sequence of motion trajectory coordinates, providing a reliable data basis for subsequent analysis of the motion patterns, deformation characteristics of subcellular structures, and the relationship with the overall mechanical response of cells, etc.

[0054] S2113, perform displacement correction based on the initial displacement field and the sequence of motion trajectory coordinates of subcellular structures, and determine the cell deformation field and the motion trajectory of subcellular structures of the target cell.

[0055] It can be understood that the obtained initial displacement field and the sequence of motion trajectory coordinates of subcellular structures can be optimized to more accurately determine the cell deformation field and the motion trajectory of subcellular structures of the target cell. The least squares method can be used to optimize the obtained initial displacement field and the sequence of motion trajectory coordinates of subcellular structures. The sequence of motion trajectory coordinates of subcellular structures can be used as the actual observed values, and the predicted positions of subcellular structures calculated based on the initial displacement field can be used as the predicted values of the theoretical model. By minimizing the sum of the squared errors between the two, adjust the parameters in the initial displacement field so that the adjusted initial displacement field can more accurately reflect the true motion of subcellular structures.

[0056] Exemplarily, for each coordinate point in the subcellular structure movement trajectory coordinate sequence, its predicted position in the current frame can be calculated according to the initial displacement field. The error between the predicted position and the actual coordinate point is calculated, and the squares of all errors are accumulated. The parameters of the initial displacement field, such as the amplitude and direction of the displacement field, are continuously adjusted in an iterative manner, so that the sum of the squared errors gradually decreases. When the sum of the squared errors reaches a preset threshold, it is considered that the displacement correction has achieved a satisfactory effect, and the optimized displacement field obtained at this time is the corrected displacement field. After obtaining the corrected displacement field, it is used to determine the cell deformation field of the target cell. The cell deformation field describes the deformation of the cell at different positions, and it is closely related to the displacement field. By calculating the spatial derivative of the corrected displacement field, such as using the finite difference method, the change rate of the displacement in different directions is calculated, so as to obtain the cell deformation field. For example, the first-order derivatives of the displacement in the x direction and the y direction are calculated, and this derivative information can reflect the stretching, compression and other deformation conditions of the cell at each position, and then the cell deformation field is constructed. For the subcellular structure movement trajectory, it is optimized based on the corrected displacement field. According to the corrected displacement field, the subcellular structure movement trajectory coordinate sequence is adjusted. If there is a deviation in the position of a certain subcellular structure due to errors in the initial movement trajectory coordinate sequence, under the action of the corrected displacement field, this position will be corrected. The adjusted movement trajectory can be smoothed. For example, methods such as moving average filtering are used. For each point in the movement trajectory coordinate sequence, the average value of several points before and after it is taken as the new coordinate value, so that the movement trajectory is smoother and more in line with the actual movement of the subcellular structure. Through the displacement correction operation of the initial displacement field and the subcellular structure movement trajectory coordinate sequence, the errors in the data can be effectively eliminated, and the cell deformation field and the subcellular structure movement trajectory of the target cell can be accurately determined.

[0057] S212. Align the coordinates of the cell deformation field and the subcellular structure movement trajectory to establish a time-varying spatial mapping relationship between the cell deformation field and the subcellular structure.

[0058] It can be understood that after obtaining the cell deformation field and the subcellular structure movement trajectory, since they may be measured and calculated in different coordinate systems, or affected by factors such as the movement of the cell itself and the change of imaging angle, there is a lack of a unified spatial reference between the two, making it difficult to directly conduct correlation analysis. Coordinate alignment operations can be performed to establish an accurate time-varying spatial mapping relationship between the two. The centroid of the cell or the position of a stable landmark subcellular structure (such as the cell nucleus) can be used as a reference point to construct a reference coordinate system. Transform the data of the cell deformation field and the subcellular structure movement trajectory. For the cell deformation field, through geometric transformation operations such as translation, rotation, and scaling, its coordinates are converted to the selected reference coordinate system. The translation operation is to move the coordinate origin of the cell deformation field to the reference point position; the rotation operation is to rotate each point in the deformation field according to the overall rotation angle of the cell to eliminate the influence of cell rotation on the direction of the deformation field; the scaling operation is to unify the scale to ensure the comparability of the deformation field data at different time points or under different experimental conditions. For the subcellular structure movement trajectory, the above geometric transformations are also performed. By calculating the relative position relationship between the subcellular structure movement trajectory and the reference point, its coordinates are adjusted to be consistent with the reference coordinate system. After completing the coordinate alignment, a time-varying spatial mapping relationship between the cell deformation field and the subcellular structure can be established. The establishment of the time-varying spatial mapping relationship is achieved by corresponding each point on the subcellular structure movement trajectory to the point at the corresponding position in the cell deformation field at the same moment. For example, at a certain moment, the subcellular structure is located at a specific position on the movement trajectory. Through the coordinate information after coordinate alignment, the corresponding position in the cell deformation field is found, so as to determine the cell deformation situation experienced by the subcellular structure at this moment. In this way, the mapping from the movement trajectory of the subcellular structure to the cell deformation field is realized, and a time-varying spatial mapping relationship between the two is established.

[0059] S213. Perform multi-scale strain analysis on the mapping relationship to generate the time-varying spatial correlation relationship between the cell deformation field and the subcellular structure movement trajectory of the target cell.

[0060] It can be understood that the core of multi-scale strain analysis lies in observing and quantifying cell deformation and subcellular structure movement from different spatial scales and time scales. The mechanical behavior of cells may exhibit different characteristics at different scales. At small scales, more attention may be paid to the local strain situation near the subcellular structure, while at large scales, the overall strain trend of the cell is emphasized.

