Image processing method for organoid recognition and processing system thereof

Abstract

The invention provides an image processing method and system for organoid recognition, and relates to the technical field of image processing, and the method comprises the steps: collecting an organoid culture image set; for each organoid culture image in the organoid culture image set, performing spot feature extraction on the organoid culture images to obtain image spot features of the organoid culture images, determining structural heterogeneous entropies of the organoid culture images according to the image spot features, and screening the organoid culture image set based on all the structural heterogeneous entropies to obtain a screened organoid culture image set; obtaining an organoid culture image subset; determining a pixel fluctuation peak of each organoid culture image in the organoid culture image subset, and inhibiting hybrid information in the organoid culture images according to the corresponding pixel fluctuation peak to obtain an image cluster after the hybrid information is inhibited; and applying the image cluster subjected to hybrid information suppression to organoid recognition, so that hybrid information in the organoid culture image can be suppressed, and the recognition effect of organoid recognition is improved.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and more specifically, to an image processing method and a processing system thereof for organoid identification. Background Art

[0002] Image processing for organoid identification refers to the process of identifying, segmenting, and quantitatively analyzing organoids by performing a series of computer processing on microscopic images taken during organoid culture. Organoids are three-dimensional structures grown in vitro from stem cells or progenitor cells, and have functional and morphological characteristics similar to real organs. Organoid research is widely used in developmental biology, disease model construction, and drug screening.

[0003] Microscope imaging of organoids is often accompanied by background noise, optical artifacts or other impurities. These interfering information will affect the boundary recognition and feature extraction of organoids. In organoid culture images, the contrast between the background and the organoid is sometimes not obvious, especially when the edge of the organoid is blurred with the background. Traditional image processing methods have difficulty separating the true boundary of the organoid. Therefore, how to suppress the mixed information in organoid culture images to improve the recognition effect of organoid identification is a difficult problem faced by the industry. Summary of the invention

[0004] The present application provides an image processing method and a processing system for organoid identification, which can suppress mixed information in organoid culture images to improve the recognition effect of organoid identification.

[0005] In a first aspect, the present application provides an image processing method for organoid identification, the image processing method comprising the following steps: Collect images of the organoid culture at different times to obtain an organoid culture image set; For each organoid culture image in the organoid culture image set, performing spot feature extraction on the organoid culture image to obtain image spot features of the organoid culture image, determining the structural heterogeneous entropy of the organoid culture image according to the image spot features, and then screening the organoid culture image set based on all the structural heterogeneous entropies to obtain a subset of organoid culture images of the organoid culture; Determining a pixel fluctuation peak of each organoid culture image in the organoid culture image subset, suppressing confounding information in the organoid culture image according to the corresponding pixel fluctuation peak, and thereby obtaining an image cluster of the organoid culture after suppressing confounding information; Image clusters after suppressing confounding information are used for organoid identification.

[0006] In this embodiment, spot feature extraction is performed on the organoid culture image to obtain the image spot features of the organoid culture image, which specifically includes: performing spot detection on the organoid culture image to obtain different detection spots in the organoid culture image; Feature extraction is performed on each detection spot to obtain image spot features of the organoid culture image.

[0007] In this embodiment, a Laplace-Gaussian filter is used to perform spot detection on the organoid culture image.

[0008] In this embodiment, feature extraction is performed on each detection spot by means of scale-invariant feature transformation.

[0009] In this embodiment, the organoid culture image set is screened based on all structural heterogeneous entropies to obtain the organoid culture image subset of the organoid culture, which specifically includes: setting a threshold for structural heterogeneity of the organoid culture; For each organoid culture image in the organoid culture image set, obtaining the structural heterogeneity entropy corresponding to each organoid culture image; An image set consisting of all organoid culture images whose structural heterogeneity entropy is less than the structural heterogeneity threshold is used as an organoid culture image subset of the organoid culture.

