A method for analyzing drug action based on three-dimensional roughness characterization of tumor organoids

Abstract

The application discloses a drug action analysis method based on tumor organoid three-dimensional roughness characterization, and relates to the technical field of tumor organoids.The original three-dimensional structure data of the tumor organoid is acquired by three-dimensional interference without fluorescent reagent marking, then the connected domain data corresponding to each organoid in the target organoid cluster is determined by segmentation to determine the morphological parameters of each organoid, wherein the morphological parameters of each organoid including at least one of solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis and sphericity are determined by internal hole filling, and the surface of each organoid after internal hole filling is subjected to smoothing processing, then the volume change caused by the filled volume after smoothing processing is determined to determine the morphological parameters of the roughness corresponding to each organoid, the roughness can greatly quantify the heterogeneity of the organoid, and the accuracy of the drug action of the tumor organoid is improved.

Description

Technical Field

[0001] This invention relates to the field of tumor organoid technology, and in particular to a method for analyzing drug effects based on three-dimensional roughness characterization of tumor organoids. Background Technology

[0002] Generally, tumor heterogeneity has a significant impact on drug response. Existing research on tumor organoid heterogeneity mainly focuses on the cytogenetics or cellular composition level, lacking analysis of tumor morphological diversity. Analysis of tumor organoid morphological diversity may be limited to capturing three-dimensional spatial structural information of tumor organoids or simulating tumor morphological change models.

[0003] Commonly used bright-field microscopy lacks the ability to image the three-dimensional structure of tumor organoids, ignoring the complexity of three-dimensional spatial relationships, leading to limitations in morphological analysis based on two-dimensional images. Three-dimensional simulation can obtain the evolution of three-dimensional models of tumor organoids based on information such as cell division and perform theoretical analysis, but it cannot analyze the actual morphology of tumor organoids. Three-dimensional imaging is a necessary condition for revealing the complex structure and morphological phenotype of tumor organoids and confirming that tumor organoids can realistically reflect their corresponding organs in vivo. Among them, confocal imaging or multiphoton microscopy based on fluorescent reagent labeling can achieve visualization of three-dimensional details of tumor organoids and is widely used in various biomedical research, including organoids. However, the permeability of fluorescent reagents is limited, making it difficult to obtain the internal morphology of dense three-dimensional tissues larger than 50 micrometers. The tissue penetration varies greatly, and the fluorescent reagent labeling detection process is damaging to cells, complex to operate, and difficult to conveniently and quickly obtain accurate morphological phenotypes of tumor organoids for application in drug action analysis.

[0004] Therefore, label-free three-dimensional structural imaging and morphological analysis of tumor organoids are crucial for exploring the relationship between the dynamic evolution of tumor organoid morphology and drug action. However, existing morphological characterizations of tumor organoids typically include morphological parameters such as volume and surface area. These parameters are weak in quantifying the heterogeneity of tumor organoids, resulting in poor targeting of drug action analysis based on existing tumor organoid morphological analysis, and impaired accuracy in analyzing the correlation between the dynamic evolution of tumor organoid morphology and drug action. Summary of the Invention

[0005] Therefore, it is necessary to provide a drug action analysis method based on the three-dimensional roughness characterization of tumor organoids to address the above-mentioned technical problems.

[0006] The present invention adopts the following technical solution:

[0007] This invention provides a method for drug action analysis based on the three-dimensional roughness characterization of tumor organoids. First, the invention obtains the original three-dimensional structural data of multiple tumor organoids using three-dimensional interference spectral data. Then, organoid segmentation and connectivity analysis are performed on the original three-dimensional structural data to determine the connected component data corresponding to each organoid within the target organoid cluster. Next, based on the connected component data corresponding to each organoid within the target organoid cluster and internal cavity filling, morphological parameters of each organoid are determined, including at least one of the following: solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis, and sphericity. The surface of each organoid after internal cavity filling is smoothed. Based on the volume change caused by the smoothing process relative to the filled volume, the morphological parameters of the roughness corresponding to each organoid are determined. Finally, based on the morphological parameters of each organoid, morphological phenotypic classification of various organs within the target organoid cluster is performed, and a growth level model of various organs within the target organoid cluster is constructed for drug action analysis.