[0061] Exemplarily, in terms of spatial scale, a multi-resolution analysis method such as wavelet transform can be adopted. Wavelet transform can decompose the data of cell deformation fields and subcellular structure movement trajectories into components of different frequencies, and each frequency component corresponds to a different spatial scale. The low-frequency components reflect the overall change trend at large scales, while the high-frequency components capture the detailed changes at small scales. By analyzing different frequency components, the strain conditions of cells at different spatial scales can be studied separately. For example, for cell deformation fields, the low-frequency wavelet coefficients can show the macroscopic deformation characteristics such as the overall stretching and compression of cells, while the high-frequency wavelet coefficients can reveal the subtle strain changes in local regions such as cell edges and around subcellular structures. For the movement trajectories of subcellular structures, wavelet analysis at different scales can help determine the differences in movement patterns of subcellular structures in the overall cell movement and local environmental changes.

[0062] Exemplarily, on the time scale, a time series analysis method such as the autoregressive moving average model (ARMA) can be utilized. The ARMA model can model the changes of subcellular structure movement trajectories and cell deformation fields over time, and analyze their correlations and change trends at different time scales. Through the parameters of the model, the rapid changes in a short period and the slow evolution over a long time can be captured. Combining the analysis results of spatial scale and time scale, a time-varying spatial correlation relationship between the cell deformation field and the movement trajectory of subcellular structures of the target cell is generated. The time-varying spatial correlation relationship not only includes the corresponding relationship between cell deformation and subcellular structure movement at different spatial scales, but also reflects their mutual influence and change laws in the time dimension. For example, the movement response of subcellular structures during rapid local strain of cells can be discovered, as well as the long-term movement trend of subcellular structures during the slow overall deformation of cells.

[0063] S220, based on the time-varying spatial correlation relationship and mechanical response data, calculate the strain gradient tensor on the surface of the target cell to obtain the initial segmentation boundary of the mechanically sensitive region.

[0064] It can be understood that when a cell is subjected to external mechanical stimuli or internal mechanical changes within itself, different regions on its surface will generate different degrees of strain, and the strain gradient can reflect the severity of these strain changes. The mechanically sensitive region is the region on the cell surface that is more sensitive and responds strongly to strain changes. The strain gradient tensor can be calculated by combining the time-varying spatial correlation relationship (which integrates the morphological changes of cells and the movement information of subcellular structures in the time and space dimensions) and mechanical response data (such as stress distribution, elastic modulus, etc.), and then the initial boundary of the mechanically sensitive region is divided, providing a key basis for in-depth study of cell mechanical properties and functions, helping to understand how cells sense and respond to mechanical signals, and the roles of these mechanically sensitive regions in cell physiological and pathological processes.

[0065] Optionally, in S220, based on the time-varying spatial correlation relationship and the mechanical response data, calculate the strain gradient tensor on the surface of the target cell to obtain the initial segmentation boundary of the mechanically sensitive region, including:

[0066] S221, based on the time-varying spatial correlation relationship, obtain the strain tensor components on the surface of the target cell.

[0067] It can be understood that the time-varying spatial correlation relationship details the morphological changes of the cell at different times and the movement trajectories of subcellular structures. This information reflects the deformation of the cell in the spatio-temporal dimension. The strain tensor is a mathematical tool for describing the deformation state of an object, which can quantify the stretching, compression, and shear deformation degrees in different directions on the cell surface. Based on the time-varying spatial correlation relationship, by analyzing the position changes of each point on the cell surface at different times, the strain tensor components can be calculated using geometric and mechanical principles. For example, by tracking the displacements of specific marker points on the cell surface at different time points and combining the relative position relationships between these points, and using tensor analysis methods, the normal strain components (measuring stretching or compression deformation) and shear strain components (measuring shear deformation) in each direction can be determined. The strain tensor components provide the basic data for subsequent calculation of the strain gradient tensor. They accurately describe the deformation state of the cell surface in different directions, which helps to further analyze the distribution characteristics of the cell surface strain.

[0068] S222, perform a spatial derivative operation on the strain tensor components on the surface of the target cell to generate a strain gradient tensor field.

[0069] It can be understood that the strain tensor components describe the strain state of each point on the cell surface. However, to determine the severity and direction of the strain change, the strain gradient can be calculated. The spatial derivative operation can quantify the rate of change of the strain tensor components in space. The derivative is calculated for each strain tensor component (including the normal strain component and the shear strain component) in each direction in space (such as the x and y directions, and also the z direction if it is a three-dimensional cell). For example, using the finite difference method, by calculating the ratio of the difference in strain tensor components at adjacent positions to the distance, the derivative of the strain tensor component in this direction can be approximated. After performing such operations on all strain tensor components and directions, these derivatives are combined to form the strain gradient tensor field. The strain gradient tensor field comprehensively reflects the intensity and direction information of the strain change on the cell surface, providing more detailed and accurate data support for determining the mechanically sensitive region. In the strain gradient tensor field, regions with larger numerical values indicate more drastic strain changes, and these regions are likely to be the mechanically sensitive regions, which will be further analyzed based on this in the follow-up.

[0070] S223, based on the strain gradient tensor field and the mechanical response data, determine the segmentation threshold of the mechanically sensitive region.