[0010] In this embodiment, determining the pixel fluctuation peak of each organoid culture image in the organoid culture image subset specifically includes: Set a pixel sliding window; For each organoid culture image in the organoid culture image subset, traverse each pixel point in the organoid culture image using the pixel sliding window, and then draw a pixel fluctuation curve of the pixel sliding window traversal process; The pixel fluctuation peak of the organoid culture image is extracted from the pixel fluctuation curve, and then the pixel fluctuation peak of each organoid culture image in the organoid culture image subset is obtained.

[0011] In this embodiment, the mixed information in the organoid culture image is suppressed according to the corresponding pixel fluctuation peak, and the image cluster after the mixed information of the organoid culture is obtained specifically includes: For each organoid culture image, determine the discreteness of the pixel fluctuation peak corresponding to the organoid culture image; Suppressing the confounding information in the organoid culture image based on the discreteness of the pixel fluctuation peak to obtain an image corresponding to the organoid culture image after suppressing the confounding information, and further obtaining an image corresponding to each organoid culture image after suppressing the confounding information; A data cluster consisting of all images after suppressing confounding information is used as an image cluster after suppressing confounding information of the organoid culture.

[0012] In this embodiment, using the image clusters after suppressing the confounding information for organoid recognition is to input the image clusters after suppressing the confounding information into a pre-trained deep learning model for organoid recognition.

[0013] In this embodiment, images of organoid cultures at different time points are collected using a live cell imaging device.

[0014] In a second aspect, the present application provides an image processing system for organoid identification, which is used to perform an image processing method for organoid identification. The image processing system comprises: An acquisition module, which acquires images of the organoid culture at different times to obtain an organoid culture image set; an image screening module, configured to extract spot features of each organoid culture image in the organoid culture image set to obtain image spot features of the organoid culture image, determine the structural heterogeneous entropy of the organoid culture image according to the image spot features, and then screen the organoid culture image set based on all the structural heterogeneous entropies to obtain a subset of organoid culture images of the organoid culture; a mixed information suppression module, for determining a pixel fluctuation peak of each organoid culture image in the organoid culture image subset, suppressing mixed information in the organoid culture image according to the corresponding pixel fluctuation peak, and thereby obtaining an image cluster of the organoid culture after suppressing mixed information; The image application module is used to use the image clusters after suppressing the mixed information for organoid identification.

[0015] The technical solution provided by the embodiments disclosed in this application has the following beneficial effects: An organoid culture image set is obtained by collecting images of organoid culture at different times; for each organoid culture image in the organoid culture image set, spot features are extracted to obtain image spot features of the organoid culture image, the structural heterogeneous entropy of the organoid culture image is determined according to the image spot features, and then the organoid culture image set is screened based on all structural heterogeneous entropies to obtain an organoid culture image subset of the organoid culture; the pixel fluctuation peak of each organoid culture image in the organoid culture image subset is determined, and the mixed information in the organoid culture image is suppressed according to the corresponding pixel fluctuation peak, so as to obtain an image cluster of the organoid culture with the mixed information suppressed; the image cluster with the mixed information suppressed is used for organoid recognition.

