An automated stem cell detection method, system, terminal and storage medium

Abstract

The present application relates to an automated stem cell detection method, system, terminal and storage medium. The method includes: acquiring a cell image, generating a cell image training set; inputting the cell image training set into a deep learning model for a first round of model training, and outputting the first round of cell prediction results of the cell image training set through the deep learning model; updating the initial cell label of the cell image according to the cell prediction result, and tracking the cell image according to the updated cell label to obtain a cell tracking result; updating the initial training label of the cell image training set according to the cell tracking result, and inputting the updated cell image training set into a deep learning model for iterative training to obtain a trained cell detection model. The embodiment of the present application does not require manual labeling, the training process is simple, good performance is achieved, and the cost of training is greatly reduced, and the training efficiency is improved.

Description

Technical Field

[0001] The present application belongs to the field of biomedical image processing technology, and in particular relates to an automated stem cell detection method, system, terminal and storage medium. Background Art

[0002] Observing the behavior of cells helps us better understand their biological mechanisms, such as tissue formation and repair, wound healing, and tumor formation. When studying cell behavior, it is very useful to track their movement trajectory, especially for stem cells. Take induced pluripotent stem cells (iPSCs) technology as an example. This technology has been used to treat diseases such as platelet deficiency, spinal cord injury, macular degeneration, Parkinson's and Alzheimer's. However, this technology still has the problem of inefficiency - the rate of cell reprogramming in most reprogramming schemes is very low, which greatly limits the research and application of induced pluripotent stem cells in scientific research and clinical fields.

[0003] At present, the detection and tracking of stem cells mainly rely on manual labeling, or training deep models based on manual labeling. The training process requires a large amount of data sets, which greatly increases the difficulty and cost of training. Summary of the invention

[0004] The present application provides an automated stem cell detection method, system, terminal and storage medium, which aim to solve at least one of the above-mentioned technical problems in the prior art to a certain extent.

[0005] In order to solve the above problems, this application provides the following technical solutions:

[0006] An automated stem cell detection method, comprising:

[0007] Acquire a cell image, generate a cell image training set, and use the initial cell label of the cell image as the initial training label of the cell image training set;

[0008] Inputting the cell image training set into a deep learning model for a first round of model training, and outputting a first round of cell prediction results of the cell image training set through the deep learning model;

[0009] updating the initial cell label of the cell image according to the cell prediction result, and performing cell tracking on the cell image according to the updated cell label to obtain a cell tracking result;

[0010] The initial training labels of the cell image training set are updated according to the cell tracking results, and the updated cell image training set is input into the deep learning model for iterative training to obtain a trained cell detection model;

[0011] The trained cell prediction model is used to perform cell detection and tracking on the cell image to be detected.

[0012] The technical solution adopted by the embodiment of the present application also includes: the obtaining of cell images also includes:

[0013] Perform brightness, contrast, scaling, rotation, cropping, and mirror filling operations on each cell image in turn to obtain n enhanced images of each cell image;

[0014] The initial cell markers of each cell image are scaled, rotated, cropped, and mirrored and filled in turn to generate cell markers for each enhanced image.

[0015] The technical solution adopted by the embodiment of the present application also includes: the initial cell label of the cell image as the initial training label of the cell image training set also includes:

[0016] Processing the fluorescent image corresponding to the cell image to obtain an initial cell label of the cell image;

[0017] Or perform cell detection on the cell image based on a cell detector to obtain an initial cell label of the cell image.

[0018] The technical solution adopted in the embodiment of the present application also includes: the deep learning model is a U-Net model, and the U-Net model uses binary cross entropy as a loss function.

[0019] The technical solution adopted by the embodiment of the present application also includes: the updating of the initial cell mark of the cell image according to the cell prediction result is specifically:

[0020] Perform weighted summation of the cell prediction results of n enhanced images of each cell image;

[0021] The weighted sum of each cell image is added to the initial cell label of the cell image to serve as a new cell label of each cell image.

[0022] The technical solution adopted by the embodiment of the present application also includes: the cell tracking of the cell image according to the updated cell marker is specifically:

[0023] Calculate the area of ​​a cell marker in the tth frame and the t+1th frame respectively;

[0024] Calculate the overlapping area of ​​the cell marker in the tth frame and the t+1th frame;

[0025] It is determined whether the ratio of the overlapping area to the area of ​​the cell marker in the tth frame is greater than a set threshold. If it is greater than the set threshold, it is determined that the cell markers in the tth frame and the t+1th frame are the same cell, and the cell tracking result is obtained.

[0026] The technical solution adopted by the embodiment of the present application also includes: the updating of the initial training labels of the cell image training set according to the cell tracking result also includes:

[0027] Performing error tracking object detection on the cell tracking results, and eliminating cell labels of the detected error tracking objects, and generating training labels for the next round of model training;

[0028] The specific steps of performing the error tracking object detection on the cell tracking result are as follows: determining whether the number of consecutive frames of the tracking object in the cell tracking result is greater than the set frame number α; if so, determining that the tracking object is a cell; otherwise, re-tracking the tracking object, and determining whether there are objects associated with the tracking object in the next consecutive β frames; if so, determining that the tracking object is a cell; if not, determining that the tracking object is an error tracking object, and clearing the cell marker of the error tracking object from the cell image.