[0071] It can be understood that the strain gradient tensor field shows the distribution of strain changes on the cell surface, and a suitable segmentation threshold can be determined to divide the mechanically sensitive regions. The mechanical response data contains various response information of the cell to mechanical stimuli such as stress, such as the stress distribution of the cell and the change of elastic modulus. The segmentation threshold can be determined through the strain gradient tensor field and the mechanical response data to make the division result more in line with the actual mechanical properties of the cell. For example, by analyzing a large amount of experimental data, the relationship between the strain gradient and the mechanical response of the cell can be found. For example, when the strain gradient reaches a certain value, the cell will produce specific physiological responses (such as gene expression changes, signal pathway activation, etc.). Statistical methods (such as histogram analysis, clustering analysis) can be used, combined with the mechanical response data, to determine a threshold that can effectively distinguish between mechanically sensitive regions and non-sensitive regions. In addition, machine learning methods can also be used. Using the data samples of known mechanically sensitive regions and non-sensitive regions to train the model, the model can automatically learn and determine the optimal segmentation threshold. This segmentation threshold will be an important basis for subsequent division of mechanically sensitive regions and directly affect the accuracy of the final analysis results.

[0072] S224. Divide the surface region of the target cell according to the segmentation threshold to obtain the initial segmentation boundary of the mechanically sensitive region.

[0073] It can be understood that after determining the segmentation threshold, the surface region of the target cell can be divided according to this threshold, so as to obtain the initial segmentation boundary of the mechanically sensitive region. Compare the strain gradient value of each point in the strain gradient tensor field with the segmentation threshold. If the strain gradient value of a certain point is greater than the segmentation threshold, the region where the point is located is marked as a possible mechanically sensitive region; otherwise, it is marked as a non-sensitive region. Through such a comparison and marking process, the cell surface is divided into different regions. Then, image processing and boundary extraction algorithms (such as contour detection algorithms) are used to extract the boundary from these marked regions. For example, the Canny edge detection algorithm is used, which can detect the edge of an object according to the change of gray value in the image (here it is the change of strain gradient value), so as to obtain the initial segmentation boundary of the mechanically sensitive region. The initial segmentation boundary preliminarily defines the range of the mechanically sensitive region and provides a basic framework for further studying the characteristics, functions of the mechanically sensitive region and its relationship with other parts of the cell.

[0074] S230. Optimize the initial segmentation boundary according to the strain gradient tensor and the culture environment parameters to determine the mechanically sensitive region on the surface of the target cell.

[0075] It can be understood that after obtaining the initial segmentation boundary of the mechanically sensitive region, since the culture environment in which the cells are located will affect the mechanical properties of the cells, and the initial segmentation boundary may have problems such as insufficient accuracy and noise interference, it is necessary to optimize it by combining the strain gradient tensor and the culture environment parameters to more accurately determine the mechanically sensitive region. The strain gradient tensor reflects the severity of the strain change on the cell surface, and the culture environment parameters (such as temperature, pH value, culture medium components, etc.) will change the physiological state and mechanical response of the cells. Considering both of them can improve the accuracy and reliability of determining the mechanically sensitive region, which is of great significance for in-depth study of the mechanical behavior of cells under different environments.

[0076] Optionally, S230, optimize the initial segmentation boundary according to the strain gradient tensor and the culture environment parameters to determine the mechanically sensitive region on the surface of the target cell, including:

[0077] S231, dynamically adjust the segmentation threshold of the initial segmentation boundary based on the culture environment parameters to generate an optimized threshold for the mechanically sensitive region.

[0078] It can be understood that the change of the culture environment parameters will affect the mechanical properties of the cells, and then change the distribution and characteristics of the mechanically sensitive region. For example, the change of temperature may lead to the change of the fluidity of the cell membrane and affect the response of the cells to mechanical stimuli; the difference in the components of the culture medium may affect the metabolic activities of the cells and the stability of the cytoskeleton, thus changing the mechanical properties of the cells. Based on this, it is necessary to dynamically adjust the segmentation threshold set during the initial segmentation boundary according to the culture environment parameters. Statistical analysis, machine learning and other methods can be used to establish a regression model (such as multiple linear regression) between the environmental parameters (such as pH value, temperature) and the strain gradient threshold to determine the influence law of different culture environment parameters (such as temperature, pH value, specific nutrient concentration, etc.) on the cell strain gradient. For example, it is found that when the temperature rises, the elastic modulus of the cells decreases, the response to stress is more sensitive, and the strain gradient threshold of the mechanically sensitive region may need to be adjusted accordingly. According to the current culture environment parameters, adjust the initial segmentation threshold according to the above relationship model. If the temperature in the current culture environment is relatively high, according to the model prediction, the response of the cells to mechanical stimuli is enhanced, then appropriately reduce the segmentation threshold so that more parts with relatively low strain gradients but still likely to be mechanically sensitive regions in the current environment are included; on the contrary, if the temperature is low, the response of the cells to mechanical stimuli is weakened, then appropriately increase the segmentation threshold to exclude some parts that may have abnormal strain gradients due to environmental factors but are not truly mechanically sensitive regions. Through dynamic adjustment, an optimized threshold for the mechanically sensitive region that more conforms to the actual situation of the cells in the current culture environment is generated.

[0079] S232, perform morphological closing operation on the initial segmentation boundary according to the optimized threshold to generate an optimized boundary for the mechanically sensitive region.

[0080] It can be understood that morphological closing operation can be used to process the initial segmentation boundary to eliminate noise and fill holes, making the boundary of the mechanically sensitive region more accurate and continuous. The morphological closing operation consists of two basic morphological operations: dilation and erosion.