[0016] It can be seen that in the present application, the mixed information in the organoid culture image can be suppressed; wherein, by spot feature extraction and calculation of structural heterogeneous entropy, the interference of mixed information in the organoid culture image can be effectively reduced, the spot feature extraction focuses on the local features in the organoid culture image, and can ignore smooth or irrelevant background information, which can help screen out areas that do not have significant biological significance, thereby suppressing background noise and irrelevant textures, and the structural heterogeneous entropy measures the complexity of the structure in the organoid culture image. By setting the heterogeneous entropy threshold, those organoid culture images containing too much interference can be removed, thereby further suppressing invalid mixed information; then, by analyzing the pixel fluctuation peak, the important structural area in the organoid culture image can be accurately identified to avoid interference in the feature extraction process, and by suppressing mixed information, the background noise and irrelevant structure in the organoid culture image can be significantly reduced, so that the main biological characteristics of the organoid culture are more prominent, so as to improve the recognition effect of organoid recognition; finally, the image cluster after suppressing the mixed information is used for organoid recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0018] Figure 1 is a flow chart of an image processing method for organoid identification provided in the present application; Figure 2 It is a module structure diagram of the image processing system for organoid identification provided in this application. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0020] The embodiment of the present application provides an image processing method and a processing system for organoid recognition, the core of which is to collect images of organoid culture at different times to obtain an organoid culture image set; for each organoid culture image in the organoid culture image set, spot feature extraction is performed on the organoid culture image to obtain the image spot feature of the organoid culture image, the structural heterogeneous entropy of the organoid culture image is determined according to the image spot feature, and then the organoid culture image set is screened based on all the structural heterogeneous entropies to obtain a subset of organoid culture images of the organoid culture; the pixel fluctuation peak of each organoid culture image in the organoid culture image subset is determined, and the mixed information in the organoid culture image is suppressed according to the corresponding pixel fluctuation peak, and then the image cluster of the organoid culture after the mixed information is suppressed is obtained; the image cluster after the mixed information is suppressed is used for organoid recognition. The above scheme can suppress the mixed information in the organoid culture image to improve the recognition effect of organoid recognition.

[0021] Embodiment 1 In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods. Figure 1 As shown in FIG. 1 , this figure is an exemplary flow chart of an image processing method for organoid identification according to this embodiment of the present application, and the image processing method includes the following steps: In step S1, images of the organoid culture at different times are collected to obtain an organoid culture image set.

[0022] In specific implementation, the images of organoid cultures at different times can be collected by a living cell imaging device; the living cell imaging device can be used to collect images of organoid cultures at different times. The living cell imaging device combines a microscope and a control system to monitor and record the dynamic changes of organoids during the culture process in real time; in this application, the organoid culture image set is a collection of multiple organoid culture images arranged in chronological order.

[0023] In step S2, for each organoid culture image in the organoid culture image set, spot feature extraction is performed on the organoid culture image to obtain image spot features of the organoid culture image, and the structural heterogeneous entropy of the organoid culture image is determined according to the image spot features, and then the organoid culture image set is screened based on all the structural heterogeneous entropies to obtain a subset of organoid culture images of the organoid culture.

[0024] In this embodiment, spot feature extraction is performed on the organoid culture image to obtain the image spot features of the organoid culture image in the following manner, namely: performing spot detection on the organoid culture image to obtain different detection spots in the organoid culture image; Feature extraction is performed on each detection spot to obtain image spot features of the organoid culture image.

[0025] In specific implementation, first, the Laplace-Gaussian filter can be used to detect spots on the organoid culture image. The Laplace-Gaussian filter combines Gaussian smoothing and the Laplace operator. It is mainly used to detect local extreme points in the organoid culture image, that is, to detect spots. The detected spots may be caused by bubbles, dust and other reasons in the organoid culture. The core idea is to reduce noise through Gaussian smoothing, and then use the Laplace operator to detect spots. In actual implementation, Gaussian filters of different scales can be applied to the organoid culture image to smooth the organoid culture image and reduce noise. The second-order derivative is calculated using the Laplace operator on the smoothed image. number, detect local extreme values, and identify spots of different sizes in the organoid culture image through multi-scale detection, that is, different detection spots in the organoid culture image; then, after the spot detection is completed, the features of each detection spot can be extracted by means of scale-invariant feature transformation. The scale-invariant feature transform (SIFT algorithm) is a feature extraction algorithm widely used in computer vision, which can extract detection spot features in organoid culture images from different scales and rotation angles, and take the set of all features as the image spot feature of the organoid culture image, which is the density distribution feature of the detection spots in the organoid culture image.

[0026] Preferably, in this embodiment, the structural heterogeneity entropy of the organoid culture image may be determined according to the image spot features in the following manner, namely: dividing the organoid culture image into a plurality of sub-regions; Determine the spot distribution characteristics of each sub-region by using the image spot characteristics; Determine the structural entropy of each sub-region according to the corresponding spot distribution characteristics; The structural heterogeneity entropy of the organoid culture image is determined based on the structural entropy of each sub-region.