[0008] This invention provides a drug action analysis device based on three-dimensional roughness characterization of tumor organoids, comprising:

[0009] The acquisition module is used to acquire the original three-dimensional structural data of multiple tumor organoids based on the three-dimensional interference spectral data of multiple tumor organoids.

[0010] The segmentation module is used to perform organoid segmentation and connectivity analysis on the original three-dimensional structural data to determine the connected component data corresponding to each organoid in the target organoid cluster among multiple tumor organoids.

[0011] The filling module is used to fill internal holes based on the connected component data corresponding to each organoid in the target organoid cluster, and to determine the morphological parameters of each organoid, including solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis and sphericity information.

[0012] The smoothing module is used to smooth the surface of each organoid after the internal pores are filled. Based on the volume change caused by the smoothing process relative to the volume after filling, the morphological parameters of the roughness corresponding to each organoid are determined.

[0013] The analysis module is used to classify the morphological phenotypes of various organs within the target organoid cluster based on the morphological parameters of each organoid, and to construct growth level models of various organoids within the target organoid cluster for drug action analysis.

[0014] The present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method for drug action analysis based on three-dimensional roughness characterization of tumor organoids.

[0015] The present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the above-mentioned drug action analysis method based on the three-dimensional roughness characterization of tumor organoids.

[0016] The above-mentioned at least one technical solution adopted in this invention can achieve the following beneficial effects:

[0017] This invention first acquires the original three-dimensional structural data of tumor organoids through three-dimensional interferometry without fluorescent reagent labeling. Then, it segments and determines the connected component data corresponding to each organoid within the target organoid cluster, thereby determining the morphological parameters of each organoid. Specifically, this invention determines the morphological parameters of each organoid by filling internal pores, including at least one of the following: solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis, and sphericity. After filling the internal pores, the surface of each organoid is smoothed. Based on the volume change caused by the smoothing process relative to the filled volume, the morphological parameter of the roughness corresponding to each organoid is determined. The roughness obtained in this way can greatly quantify the heterogeneity of tumor organoids. Subsequent morphological phenotypic classification and drug effect analysis based on morphological parameters enhance the targeting of drug effect analysis of tumor organoids and improve the accuracy of the correlation analysis between the dynamic evolution of tumor organoid morphology and drug effects. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0019] Figure 1 A schematic flowchart of a drug action analysis method based on three-dimensional roughness characterization of tumor organoids provided by the present invention;

[0020] Figure 2 A schematic diagram illustrating the determination of organoid roughness morphological parameters provided by the present invention;

[0021] Figure 3 This invention provides a simplified schematic diagram for calculating the roughness of a three-dimensional body.

[0022] Figure 4 This invention provides a schematic diagram for calculating the roughness of a burr ball.

[0023] Figure 5 This invention provides a schematic diagram of organoid roughness calculation;

[0024] Figure 6 This invention provides another schematic diagram for calculating the roughness of organoids;

[0025] Figure 7 This is a schematic diagram of a drug action analysis device based on three-dimensional roughness characterization of tumor organoids, provided by the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0027] Currently, the visualization of three-dimensional details of organoids achieved by confocal imaging or multiphoton microscopy based on fluorescent reagent labeling is insufficient to obtain accurate morphological phenotypes of tumor organoids for application in drug action analysis.

[0028] Therefore, there is an urgent need for a method that can perform high-throughput, longitudinal, label-free, three-dimensional imaging of organoids in a short time, and then use the morphological phenotype of tumor organoids to describe the growth level, drug effects, or invasiveness of tumor organoids.

[0029] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0030] Figure 1 This is a schematic diagram of a drug action analysis method based on three-dimensional roughness characterization of tumor organoids according to the present invention, which specifically includes the following steps:

[0031] S101: Obtain the original three-dimensional structural data of multiple tumor organoids based on the three-dimensional interference spectral data of multiple tumor organoids.

[0032] S102: Perform organoid segmentation and connectivity analysis on the original three-dimensional structural data to determine the connected component data corresponding to each organoid in the target organoid cluster among multiple tumor organoids.

[0033] S103: Based on the connected component data corresponding to each organoid in the target organoid cluster and the filling of internal pores, determine the morphological parameters of each organoid, including at least one of the following: solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis, and sphericity.