[0081] Exemplarily, the dilation operation expands the boundaries of objects in the image. For the mechanically sensitive region defined by the initial segmentation boundary, the dilation operation can fill the boundary holes caused by noise or inaccurate segmentation and connect some discrete small regions. For example, if there are some small holes in the initial segmentation boundary, which may be caused by measurement errors or noise interference of local strain gradients, the dilation operation can fill these holes and make the shape of the mechanically sensitive region more complete. The erosion operation shrinks the object boundary inward. Performing the erosion operation after dilation can remove some false edges and noise points introduced by dilation, making the boundary smoother and more accurate. Through the combination of dilation and erosion (i.e., closing operation), the quality of the initial segmentation boundary can be effectively improved. The dilation and erosion operations can be performed with a suitable structuring element (such as a circle, square, etc.). By performing morphological closing operation on the initial segmentation boundary, an optimized boundary that is more accurate and can better represent the actual mechanically sensitive region is generated.

[0082] S233. Perform boundary fusion on the optimized boundary according to the strain gradient tensor to determine the mechanically sensitive region on the surface of the target cell.

[0083] It can be understood that in the strain gradient tensor, the change trends of the strain gradients on both sides of the optimized boundary can be identified. If there are obvious differences in the strain gradients on both sides of the boundary and this difference conforms to the characteristics of the mechanically sensitive region, it indicates that the current boundary division may be reasonable; however, if the change of the strain gradient is not obvious or there are abnormalities, the boundary may need to be adjusted. The boundary fusion can be performed according to the information of the strain gradient tensor. For adjacent parts that may belong to the mechanically sensitive region, if the change of the strain gradient between them is continuous and conforms to the characteristics of the mechanically sensitive region, the boundaries of these parts are fused to form a larger and more continuous mechanically sensitive region. For example, there are some small segments of the boundary on the optimized boundary, and the strain gradient changes of the regions they define are similar and significantly different from the surrounding regions. These small segments of the boundary can be merged through boundary fusion to make the boundary of the mechanically sensitive region more coherent.

[0084] S240. Extract the motion trajectories of subcellular structures from the time-varying spatial correlation relationship, calculate the curvature change parameters of the subcellular structures, and determine the local deformation characteristics of the subcellular structures.

[0085] It is understandable that the movement and deformation of subcellular structures play a crucial role in the physiological activities of cells. Studying their movement trajectories and local deformation characteristics helps to deeply understand cell functions and physiological processes. The time-varying spatial correlation relationship integrates the information of cells in the time and space dimensions, providing a rich data basis for extracting the movement trajectories of subcellular structures. By calculating the curvature change parameters of subcellular structures, the degree of curvature change of their movement trajectories can be quantified, thereby determining the local deformation characteristics. The local deformation characteristics can reflect the morphological changes of subcellular structures during movement, which is of great significance for revealing the mechanical mechanisms and physiological functions within cells.

[0086] Optionally, in S240, extract the movement trajectory of the subcellular structure from the time-varying spatial correlation relationship, calculate the curvature change parameters of the subcellular structure, and determine the local deformation characteristics of the subcellular structure, including:

[0087] In S241, extract the movement trajectory of the subcellular structure from the time-varying spatial correlation relationship, and determine the movement trajectory coordinate sequence of the subcellular structure of the target cell.

[0088] It is understandable that the time-varying spatial correlation relationship details the spatial position information of subcellular structures at different times and their association with the overall cell deformation. The part related to the movement of subcellular structures can be screened out from the time-varying spatial correlation relationship to obtain the specific positions of subcellular structures at each time point, and then the movement trajectory coordinate sequence can be determined. Since the time-varying spatial correlation relationship may contain a large amount of redundant information, specific algorithms and rules can be used to extract useful data. For example, according to the characteristic markers of subcellular structures or preset recognition methods, the positions of subcellular structures can be accurately found in the spatial data at different time points. For subcellular structures such as mitochondria, their specific fluorescence markers can be utilized, and the positions of mitochondria can be located in the image data corresponding to the time-varying spatial correlation relationship through image processing techniques. Arranging the position information at each time point in chronological order forms the movement trajectory coordinate sequence of the subcellular structure. This coordinate sequence precisely records the position changes of the subcellular structure at different times, providing basic data for subsequent analysis of its movement characteristics and deformation characteristics.

[0089] In S242, perform spline interpolation on the movement trajectory coordinate sequence to generate the movement trajectory curve of the subcellular structure, and calculate the trajectory curvature parameters of the subcellular structure based on the movement trajectory curve.

[0090] It can be understood that the obtained coordinate sequence of the subcellular structure movement trajectories is usually discrete data points, and it is difficult for these discrete points to intuitively reflect the continuity and smoothness of the subcellular structure movement. Spline interpolation is a commonly used mathematical method, through which a smooth curve can be constructed between these discrete coordinate points to more accurately describe the movement trajectory of the subcellular structure. The principle of spline interpolation is to use piecewise polynomial functions to fit the discrete data points, so that the curve maintains a certain smoothness while passing through each data point. There are methods such as cubic spline interpolation for spline interpolation. Cubic spline interpolation constructs cubic polynomial functions between each data point to ensure that the curve has continuous first and second derivatives throughout the interval, thus ensuring the smoothness of the curve. In practical applications, an appropriate spline interpolation method is selected according to the characteristics of the movement trajectory coordinate sequence to generate the movement trajectory curve of the subcellular structure. After obtaining the movement trajectory curve, the trajectory curvature parameter can be calculated to quantify the degree of curve bending. Curvature is an important parameter for describing the degree of curve bending. For the movement trajectory curve of the subcellular structure, the curvature parameter can reflect its bending change during the movement process. There are various methods for calculating curvature. For example, for a parametric curve, the first and second derivatives of the curve can be used to calculate the curvature. Through the calculated trajectory curvature parameter, the degree of bending of the subcellular structure movement trajectory at different positions and times can be intuitively understood, providing a quantitative basis for analyzing its local deformation characteristics.