[0027] It should be noted that in the present application, structural heterogeneity entropy is used to describe the complexity of the image structure in the organoid culture image. A higher structural heterogeneity entropy indicates that the spot distribution and local structure of the organoid culture image are more complex, indicating that there are obvious dust bubbles, morphological heterogeneity or tissue remodeling in the organoid culture, which is not conducive to the identification of the organoid culture. A lower structural heterogeneity entropy indicates that the spot distribution is more uniform and the local structure of the organoid culture image is relatively simple, which may indicate that the organoid culture is in a relatively stable growth stage; the spot distribution feature indicates the density feature of the spots detected in the corresponding sub-region; the structural entropy indicates the complexity of the image structure in the corresponding sub-region.

[0028] In the specific implementation, first, the organoid culture image is divided into multiple sub-regions, that is, the organoid culture image is divided into multiple sub-regions (such as 3×3 or 5×5 pixel windows), each sub-region contains multiple detected spots; then, the spot distribution characteristics of each sub-region can be determined by the image spot characteristics, that is, the spot distribution characteristics of each sub-region can be obtained in the image spot characteristics; then, the structural entropy of each sub-region can be determined according to the corresponding spot distribution characteristics, and the structural entropy can be determined according to the following method: structural entropy of the ith sub-region = spot distribution characteristics of the ith sub-region * log spot distribution characteristics of the ith sub-region, and the structural entropy of each sub-region can be obtained by the above method; finally, the structural heterogeneous entropy of the organoid culture image can be determined based on the structural entropy of each sub-region, that is, the structural entropy of each sub-region is weighted averaged, so that the result is used as the structural heterogeneous entropy of the organoid culture image, and the weight of the structural entropy of each sub-region can be set by data analysis and historical experience, which will not be repeated here; the structural heterogeneous entropy of each organoid culture image in the organoid culture image set can be obtained by the above method.

[0029] In this embodiment, the organoid culture image set is screened based on all structural heterogeneous entropies to obtain the organoid culture image subset of the organoid culture, which can be specifically obtained in the following manner, namely: setting a threshold for structural heterogeneity of the organoid culture; For each organoid culture image in the organoid culture image set, obtaining the structural heterogeneity entropy corresponding to each organoid culture image; An image set consisting of all organoid culture images whose structural heterogeneity entropy is less than the structural heterogeneity threshold is used as an organoid culture image subset of the organoid culture.

[0030] In the specific implementation, first, a threshold of structural heterogeneity entropy, i.e., structural heterogeneity threshold, needs to be set for the image screening of organoid culture, which is used to distinguish organoid culture images with different heterogeneity levels. An appropriate structural heterogeneity threshold can be selected according to the experimental purpose or the growth characteristics of the organoid culture. For example, if the goal is to screen organoid culture images with relatively simple or uniform structures, a lower heterogeneity entropy threshold can be set; if organoid culture images with complex structures are to be screened out, the structural heterogeneity threshold can be set higher, which will not be repeated here; then, for each organoid culture image in the organoid culture image set, the structural heterogeneity entropy corresponding to each organoid culture image is obtained; finally, the structural heterogeneity entropy corresponding to each organoid culture image is compared with the structural heterogeneity threshold respectively, so that the image set consisting of all organoid culture images with structural heterogeneity entropy less than the structural heterogeneity threshold is taken as the organoid culture image subset of the organoid culture.

[0031] It should be noted that confounding information may come from noise, background interference or irrelevant image structure. Confounding information will affect the recognition and analysis of organoid culture images. Through spot feature extraction and calculation of structural heterogeneous entropy, the interference of confounding information in organoid culture images can be effectively reduced; among them, spot feature extraction focuses on local features in organoid culture images and can ignore smooth or irrelevant background information, which can help screen out areas that do not have significant biological significance, thereby suppressing background noise and irrelevant textures; structural heterogeneous entropy measures the complexity of the structure in organoid culture images. By setting the heterogeneous entropy threshold, organoid culture images containing too much interference can be removed, thereby further suppressing invalid confounding information.