[0034] S104: Smooth the surface of each organoid after filling the internal pores, and determine the morphological parameters of the roughness corresponding to each organoid based on the volume change caused by the smoothing process relative to the volume after filling.

[0035] S105: Based on the morphological parameters of each organoid, classify the morphological phenotypes of various organs within the target organoid cluster, and construct growth level models of various organoids within the target organoid cluster for drug action analysis.

[0036] For ease of explanation, the following description focuses solely on the server as the executing entity. The server mentioned in this invention can be a server set up on a business platform, or a device such as a desktop computer or laptop computer capable of executing the solution of this invention.

[0037] Generally, when performing drug action analysis based on the morphological phenotype of tumor organoids, the server can first acquire three-dimensional interference spectral data of multiple tumor organoids under normal culture conditions using three-dimensional optical imaging technology. .

[0038] Then, the three-dimensional interferometric spectral data Wavenumber resampling, DC removal, and Fourier transform were performed to convert the frequency domain to the depth domain, thereby obtaining the original three-dimensional structural data of multiple tumor organoids. .

[0039] Then, the original three-dimensional structural data Perform organoid segmentation to obtain binary data of the target organoid cluster contained in multiple tumor organoids. .

[0040] Thus, the binarized data containing the target organoid clusters Perform connectivity analysis to obtain the number of individual organoids within the target organoid cluster. N and the connected component data corresponding to each organoid within the target organoid cluster. .

[0041] Next, we can analyze each organoid. Morphological analysis was performed to obtain multidimensional morphological parameters and determine the morphological phenotype of each organoid.

[0042] Based on this, various organs can be phenotypically classified, such as by k-means clustering analysis based on machine learning. This allows for statistical analysis of each organ class by category, and the construction of linear models characterizing the growth levels of various organ classes within the target organ class cluster. Finally, the constructed linear models can be used to evaluate the changes in growth levels of various organ classes within the target organ class cluster at different times and the effects of drugs.

[0043] In one or more embodiments of the present invention, the server performs processing on each organoid... When performing morphological analysis to obtain its multidimensional morphological parameters, one can first analyze those containing only a single organoid. Filling internal holes in 3D images Therefore, individual organoids can be calculated based on the three-dimensional images after the internal pores are filled. The solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis, and sphericity information.

[0044] Regarding the roughness of each organoid, refer to Figure 2 , Figure 2 This is a schematic diagram illustrating how to determine the roughness morphological parameters of organoids in this invention.

[0045] The server can first fill the internal cavities of various organs Perform three-dimensional Fourier transform and spectral centering to obtain the frequency spectra of various organs. And based on the frequency spectrum of various organs Determine the raw power spectrum of each organoid's 3D image. .

[0046] Then, by using different cutoff frequencies in the low-pass frequency domain based on the power spectrum ratio, the surface of each organoid after the internal pores are filled is smoothed to obtain the filtered spectrum and determine the power spectrum retained after filtering.

[0047] Specifically, after the server fills the internal holes, a single organoid... When performing smoothing, a low-pass frequency domain filter based on the power spectral density can be selected, such as a cutoff frequency of 1000 MHz. Butterworth low-pass filter: .

[0048] The server can select different cutoff frequencies. and using the corresponding filter and Multiply the results to obtain the filtered spectrum. And calculate the power spectrum retained by the filtered image. .

[0049] Then the server can determine the power spectrum ratio. (This is just an example; other values ​​can be determined as needed.) Determine the preferred cutoff frequency. This ensures smoothness between different organoids. And it optimizes the cutoff frequency. The corresponding smoothed spectrum Perform Fourier transform and binarization to determine the smoothed data of various organs under the corresponding power spectrum proportions. .

[0050] Finally, the server can fill and smooth the data inside various organs after filling the pores. Smoothing data for various organs XOR processing is performed, and the XOR result represents the smoothed volume changes of various organs. It can be based on the volume changes of various organs after smoothing. Surface area before smoothing after filling internal pores of various organs ratio The morphological parameters of the roughness corresponding to each organoid are determined. The above roughness determination process can obtain accurate roughness of tumor organoids without labeling.