[0091] S243, perform a time derivative operation according to the trajectory curvature parameter of the subcellular structure to determine the local deformation characteristics of the subcellular structure.

[0092] It can be understood that the local deformation of subcellular structures is not only related to the curvature of the movement trajectory but also closely related to the change of curvature over time. By performing a time derivative operation on the trajectory curvature parameter, the rate of change of curvature over time can be obtained, thereby more accurately determining the local deformation characteristics of subcellular structures. The time derivative operation reflects the change trend of the trajectory curvature in the time dimension. When local deformation occurs in subcellular structures, the curvature of their movement trajectories changes over time, and this change can be more prominently reflected through the time derivative operation. For example, when mitochondria are in the process of fission or fusion, the curvature of their movement trajectories changes rapidly. After performing a time derivative operation on the curvature parameter, the rate and direction of this change can be captured. According to the result of the time derivative operation, the local deformation characteristics of subcellular structures can be determined. If the time derivative of the curvature is large, it indicates that the degree of bending of the subcellular structure has changed significantly in a short period of time, which may mean that the subcellular structure is undergoing relatively intense local deformation, such as the constriction or elongation of mitochondria; conversely, if the time derivative is small, it means that the change in the degree of bending of the subcellular structure is relatively slow and the local deformation is relatively small. By using the time derivative of the trajectory curvature parameter as an important indicator, the local deformation characteristics of subcellular structures can be effectively determined, providing key information for in-depth study of the dynamic changes of subcellular structures and the physiological processes of cells.

[0093] S300. Generate a mechanical phenotype descriptor for the target cell according to the mechanically sensitive region, local deformation characteristics, and culture environment parameters.

[0094] It can be understood that the mechanical phenotype of a cell is an embodiment of its comprehensive characteristics in a mechanical environment, reflecting the cell's response to mechanical stimuli and its own mechanical properties. The mechanically sensitive region reflects the parts of the cell surface that are sensitive to mechanical stimuli, the local deformation characteristics show the changes in subcellular structures under mechanical action, and the culture environment parameters affect the physiological state and mechanical behavior of the cell. Generating a mechanical phenotype descriptor by integrating the information of the mechanically sensitive region, local deformation characteristics, and culture environment parameters can comprehensively and quantitatively characterize the mechanical phenotype of the cell, providing a unified quantitative index for subsequent research on the mechanical properties, functions of the cell, and its changes in different environments, which helps to deeply understand the physiological and pathological processes of the cell and the mechanical heterogeneity between cells.

[0095] In a possible implementation, S300. Generate a mechanical phenotype descriptor for the target cell according to the mechanically sensitive region, local deformation characteristics, and culture environment parameters, including:

[0096] S310. Perform feature correlation according to the mechanically sensitive region, local deformation characteristics, and culture environment parameters to generate a heterogeneity correlation matrix for the target cell.

[0097] It is understandable that there are complex interrelationships among the mechanically sensitive regions of cells, local deformation characteristics, and culture environment parameters, and these relationships affect the mechanical phenotypes and heterogeneity of cells. The purpose of feature correlation is to explore the internal connections among these factors. By analyzing the correlations, causal relationships, etc. among different factors, a matrix is constructed to represent the degree of association among them, that is, the heterogeneity association matrix.

[0098] Exemplarily, the characteristics of the mechanically sensitive region (such as area, position, strain gradient, etc.), local deformation characteristics (such as the curvature change rate and deformation amplitude of subcellular structures, etc.), and culture environment parameters (temperature, pH value, nutrient concentration, etc.) can be used as the element dimensions of the matrix. Calculate the degree of association among the elements. For example, the Pearson correlation coefficient is used to measure the linear correlation between two features. If the change trends of two features are similar, the correlation coefficient is higher, and the corresponding element value in the matrix is also larger; conversely, if there is no obvious association between two features, the element value is smaller. In this way, a heterogeneity association matrix that can comprehensively reflect the association relationships among various factors is generated, and this matrix provides the basic data structure for subsequent in-depth analysis of cell heterogeneity.

[0099] S320, perform feature dimensionality reduction on the heterogeneity association matrix, and perform time-frequency joint feature encoding on the dimension-reduced heterogeneity association matrix to generate a mechanical phenotype descriptor of the target cell.

[0100] It can be understood that the heterogeneous association matrix contains the association information between multiple features. However, its dimension is relatively high, and there is information redundancy, which is not conducive to subsequent analysis and processing. Feature dimensionality reduction aims to reduce the dimension of the data while retaining the key information. Feature dimensionality reduction methods can include principal component analysis (PCA), linear discriminant analysis (LDA), etc. Taking principal component analysis as an example, it transforms the original high-dimensional data into a set of uncorrelated low-dimensional data through linear transformation. These low-dimensional data are called principal components. Each principal component is a linear combination of the original features and retains as much variance information of the original data as possible. By performing dimensionality reduction on the heterogeneous association matrix through principal component analysis, redundant information can be removed, the data structure can be simplified, and at the same time, the key features that have a greater impact on the cell mechanical phenotype can be retained. Time-frequency joint feature encoding is a method for further processing the data after dimensionality reduction. The mechanical behavior of cells has specific features in both the time and frequency dimensions. Time-frequency joint feature encoding can capture this information simultaneously. For example, using methods such as wavelet transform to convert the data after dimensionality reduction from the time domain to the time-frequency domain. Wavelet transform can decompose the signal into the coefficients of different frequency components at different time points, and these coefficients reflect the energy distribution of the signal at different times and frequencies. By encoding these time-frequency coefficients and combining them into a feature vector, this feature vector is the mechanical phenotype descriptor of the target cell. The mechanical phenotype descriptor integrates the information of the cell mechanical features in time and frequency, more comprehensively describes the mechanical phenotype of the cell, and provides a more effective data representation form for subsequent analysis and model training.