[0032] In step S3, the pixel fluctuation peak of each organoid culture image in the organoid culture image subset is determined, and the mixed information in the organoid culture image is suppressed according to the corresponding pixel fluctuation peak, so as to obtain an image cluster of the organoid culture with suppressed mixed information.

[0033] In this embodiment, the pixel fluctuation peak of each organoid culture image in the organoid culture image subset may be determined in the following manner, namely: Set a pixel sliding window; For each organoid culture image in the organoid culture image subset, traverse each pixel point in the organoid culture image using the pixel sliding window, and then draw a pixel fluctuation curve of the pixel sliding window traversal process; The pixel fluctuation peak of the organoid culture image is extracted from the pixel fluctuation curve, and then the pixel fluctuation peak of each organoid culture image in the organoid culture image subset is obtained.

[0034] In the specific implementation, first, in order to capture the local changes of each pixel in the organoid culture image, a pixel sliding window of appropriate size can be set. The pixel sliding window is used for pixel fluctuation detection in the local area. The selection of the pixel sliding window size needs to balance the ability to capture local features and the computational complexity. Smaller windows are more sensitive to subtle changes, while larger windows are suitable for larger-scale structures. The pixel sliding window size selected in this application is 3×3 pixels. The window size can also be set to other values ​​in actual implementation, which is not limited here; then, for each organoid culture image in the organoid culture image subset, a pixel sliding window is used to traverse each pixel in the organoid culture image, that is, a sliding window is used to traverse each pixel in the organoid culture image, and the pixel value fluctuation in the area is calculated. The local second-order derivative of the values ​​of all pixels in the area can be calculated. , and use the local second-order derivative as the pixel value fluctuation of the area. The pixel value fluctuation is used to measure the pixel value fluctuation, so that multiple pixel value fluctuations in the pixel sliding window traversal process can be obtained, all pixel value fluctuations are plotted into a coordinate system, and all pixel value fluctuations are connected and fitted to obtain the pixel fluctuation curve of the pixel sliding window traversal process; finally, the pixel fluctuation peak of the organoid culture image can be extracted from the pixel fluctuation curve, that is, the peak detection algorithm is used to identify the point with the largest local fluctuation amplitude in the pixel fluctuation curve, that is, the pixel fluctuation peak. For each organoid culture image, the pixel fluctuation peak of the organoid culture image has multiple values, and each value identifies the point with the largest local fluctuation amplitude in the pixel fluctuation curve. The pixel fluctuation peak of each organoid culture image in the organoid culture image subset can be obtained in the above manner.

[0035] In this embodiment, the mixed information in the organoid culture image is suppressed according to the corresponding pixel fluctuation peak, and then the image cluster of the organoid culture after the mixed information is suppressed is obtained in the following manner, namely: For each organoid culture image, determine the discreteness of the pixel fluctuation peak corresponding to the organoid culture image; Suppressing the confounding information in the organoid culture image based on the discreteness of the pixel fluctuation peak to obtain an image corresponding to the organoid culture image after suppressing the confounding information, and further obtaining an image corresponding to each organoid culture image after suppressing the confounding information; A data cluster consisting of all images after suppressing confounding information is used as an image cluster after suppressing confounding information of the organoid culture.

[0036] In the specific implementation, first, for each organoid culture image in the organoid culture image subset, all values ​​of the pixel fluctuation peak of the organoid culture image are obtained, so that the variance of all values ​​can be used as the discreteness of the pixel fluctuation peak, and the discreteness is used to measure the distribution of the pixel fluctuation peak in the organoid culture image; then, based on the discreteness of the pixel fluctuation peak, the confounding information in the organoid culture image is suppressed to obtain the image of the organoid culture image after suppressing the confounding information. In the actual implementation, the image after suppressing the confounding information can be determined according to the following formula:

[0037] in, represents the image after suppressing the mixed information corresponding to the organoid culture image n, and represent the pixel points with coordinates (i, j) and (m, g) in the organoid culture image n, respectively. The pixel peak fluctuation of the two-dimensional positioning map of the head is represented. The image after suppressing confounding information corresponding to each organoid culture image can be obtained by the above method. Finally, the data cluster composed of all the images after suppressing confounding information can be used as the image cluster of the organoid culture after suppressing confounding information.