[0051] Figure 3 This is a schematic diagram illustrating a simple three-dimensional volume roughness calculation method according to the present invention. Figure 3 The left side shows the initial simple three-dimensional solid with a surface area of ​​0.099 mm². 2 When the optimal power spectrum ratio is selected as 90%, the following can be obtained. Figure 3 The volume change of the intermediate filtered result is 0.00000875 mm. 3 The roughness calculation result is 0.000088 mm; the optimal power spectrum ratio can be obtained when 85% is selected. Figure 3 The filtered result on the right shows a volume change of 0.00003900 mm. 3 The roughness calculation result is 0.000394 mm.

[0052] Figure 4 This is a schematic diagram illustrating the roughness calculation of a burr ball according to the present invention. Figure 4 The initial burr ball on the left has a radius of 100. The optimal power spectrum ratio is 85%. Figure 4 The filtered result on the right shows a volume change of 0.01460750 mm. 3 The roughness calculation result is 0.004875mm.

[0053] Figure 5 This is a schematic diagram illustrating the roughness calculation of an organoid in this invention. Figure 5 The left side of the image shows the initial organoids obtained after segmentation, which can be obtained when the power spectrum ratio is 90%. Figure 5 The intermediate organoid filtering results show a volume change of 0.00114262 mm before and after filtering. 3 The roughness calculation result is 0.001423 mm; this can be obtained when the power spectrum ratio is 85%. Figure 5 The organoid filtering result on the right shows a volume change of 0.00290750 mm before and after filtering. 3 The surface roughness is 0.003621 mm.

[0054] Figure 6 This is a schematic diagram illustrating another method for calculating organoid roughness in this invention. Figure 6The left side of the image shows another initial organoid obtained after segmentation, which can be obtained when the power spectrum ratio is 90%. Figure 6 The intermediate organoid filtering results show a volume change of 0.00048875 mm before and after filtering. 3 The roughness calculation result is 0.001361 mm; this can be obtained when the power spectrum ratio is 85%. Figure 5 The organoid filtering result on the right shows a volume change of 0.00068650 mm before and after filtering. 3 The surface roughness is 0.001912 mm.

[0055] Furthermore, in one or more embodiments of the present invention, when the server performs phenotypic classification on various organs within the target organ cluster, it may employ clustering analysis methods including but not limited to k-means clustering and DBSCAN clustering.

[0056] Taking the improved K-means clustering as an example, the server can first perform data normalization and correlation analysis on the multidimensional morphological parameters, and then determine and remove redundant features in the morphological parameters based on the correlation threshold. Specifically, the server can perform data normalization and correlation analysis on the solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis, sphericity, and roughness parameters obtained in the previous steps, and remove redundant features with a correlation ≥ 95%.

[0057] Then, the server can perform balancing processing on the data of each organoid within the target organoid cluster, and based on the preferred number of clusters and preferred organoids as the initial cluster centers, perform cluster analysis using K-means clustering according to the morphological parameters of each organoid to determine the classification of each organoid within the target organoid cluster.

[0058] The optimal number of clusters and cluster centers can be determined by analyzing organoid morphological characteristics based on existing data. P And to select representative organoids from each type. As the initial cluster center.

[0059] The balance handling is mainly achieved through two methods. Sample imbalance handling 1: When the proportion of organoids smaller than the size threshold within the target organoid cluster is greater than the first threshold, downsampling is performed on the organoids in that cluster. That is, considering that organoids in the organoid cluster have different sizes, and the proportion of small-sized organoids is significantly larger, downsampling is performed on the small-sized organoids.

[0060] Imbalanced sample handling 2: When the proportion of organoids of any category within the target organoid cluster is less than the second threshold and their importance is greater than the importance threshold, organoids of that category are classified separately. That is, considering that organoids in the organoid cluster have different sizes, if the proportion of organoids of a certain category is extremely small but their existence is extremely important, such as large-sized hollow organoids, then organoids of that category need to be manually classified into one category.

[0061] After addressing the imbalanced sample size, K-means clustering was performed on the organoids to obtain a clustering model with P mean vectors for each class. This clustering model can then be used to perform cluster analysis on newly collected organoid data to obtain the category information of individual organoids.