[0101] S400, use the mechanical phenotype analysis model to process the mechanical phenotype descriptor to obtain the heterogeneous prediction result of the target cell; wherein, the mechanical phenotype analysis model is a machine learning model pre-trained using the sample mechanical phenotype descriptors of the sample cells.

[0102] It can be understood that machine learning models have powerful pattern recognition and prediction capabilities. By pre-training the model with a large number of sample mechanical phenotype descriptors of sample cells, the model can learn the potential relationship between the mechanical phenotype descriptors and cell heterogeneity. When the mechanical phenotype descriptors of the target cells are input, the model predicts the heterogeneity of the target cells based on the patterns it has learned. Different types of machine learning models, such as neural networks, support vector machines, etc., can be used to construct a mechanical phenotype analysis model. Taking a neural network as an example, it consists of multiple neurons and learns the patterns in the data by adjusting the connection weights between the neurons. During the training process, the mechanical phenotype descriptors of the sample cells are used as input, and the corresponding cell heterogeneity labels (such as the degree of cell differentiation, functional state, etc.) are used as output. The model continuously adjusts the weights to minimize the error between the predicted result and the actual label. After training is completed, the model has the ability to predict cell heterogeneity based on the input mechanical phenotype descriptors. For the target cells, their mechanical phenotype descriptors are input into the trained model. The model undergoes internal calculations and processing and outputs the predicted result of the heterogeneity of the target cells, helping researchers understand the differences in mechanical properties between the target cells and other cells and their potential biological states, and improving the accuracy and generalization of heterogeneity classification.

[0103] Before using the mechanical phenotype analysis model to process the mechanical phenotype descriptors and obtain the predicted result of the heterogeneity of the target cells, the method further includes:

[0104] S510, obtaining the mechanical phenotype descriptors of the sample cells.

[0105] It can be understood that the mechanical phenotype descriptors of the sample cells are the basic data for training the mechanical phenotype analysis model. The process of obtaining the mechanical phenotype descriptors of the sample cells can be encoded by obtaining information such as the mechanically sensitive regions, local deformation characteristics, and culture environment parameters of the sample cells. For example, when studying the mechanical phenotype of tumor cells, the sample cells can include tumor cells at different tumor stages and from different tissue sources, as well as normal cells as controls. By obtaining the mechanical phenotype descriptors of a large number of diverse sample cells, the model can learn a wider range of relationships between cell mechanical phenotypes and heterogeneity, improve the generalization ability of the model, and make it more accurate and reliable when predicting the heterogeneity of target cells.

[0106] S520, determining the heterogeneity analysis label of the sample cells.

[0107] It can be understood that the heterogeneity analysis label is information used to mark the heterogeneous characteristics of sample cells, which provides a supervision signal for model training. The heterogeneity of sample cells can be reflected in multiple aspects, such as the degree of cell differentiation, proliferation ability, drug sensitivity, disease state, etc. For the degree of cell differentiation, it can be determined by detecting the expression level of cell-specific markers. For example, during the differentiation of neural stem cells, the expression of specific neural marker proteins changes. By using experimental methods such as immunofluorescence and Western blot to detect the expression of these markers, the differentiation stage of the cells can be judged and used as a heterogeneity analysis label. For tumor cells, the heterogeneity analysis label can be determined according to the grading, staging, and gene mutation status of the tumor.

[0108] S530, using the mechanical phenotype descriptor of the sample cells as the input and the corresponding heterogeneity analysis label as the expected output, trains the initial machine learning model to obtain the trained mechanical phenotype analysis model.

[0109] It can be understood that using the mechanical phenotype descriptor of the sample cells as the input of the model and the corresponding heterogeneity analysis label as the expected output, the model continuously adjusts its internal parameters (such as the weights of the neural network) to make the predicted output of the model as close as possible to the expected output. During the training process, a preset loss function can be used to measure the difference between the model prediction result and the expected output, such as mean square error (MSE), cross-entropy loss, etc. The model calculates the gradient of the loss function with respect to the parameters through the backpropagation algorithm and updates the parameters according to the gradient to gradually reduce the loss function. As the training progresses, the model continuously learns the patterns in the sample data and understands the relationship between the mechanical phenotype descriptor and the heterogeneity analysis label more and more deeply. After multiple rounds of training, when the loss of the model on the training set reaches a certain convergence criterion, or the performance on the validation set no longer improves, it is considered that the model training is completed, and the trained mechanical phenotype analysis model is obtained. The training data can be shown in the following table:

[0110]

[0111] The mechanical phenotype descriptor is a feature vector that fuses the mechanically sensitive region, local deformation characteristics, and culture environment parameters. For example:

[0112] The first 3 dimensions: the proportion of the area of the mechanically sensitive region, the average value of the strain gradient, and the local deformation frequency;

[0113] The last 3 dimensions: the culture temperature (°C), pH value, and glucose concentration (mM);

[0114] The heterogeneity label is the degree of differentiation: including low differentiation, medium differentiation, and high differentiation.

[0115] The trained model has the ability to predict cell heterogeneity based on the input mechanical phenotype descriptors, which can be used for predictive analysis of the heterogeneity of target cells and provide valuable references for cell biology research.