[0038] It should be noted that by analyzing the pixel fluctuation peaks, the important structural areas in the organoid culture images can be accurately identified to avoid interference in the feature extraction process. By suppressing mixed information, the background noise and irrelevant structures in the organoid culture images can be significantly reduced, making the main biological characteristics of the organoid culture more prominent, thereby improving the recognition effect of organoid identification.

[0039] In step S4, the image clusters after suppressing the confounding information are used for organoid identification.

[0040] In specific implementation, the image clusters after suppressing the confounding information are used for organoid identification by inputting the image clusters after suppressing the confounding information into a pre-trained deep learning model for organoid identification; first, a deep learning model suitable for image classification can be selected, such as a convolutional neural network (CNN); then, the image clusters after suppressing the confounding information are input into the deep learning model for identification and classification, so that the identification results can be visualized to complete the identification of the organoid.

[0041] It can be seen that in the present application, the mixed information in the organoid culture image can be suppressed; wherein, by spot feature extraction and calculation of structural heterogeneous entropy, the interference of mixed information in the organoid culture image can be effectively reduced, the spot feature extraction focuses on the local features in the organoid culture image, and can ignore smooth or irrelevant background information, which can help screen out areas that do not have significant biological significance, thereby suppressing background noise and irrelevant textures, and the structural heterogeneous entropy measures the complexity of the structure in the organoid culture image. By setting the heterogeneous entropy threshold, those organoid culture images containing too much interference can be removed, thereby further suppressing invalid mixed information; then, by analyzing the pixel fluctuation peak, the important structural area in the organoid culture image can be accurately identified to avoid interference in the feature extraction process, and by suppressing mixed information, the background noise and irrelevant structure in the organoid culture image can be significantly reduced, so that the main biological characteristics of the organoid culture are more prominent, so as to improve the recognition effect of organoid recognition; finally, the image cluster after suppressing the mixed information is used for organoid recognition.

[0042] In summary, the technical solution adopted in the present application can suppress the mixed information in the organoid culture image to improve the recognition effect of organoid identification.

[0043] Embodiment 2 The present application provides an image processing system for organoid identification, referring to Figure 2 As shown, this figure is a schematic diagram of an image processing system according to this embodiment of the present application, and the image processing system includes: An acquisition module 100 acquires images of the organoid culture at different times to obtain an organoid culture image set; An image screening module 200 is used for performing spot feature extraction on each organoid culture image in the organoid culture image set to obtain image spot features of the organoid culture image, determining the structural heterogeneous entropy of the organoid culture image according to the image spot features, and then screening the organoid culture image set based on all the structural heterogeneous entropies to obtain a subset of organoid culture images of the organoid culture; A confounding information suppression module 300 is used to determine the pixel fluctuation peak of each organoid culture image in the organoid culture image subset, suppress the confounding information in the organoid culture image according to the corresponding pixel fluctuation peak, and thus obtain an image cluster of the organoid culture after suppressing the confounding information; The image application module 400 is used to use the image clusters after suppressing the mixed information for organoid identification.