[0062] Furthermore, in one or more embodiments of the present invention, when the server constructs a growth level model of various organoids within the target organoid cluster, the server can statistically analyze the mean, median, and standard deviation of different morphological parameters of each type of organoid within the target organoid cluster, thereby standardizing the morphological parameters of that type of organoid. Specifically, based on the organoid data of each category within the target organoid cluster, the server can obtain the number of organoids in each category, as well as information such as the mean, median, and standard deviation of various morphological parameters such as solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis, sphericity, and roughness.

[0063] The server can then perform correlation analysis on the number of organoids and the characteristics of different morphological parameters, and determine and remove redundant features of each organoid based on the correlation threshold. For example, redundant features with a correlation ≥ 95% can be removed.

[0064] Then, based on the correlation coefficient matrix of the residual features of various organoids, eigenvalue decomposition is performed to obtain multiple eigenvalues ​​and corresponding eigenvectors for principal component analysis. The eigenvectors corresponding to the eigenvalues ​​that reach the preset cumulative contribution rate are determined as principal components. Typically, the eigenvectors corresponding to the p largest eigenvalues ​​are selected as principal components based on the magnitude of the eigenvalues. Generally, eigenvalues ​​with a cumulative contribution rate reaching a certain proportion (e.g., 85%) can be selected.

[0065] Finally, using the selected feature vectors, the scores of each organoid group on these principal components can be calculated. These scores constitute a new dataset after dimensionality reduction. This allows us to determine the score of each organoid group on each principal component. By weighting and summing the scores based on the contribution rate of each principal component, the growth level of each organoid group can be determined.

[0066] Correlation analysis reveals certain correlations among multiple morphological parameters of different organoid communities. Therefore, principal component analysis can be applied to reduce the dimensionality of multiple categories of morphological parameters to simplify the data structure and improve the accuracy and interpretability of the comprehensive evaluation of organoid community growth levels.

[0067] Finally, the drug effects of tumor organoids can be analyzed based on the growth levels under different drug conditions or drug concentrations. There are already well-established techniques for analyzing the drug effects of tumor organoids based on growth levels, which will not be elaborated upon in this invention.

[0068] based on Figure 1 The method for drug action analysis based on the three-dimensional roughness characterization of tumor organoids, as shown in this invention, first acquires the original three-dimensional structural data of tumor organoids through three-dimensional interferometry without fluorescent reagent labeling. Then, it segments and determines the connected component data corresponding to each organoid within the target organoid cluster, thereby determining the morphological parameters of each organoid. Specifically, this invention determines the morphological parameters of each organoid by filling internal pores, including at least one of the following: solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis, and sphericity. After filling the internal pores, the surface of each organoid is smoothed. Based on the volume change caused by the smoothing process relative to the filled volume, the morphological parameters of the roughness corresponding to each organoid are determined. The roughness obtained in this way can greatly quantify the heterogeneity of organoids. Subsequent morphological phenotypic classification and drug action analysis based on morphological parameters enhance the targeting of drug action analysis of tumor organoids and improve the accuracy of the correlation analysis between the dynamic evolution of tumor organoid morphology and drug action.

[0069] When applying the drug action analysis method based on three-dimensional roughness characterization of tumor organoids provided by this invention, it is not necessary to consider... Figure 1 The steps shown are executed in sequence. The specific execution order of each step can be determined as needed, and this invention does not impose any restrictions on it.

[0070] The above describes a drug action analysis method based on three-dimensional roughness characterization of tumor organoids, provided by one or more embodiments of the present invention. Based on the same idea, the present invention also provides a corresponding drug action analysis device based on three-dimensional roughness characterization of tumor organoids, such as... Figure 7 As shown.

[0071] Figure 7 A schematic diagram of a drug action analysis device based on three-dimensional roughness characterization of tumor organoids provided by the present invention includes:

[0072] The acquisition module 201 is used to acquire the original three-dimensional structural data of multiple tumor organoids based on the three-dimensional interference spectral data of multiple tumor organoids.

[0073] The segmentation module 202 is used to perform organoid segmentation and connectivity analysis on the original three-dimensional structural data to determine the connected component data corresponding to each organoid in the target organoid cluster among multiple tumor organoids.

[0074] The filling module 203 is used to determine the morphological parameters of each organoid based on the internal hole filling according to the connected component data corresponding to each organoid in the target organoid cluster, including at least one of the following: solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis and sphericity information.