[0116] Corresponding to the deep learning-based cell mechanical phenotype analysis method in the above embodiment, the embodiment of the present application also provides a deep learning-based cell mechanical phenotype analysis system, and each unit of this system can implement each step of the deep learning-based cell mechanical phenotype analysis method. Figure 3 The structural block diagram of the deep learning-based cell mechanical phenotype analysis system provided by the embodiment of the present application is shown. For ease of description, only the parts related to the embodiment of the present application are shown.

[0117] Refer to Figure 3 , the deep learning-based cell mechanical phenotype analysis system includes:

[0118] An acquisition unit, configured to acquire the time-series images, mechanical response data, and culture environment parameters of the target cells;

[0119] A segmentation unit, configured to perform region segmentation on the time-series images based on the mechanical response data and the culture environment parameters to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of the subcellular structures;

[0120] A generation unit, configured to generate mechanical phenotype descriptors of the target cells according to the mechanically sensitive regions, the local deformation characteristics, and the culture environment parameters;

[0121] A result unit, configured to process the mechanical phenotype descriptors using a mechanical phenotype analysis model to obtain a heterogeneity prediction result of the target cells; wherein, the mechanical phenotype analysis model is a machine learning model pre-trained using the sample mechanical phenotype descriptors of the sample cells.

[0122] It should be noted that the information interaction, execution process, etc. between the above systems / units, due to being based on the same concept as the method embodiment of the present application, for their specific functions and the technical effects brought, refer to the method embodiment part for details, and will not be elaborated here.

[0123] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated in a processing unit, or each unit module can exist physically alone, or two or more unit modules can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiments and will not be repeated here.

[0124] The embodiment of this application also provides an electronic device, Figure 4 which is a schematic structural diagram of the electronic device provided in an embodiment of this application. As Figure 4 shown, the electronic device 6 in this embodiment includes: at least one processor 60 ( Figure 4 only one is shown in the figure), at least one memory 61 ( Figure 4 only one is shown in the figure), and a computer program 62 stored in the at least one memory 61 and executable on the at least one processor 60. When the processor 60 executes the computer program 62, the electronic device 6 implements the steps in any of the above-mentioned embodiments of the cell mechanics phenotype analysis method based on deep learning, or enables the electronic device 6 to implement the functions of each unit in the above-mentioned system embodiments.

[0125] Exemplarily, the computer program 62 can be divided into one or more units. The one or more units are stored in the memory 61 and executed by the processor 60 to complete this application. The one or more units can be a series of computer program instruction segments capable of completing specific functions, and these instruction segments are used to describe the execution process of the computer program 62 in the electronic device 6.

[0126] The electronic device 6 can be a computing device or a terminal device such as a desktop computer, a notebook, a palm computer, and a cloud server. The electronic device may include, but is not limited to, a processor 60 and a memory 61. Those skilled in the art can understand that Figure 4 this is only an example of the electronic device 6 and does not constitute a limitation on the electronic device 6. It may include more or fewer components than shown in the figure, or combine some components, or different components. For example, it may also include input and output devices, network access devices, buses, etc.

[0127] The processor 60 may be a Central Processing Unit (CPU), and the processor 60 may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0128] In some embodiments, the memory 61 may be an internal storage unit of the electronic device 6, such as a hard disk or memory of the electronic device 6. In other embodiments, the memory 61 may also be an external storage device of the electronic device 6, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc., equipped on the electronic device 6. Further, the memory 61 may also include both the internal storage unit and the external storage device of the electronic device 6. The memory 61 is used to store an operating system, application programs, a BootLoader, data, and other programs, such as the program code of the computer program. The memory 61 may also be used to temporarily store data that has been output or is to be output.

[0129] The embodiment of the present application also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0130] The embodiment of the present application provides a computer program product, and when the computer program product runs on an electronic device, the electronic device implements the steps in any of the above method embodiments.

[0131] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above-described embodiment methods of this application, a computer program can be used to instruct relevant hardware to complete. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-described method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can at least include: any entity or device that can carry the computer program code to an electronic device, a recording medium, a computer memory, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), an electrical carrier signal, a telecommunication signal, and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk, or an optical disc, etc. In some jurisdictions, according to legislation and patent practice, the computer-readable medium cannot be an electrical carrier signal and a telecommunication signal.

[0132] In the above embodiments, the descriptions of the various embodiments have their own emphases. For parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0133] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A professional person can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0134] In the embodiments provided in this application, it should be understood that the disclosed method and device for cell mechanics phenotype analysis based on deep learning can be implemented in other ways. For example, the above-described method and device embodiments for cell mechanics phenotype analysis based on deep learning are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical, mechanical, or other form.

[0135] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or may be distributed over multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0136] The foregoing embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included within the protection scope of the present application.

Claims

1. A method for cell mechanical phenotype analysis based on deep learning, characterized in that, Including: Obtaining time-series images, mechanical response data, and culture environment parameters of target cells; Based on the mechanical response data and culture environment parameters, performing region segmentation on the time-series images to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of subcellular structures; Generating a mechanical phenotype descriptor of the target cells according to the mechanically sensitive regions, the local deformation characteristics, and the culture environment parameters; Processing the mechanical phenotype descriptor using a mechanical phenotype analysis model to obtain a heterogeneity prediction result of the target cells; wherein, the mechanical phenotype analysis model is a machine learning model pre-trained using sample mechanical phenotype descriptors of sample cells.