[0029] Another technical solution adopted by the embodiment of the present application is: an automated stem cell detection system, comprising:

[0030] Data acquisition module: used to acquire cell images, generate a cell image training set, and use the initial cell labels of the cell images as initial training labels of the cell image training set;

[0031] Model training module: used for inputting the cell image training set into the deep learning model for a first round of model training, and outputting the first round of cell prediction results of the cell image training set through the deep learning model;

[0032] A cell tracking module: used for updating the initial cell mark of the cell image according to the cell prediction result, and performing cell tracking on the cell image according to the updated cell mark to obtain a cell tracking result;

[0033] Data update module: used to update the initial training labels of the cell image training set according to the cell tracking results, and input the updated cell image training set into the deep learning model for iterative training to obtain a trained cell detection model, and perform cell detection and tracking on the cell image to be detected according to the trained cell prediction model.

[0034] Another technical solution adopted by the embodiment of the present application is: a terminal, the terminal includes a processor and a memory coupled to the processor, wherein:

[0035] The memory stores program instructions for implementing the automated stem cell detection method;

[0036] The processor is used to execute the program instructions stored in the memory to control automated stem cell detection.

[0037] Another technical solution adopted by the embodiment of the present application is: a storage medium storing program instructions executable by a processor, wherein the program instructions are used to execute the automated stem cell detection method.

[0038] Compared with the prior art, the beneficial effects of the embodiments of the present application are as follows: the automated stem cell detection method, system, terminal and storage medium of the embodiments of the present application improve the credibility of the labeling by weighted summing the cell prediction results of the n enhanced images corresponding to each cell image; by adding the weighted summation result of each cell image to the initial cell label of the cell image, the performance degradation of the model is prevented; the cells are tracked according to the added results, the training labels are updated according to the tracking results and iterative training is performed again to obtain the final cell detection model. The embodiments of the present application do not require manual labels, the training process is simple, the labor cost is reduced while obtaining better performance, and the training cost is greatly reduced, and the training efficiency is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flow chart of the automated stem cell detection method of an embodiment of the present application;

[0040] Figure 2 A schematic diagram of overlapping area calculation according to an embodiment of the present application;

[0041] Figure 3 This is a schematic diagram of the cell tracking results of an embodiment of the present application;

[0042] Figure 4 This is a schematic diagram of the structure of the automated stem cell detection system according to an embodiment of the present application;

[0043] Figure 5 This is a schematic diagram of the terminal structure of an embodiment of the present application;

[0044] Figure 6 A schematic diagram of the structure of a storage medium according to an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0046] See also Figure 1 , is a flow chart of the automated stem cell detection method of an embodiment of the present application. The automated stem cell detection method of an embodiment of the present application comprises the following steps:

[0047] S1: Acquire a certain number of cell images and obtain the initial cell label of each cell image;

[0048] In this step, the initial cell marker of the cell image is obtained by processing the fluorescent image corresponding to the cell image to obtain the corresponding initial cell marker, or performing cell detection on the cell image based on an unsupervised cell detector to obtain the initial cell marker.

[0049] S2: Perform data augmentation on the cell images to obtain an augmented cell image training set, and use the initial cell labels as initial training labels for the cell image training set;

[0050] In this step, the purpose of data enhancement is to increase the robustness of the model. The data enhancement method of the cell image is: brightness, contrast, scaling, rotation, cutting and mirror filling operations are performed on each cell image in turn, and n enhanced images corresponding to each cell image are obtained, and the initial cell labeling of each cell image is scaled, rotated, cut and mirrored to generate the cell labeling of each enhanced image. The parameter selection for each step of the data enhancement operation follows the principle of non-degeneration, that is: the cell prediction result obtained after the first round of model training using the enhanced cell image is compared with the initial cell labeling corresponding to the cell image, and the number of predicted cell labels is not reduced. Similarly, in addition to brightness and contrast, the initial label of each cell image is also subjected to the same operation.

[0051] S3: Input the cell image training set into the deep learning model for the first round of model training, and output the first round of cell prediction results of the cell image training set through the deep learning model;

[0052] In this step, the deep learning model is the U-Net model, and binary cross entropy is used as the loss function during model training.

[0053] S4: performing weighted summation on the cell prediction results of the n enhanced images corresponding to each cell image, and adding the weighted summation result of each cell image to the initial cell label of the cell image to obtain a new cell label of each cell image;

[0054] In this step, the embodiment of the present application uses weighted summation to improve the credibility of cell labeling. During weighted summation, the weight of the pixel value at the corresponding position of each enhanced image is 1 / n. Due to the uncertainty of model training, the direction of parameter update is likely to deviate from the expected direction. The purpose of adding the weighted summation result to the cell labeling during model training is to prevent the gradual regression of subsequent model training performance when the current round of cell prediction results is worse than the cell labeling. At the same time, since some cell features are more complex, it is difficult for the model to fully learn cell features in a limited round of parameter update process. Therefore, the prediction effect of such complex cells in the cell prediction results is not good. By updating the cell labeling, complex cells can be fully learned in the next round of model training to prevent the performance degradation of the model.