[0044] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0045] A person skilled in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, the storage medium including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically-erasable programmable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0046] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

Claims

1. An image processing method for organoid identification, characterized in that: The processing method comprises the following steps: Collect images of the organoid culture at different times to obtain an organoid culture image set; For each organoid culture image in the organoid culture image set, performing spot feature extraction on the organoid culture image to obtain image spot features of the organoid culture image, determining the structural heterogeneous entropy of the organoid culture image according to the image spot features, and then screening the organoid culture image set based on all the structural heterogeneous entropies to obtain a subset of organoid culture images of the organoid culture; Determining a pixel fluctuation peak of each organoid culture image in the organoid culture image subset, suppressing confounding information in the organoid culture image according to the corresponding pixel fluctuation peak, and thereby obtaining an image cluster of the organoid culture after suppressing confounding information; The image clusters after suppressing the mixed information are used for organoid identification; The organoid culture image set is screened based on all structural heterogeneous entropies to obtain the organoid culture image subset of the organoid culture, which specifically includes: setting a threshold for structural heterogeneity of the organoid culture; For each organoid culture image in the organoid culture image set, obtaining the structural heterogeneity entropy corresponding to each organoid culture image; taking an image set consisting of all organoid culture images whose structural heterogeneity entropy is less than the structural heterogeneity threshold as an organoid culture image subset of the organoid culture; Wherein, determining the pixel fluctuation peak of each organoid culture image in the organoid culture image subset specifically includes: Set a pixel sliding window; For each organoid culture image in the organoid culture image subset, traverse each pixel point in the organoid culture image using the pixel sliding window, and then draw a pixel fluctuation curve during the pixel sliding window traversal process; The pixel fluctuation peak of the organoid culture image is extracted from the pixel fluctuation curve, and then the pixel fluctuation peak of each organoid culture image in the organoid culture image subset is obtained.

2. An image processing method for organoid identification according to claim 1, characterized in that: Performing spot feature extraction on the organoid culture image to obtain the image spot features of the organoid culture image specifically includes: performing spot detection on the organoid culture image to obtain different detection spots in the organoid culture image; Feature extraction is performed on each detection spot to obtain image spot features of the organoid culture image.

3. An image processing method for organoid identification according to claim 2, characterized in that: Blob detection on organoid culture images using Laplace-Gaussian filter.

4. An image processing method for organoid identification according to claim 2, characterized in that: The features of each detection spot are extracted by means of scale-invariant feature transformation.

5. The image processing method for organoid identification according to claim 1, characterized in that: Suppressing the mixed information in the organoid culture image according to the corresponding pixel fluctuation peak, and then obtaining the image cluster of the organoid culture after suppressing the mixed information specifically includes: For each organoid culture image, determine the discreteness of the pixel fluctuation peak corresponding to the organoid culture image; Suppressing the confounding information in the organoid culture image based on the discreteness of the pixel fluctuation peak to obtain an image corresponding to the organoid culture image after suppressing the confounding information, and further obtaining an image corresponding to each organoid culture image after suppressing the confounding information; A data cluster consisting of all images after suppressing confounding information is used as an image cluster after suppressing confounding information of the organoid culture.

6. The image processing method for organoid identification according to claim 1, characterized in that: Using the image clusters after suppressing the mixed information for organoid recognition is to input the image clusters after suppressing the mixed information into a pre-trained deep learning model for organoid recognition.

7. The image processing method for organoid identification according to claim 1, characterized in that: Images of organoid cultures at different times were collected using live cell imaging equipment.

8. An image processing system for organoid identification, used to execute the image processing method for organoid identification according to any one of claims 1 to 7, characterized in that: The processing system comprises: An acquisition module, which acquires images of the organoid culture at different times to obtain an organoid culture image set; an image screening module, configured to extract spot features of each organoid culture image in the organoid culture image set to obtain image spot features of the organoid culture image, determine the structural heterogeneous entropy of the organoid culture image according to the image spot features, and then screen the organoid culture image set based on all the structural heterogeneous entropies to obtain a subset of organoid culture images of the organoid culture; a mixed information suppression module, for determining a pixel fluctuation peak of each organoid culture image in the organoid culture image subset, suppressing mixed information in the organoid culture image according to the corresponding pixel fluctuation peak, and thereby obtaining an image cluster of the organoid culture after suppressing mixed information; The image application module is used to use the image clusters after suppressing the mixed information for organoid identification.

Images