2. The method for analyzing cell mechanical phenotypes based on deep learning according to claim 1, wherein The performing region segmentation on the time-series images based on the mechanical response data and culture environment parameters to determine the mechanically sensitive regions on the surface of the target cells and the local deformation characteristics of subcellular structures includes: Performing time-varying spatial registration processing on the time-series images to generate a time-varying spatial correlation relationship between the cell deformation field and the movement trajectories of subcellular structures of the target cells; Based on the time-varying spatial correlation relationship and the mechanical response data, calculating the strain gradient tensor on the surface of the target cells to obtain an initial segmentation boundary of the mechanically sensitive region; Optimizing the initial segmentation boundary according to the strain gradient tensor and the culture environment parameters to determine the mechanically sensitive regions on the surface of the target cells; Extracting the movement trajectories of subcellular structures from the time-varying spatial correlation relationship, calculating the curvature change parameters of the subcellular structures, and determining the local deformation characteristics of the subcellular structures.

3. The method for analyzing cell mechanical phenotypes based on deep learning according to claim 2, wherein The performing time-varying spatial registration processing on the time-series images to generate a time-varying spatial correlation relationship between the cell deformation field and the movement trajectories of subcellular structures of the target cells includes: Performing displacement field calculation and feature matching on the time-series images to determine the cell deformation field and the movement trajectories of subcellular structures of the target cells; Aligning the coordinates of the cell deformation field and the movement trajectories of the subcellular structures to establish a time-varying spatial mapping relationship between the cell deformation field and the subcellular structures; Performing multi-scale strain analysis on the mapping relationship to generate a time-varying spatial correlation relationship between the cell deformation field and the movement trajectories of subcellular structures of the target cells.

4. The method for analyzing cell mechanical phenotypes based on deep learning according to claim 3, wherein The performing displacement field calculation and feature matching on the time-series images to determine the cell deformation field and the movement trajectories of subcellular structures of the target cells includes: Performing pixel-by-pixel displacement calculation on adjacent time-series images to generate an initial displacement field of the target cells; Identifying subcellular structure feature points from the time-series images and determining a sequence of movement trajectory coordinates of the subcellular structures according to the subcellular structure feature points; Performing displacement correction based on the initial displacement field and the sequence of movement trajectory coordinates of the subcellular structures to determine the cell deformation field and the movement trajectories of subcellular structures of the target cells.

5. The method for analyzing cell mechanical phenotypes based on deep learning according to claim 2, wherein The calculating the strain gradient tensor on the surface of the target cells based on the time-varying spatial correlation relationship and the mechanical response data to obtain an initial segmentation boundary of the mechanically sensitive region includes: Based on the time-varying spatial correlation relationship, obtaining the strain tensor components on the surface of the target cells; Perform a spatial derivative operation on the strain tensor components on the surface of the target cell to generate a strain gradient tensor field; Based on the strain gradient tensor field and the mechanical response data, determine the segmentation threshold of the mechanically sensitive region; Divide the surface region of the target cell according to the segmentation threshold to obtain an initial segmentation boundary of the mechanically sensitive region.

6. The method for cell mechanical phenotype analysis based on deep learning according to claim 5, wherein The optimizing the initial segmentation boundary according to the strain gradient tensor and the culture environment parameters to determine the mechanically sensitive region on the surface of the target cell includes: Dynamically adjust the segmentation threshold of the initial segmentation boundary based on the culture environment parameters to generate an optimized threshold for the mechanically sensitive region; Perform a morphological closing operation on the initial segmentation boundary according to the optimized threshold to generate an optimized boundary of the mechanically sensitive region; Perform boundary fusion on the optimized boundary according to the strain gradient tensor to determine the mechanically sensitive region on the surface of the target cell.

7. The method for cell mechanical phenotype analysis based on deep learning according to claim 2, wherein, The extracting the subcellular structure motion trajectory from the time-varying spatial correlation relationship, calculating the curvature change parameter of the subcellular structure, and determining the local deformation characteristics of the subcellular structure includes: Extract the subcellular structure motion trajectory from the time-varying spatial correlation relationship to determine the motion trajectory coordinate sequence of the subcellular structure of the target cell; Perform spline interpolation on the motion trajectory coordinate sequence to generate a motion trajectory curve of the subcellular structure, and calculate the trajectory curvature parameter of the subcellular structure according to the motion trajectory curve; Perform a time derivative operation according to the trajectory curvature parameter of the subcellular structure to determine the local deformation characteristics of the subcellular structure.

8. The method for analyzing cell mechanical phenotypes based on deep learning according to claim 1, wherein The generating the mechanical phenotype descriptor of the target cell according to the mechanically sensitive region, the local deformation characteristics, and the culture environment parameters includes: Perform feature association according to the mechanically sensitive region, the local deformation characteristics, and the culture environment parameters to generate a heterogeneity association matrix of the target cell; Perform feature dimensionality reduction on the heterogeneity association matrix, and perform time-frequency joint feature encoding on the dimensionality-reduced heterogeneity association matrix to generate the mechanical phenotype descriptor of the target cell.

9. The method for cell mechanical phenotype analysis based on deep learning according to claim 1, wherein Before using the mechanical phenotype analysis model to process the mechanical phenotype descriptor to obtain the heterogeneity prediction result of the target cell, the method further includes: Obtain the mechanical phenotype descriptor of the sample cell; Determine the heterogeneity analysis label of the sample cell; Use the mechanical phenotype descriptor of the sample cell as the input and the corresponding heterogeneity analysis label as the expected output to train the initial machine learning model to obtain the trained mechanical phenotype analysis model.

10. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, the method described in any one of claims 1 to 9 is implemented.

Images