[0075] The smoothing module 204 is used to smooth the surface of each organoid after the internal pores are filled. Based on the volume change caused by the smoothing process relative to the volume after filling, the morphological parameters of the roughness of each organoid are determined.

[0076] Analysis module 205 is used to classify the morphological phenotypes of various organs within the target organoid cluster based on the morphological parameters of each organoid, and to construct growth level models of various organs within the target organoid cluster for drug action analysis.

[0077] Specific limitations regarding the drug action analysis device based on the three-dimensional roughness characterization of tumor organoids can be found in the limitations of the drug action analysis method based on the three-dimensional roughness characterization of tumor organoids mentioned above, and will not be repeated here. Each module in the aforementioned drug action analysis device based on the three-dimensional roughness characterization of tumor organoids can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0078] The present invention also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described... Figure 1 A method for analyzing drug effects based on three-dimensional roughness characterization of tumor organoids is provided.

[0079] This invention also provides a computer device. At the hardware level, the computer device includes a processor, an internal bus, a network interface, memory, and non-volatile memory, and may also include other hardware required for various operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then executes it to achieve the above-mentioned functions. Figure 1 A method for analyzing drug effects based on three-dimensional roughness characterization of tumor organoids is provided.

[0080] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0081] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this invention.

Claims

1. A method for analyzing drug effects based on three-dimensional roughness characterization of tumor organoids, characterized in that, include: Based on the three-dimensional interferometric spectral data of multiple tumor organoids, the original three-dimensional structural data of multiple tumor organoids were obtained; Organoid segmentation and connectivity analysis were performed on the original three-dimensional structural data to determine the connected component data corresponding to each organoid in the target organoid cluster among multiple tumor organoids. Based on the connected component data corresponding to each organoid in the target organoid cluster and the filling of internal pores, determine the morphological parameters of each organoid, including at least one of solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis and sphericity. Three-dimensional Fourier transform and spectral centering are performed on various organs after the internal pores are filled to obtain the frequency spectrum of each organ to determine the original power spectrum of each organ class; by using low-pass frequency domain filtering based on the power spectrum ratio and different cutoff frequencies, the surface of each organ class after the internal pores are filled is smoothed to obtain the filtered spectrum and determine the power spectrum retained after filtering. The preferred cutoff frequency is determined based on the power spectrum ratio. Fourier transform and binarization are performed on the smoothed spectrum corresponding to the preferred cutoff frequency to determine the smoothed data of various organs under the corresponding power spectrum ratio. The data before smoothing after filling the internal pores of various organs are XORed with the smoothed data of various organs to determine the volume change of various organs after smoothing. Based on the ratio of the volume change of various organs after smoothing to the surface area before smoothing after filling the internal pores of various organs, the morphological parameters of the roughness corresponding to each type of organ are determined. Based on the morphological parameters of each organoid, morphological phenotypic classification is performed on various organs within the target organoid cluster. For each organoid class within the target organoid cluster, the number of organoids in that class and the mean, median, and standard deviation of different morphological parameters of that class are counted, and the morphological parameters of that class are standardized. Correlation analysis is performed on the number of organoids in each class and the characteristics of different morphological parameters. Redundant features of each organoid class are identified and removed based on the correlation threshold. Eigenvalue decomposition is performed based on the correlation coefficient matrix of the remaining features of each organoid class to obtain multiple eigenvalues ​​and corresponding eigenvectors for principal component analysis. The eigenvectors corresponding to the eigenvalues ​​that reach the preset cumulative contribution rate are determined as principal components. The scores of each organoid on each principal component are determined, and the corresponding scores are weighted and summed according to the contribution rate of each principal component to determine the growth level of each organoid for drug effect analysis.

2. The method for drug action analysis based on three-dimensional roughness characterization of tumor organoids as described in claim 1, characterized in that, The step of obtaining the original three-dimensional structural data of multiple tumor organoids based on the three-dimensional interference spectral data of multiple tumor organoids specifically includes: Three-dimensional interferometric spectral data of multiple tumor organoids were obtained based on three-dimensional optical imaging technology; The three-dimensional interferometric spectral data were converted from the frequency domain to the depth domain to obtain the original three-dimensional structural data of multiple tumor organoids.

3. The method for drug action analysis based on three-dimensional roughness characterization of tumor organoids as described in claim 1, characterized in that, The process of segmenting and analyzing the original three-dimensional structural data to determine the connected component data corresponding to each organoid within the target organoid cluster in multiple tumor organoids specifically includes: Organoid segmentation is performed on the original three-dimensional structural data to obtain binarized data of target organoid clusters contained in multiple tumor organoids; Connectivity analysis is performed on the binarized data containing the target organoid cluster to obtain the connected component data corresponding to each organoid within the target organoid cluster.

4. The method for drug action analysis based on three-dimensional roughness characterization of tumor organoids as described in claim 1, characterized in that, The morphological phenotypic classification based on the morphological parameters of each organoid within the target organoid cluster specifically includes: The morphological parameters of each organoid within the target organoid cluster are normalized and subjected to correlation analysis. Redundant features in the morphological parameters are determined and removed based on the correlation threshold. The data of each organoid within the target organoid cluster are balanced, and based on the selected number of clusters and the selected organoids as the initial cluster centers, K-means clustering is used to perform cluster analysis based on the morphological parameters of each organoid to determine the classification of each organoid within the target organoid cluster.

5. The drug action analysis method based on three-dimensional roughness characterization of tumor organoids as described in claim 4, characterized in that, The balancing process for the data of each organoid within the target organoid cluster specifically includes: When the proportion of organoids smaller than the size threshold within the target organoid cluster is greater than the first threshold, downsampling is performed on the organoids belonging to that cluster. When the proportion of organoids of any category within the target organoid cluster is less than the second threshold and the importance is greater than the importance threshold, organoids of that category are classified separately or oversampled.

6. A drug action analysis device based on three-dimensional roughness characterization of tumor organoids, characterized in that, include: The acquisition module is used to acquire the original three-dimensional structural data of multiple tumor organoids based on the three-dimensional interference spectral data of multiple tumor organoids. The segmentation module is used to perform organoid segmentation and connectivity analysis on the original three-dimensional structural data to determine the connected component data corresponding to each organoid in the target organoid cluster among multiple tumor organoids. The filling module is used to fill internal holes based on the connected component data corresponding to each organoid in the target organoid cluster, and to determine the morphological parameters of each organoid, including solid volume, cavity volume, filled volume, filled surface area, longest axis, shortest axis and sphericity information. The smoothing module is used to perform three-dimensional Fourier transform and spectrum centering on various organs after the internal pores are filled, to obtain the frequency spectrum of each organ and determine the original power spectrum of each organoid; by using different cutoff frequencies through low-pass frequency domain filtering based on the power spectrum ratio, the surface of each organoid after the internal pores are filled is smoothed to obtain the filtered spectrum and determine the power spectrum retained after filtering. The preferred cutoff frequency is determined based on the power spectrum ratio. Fourier transform and binarization are performed on the smoothed spectrum corresponding to the preferred cutoff frequency to determine the smoothed data of various organs under the corresponding power spectrum ratio. The data before smoothing after filling the internal pores of various organs are XORed with the smoothed data of various organs to determine the volume change of various organs after smoothing. Based on the ratio of the volume change of various organs after smoothing to the surface area before smoothing after filling the internal pores of various organs, the morphological parameters of the roughness corresponding to each type of organ are determined. The analysis module is used to classify the morphological phenotypes of various organs within the target organoid cluster based on the morphological parameters of each organoid. For each organoid class within the target organoid cluster, it counts the number of organoids of that class and the mean, median, and standard deviation of different morphological parameters of that class of organoids, and standardizes the morphological parameters of that class of organoids. It performs correlation analysis on the number of organoids of each class and the characteristics of different morphological parameters, determines redundant features of each organoid class based on the correlation threshold, and removes them. Based on the correlation coefficient matrix of the remaining features of each organoid class, it performs eigenvalue decomposition to obtain multiple eigenvalues ​​and corresponding eigenvectors for principal component analysis, and determines the eigenvectors corresponding to the eigenvalues ​​that reach the preset cumulative contribution rate as the principal components. The scores of each organoid on each principal component are determined, and the corresponding scores are weighted and summed according to the contribution rate of each principal component to determine the growth level of each organoid for drug effect analysis.

7. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method as described in any one of claims 1 to 5.

8. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of claims 1 to 5.

Images