[0055] S5: performing cell tracking on the cell image according to the overlapping area of ​​the new cell markers in adjacent frames to obtain a cell tracking result;

[0056] In this step, cell tracking is performed by calculating the overlap area of ​​cell markers in adjacent frames. Figure 2 The figure is a schematic diagram of overlapping area calculation in an embodiment of the present application. First, the area A of a cell marker in the tth frame and the t+1th frame (ie, the next frame) is calculated respectively. t , A t+1 Then calculate the overlapping area of ​​the cell marker in the tth frame and the t+1th frame, that is, A t ∩A t+1 , and judge whether the ratio of the overlapping area to the area of ​​the cell mark in the tth frame is greater than the set threshold A, that is, judge If it is greater than the set threshold, it is determined that the cell markers in the tth frame and the t+1th frame are the same cell, and so on to obtain the cell tracking result. Preferably, the embodiment of the present application sets the threshold to 0.1.

[0057] S6: Detecting the wrongly tracked objects in the cell tracking results, and eliminating the cell labels of the detected wrongly tracked objects to generate training labels for the next round of model training;

[0058] In this step, bubbles and impurities generated by stem cell activity are easily misidentified as cells by the model, i.e., wrong tracking objects. Unlike the high continuity of cell tracking, the tracking results of such wrong tracking objects have a shorter duration. Figure 3As shown, it is a schematic diagram of cell tracking results, where (a) is the cell tracking result under ideal conditions, which has a high degree of continuity, (b), (c), and (d) are actual cell tracking results, and the number of continuous frames of cell tracking is short and the continuity is not strong. Therefore, the embodiment of the present application removes such erroneous tracking objects by analyzing the cell tracking results. Specifically: determine whether the number of continuous frames of the tracking object in the cell tracking result is greater than the set frame number α. If it is greater than α, the tracking object is determined to be a cell; conversely, if it is less than α, the tracking object is re-tracked, and it is determined whether there are objects associated with the tracking object in the next continuous β frames. If so, the tracking object is determined to be a cell; if not, the tracking object is determined to be an erroneous tracking object, and the cell marker of the erroneous tracking object is cleared from the cell image. Preferably, the embodiment of the present application sets α=3, β=5, which can be set according to actual operations.

[0059] S7: Input the training set cell images with updated training labels into the deep learning model for the next round of training, and output new cell prediction results through the deep learning model;

[0060] S8: iteratively execute S4-S7 until the set number of model training times is reached to obtain a trained cell prediction model;

[0061] In the embodiment of the present application, the number of model training times is set to 5 times, that is, a cell prediction model with better performance can be obtained after 5 rounds of training. The training labels of the cell images in the training set are updated according to the results of each round of training. No human intervention is required, which greatly reduces the cost of training and improves the training efficiency.

[0062] S9: Perform cell detection and tracking on the cell image to be detected based on the trained cell prediction model.

[0063] Based on the above, the automated stem cell detection method of the embodiment of the present application improves the credibility of the labeling by weighted summing the cell prediction results of the n enhanced images corresponding to each cell image; prevents the performance degradation of the model by adding the weighted summation result of each cell image to the initial cell label of the cell image; performs cell tracking according to the added result, updates the training label according to the tracking result and re-iterates the training to obtain the final cell detection model. The embodiment of the present application does not require manual labeling, the training process is simple, reduces labor costs while obtaining better performance, and greatly reduces the cost of training and improves training efficiency.

[0064] See also Figure 4 , is a schematic diagram of the structure of the automated stem cell detection system of the embodiment of the present application. The automated stem cell detection system 40 of the embodiment of the present application includes:

[0065] Data acquisition module 41: used to acquire cell images, generate a cell image training set, and use the initial cell labels of the cell images as initial training labels of the cell image training set;

[0066] Model training module 42: used for inputting the cell image training set into the deep learning model for a first round of model training, and outputting the first round of cell prediction results of the cell image training set through the deep learning model;

[0067] The cell tracking module 43 is used to update the initial cell mark of the cell image according to the cell prediction result, and perform cell tracking on the cell image according to the updated cell mark to obtain the cell tracking result;

[0068] Data updating module 44: used to update the initial training labels of the cell image training set according to the cell tracking results, and input the updated cell image training set into the deep learning model for iterative training to obtain a trained cell detection model, and perform cell detection and tracking on the cell image to be detected according to the trained cell prediction model.

[0069] See also Figure 5 , is a schematic diagram of the terminal structure of an embodiment of the present application. The terminal 50 includes a processor 51 and a memory 52 coupled to the processor 51.

[0070] The memory 52 stores program instructions for implementing the above-mentioned automated stem cell detection method.

[0071] The processor 51 is used to execute program instructions stored in the memory 52 to control the automated stem cell detection.

[0072] The processor 51 may also be referred to as a CPU (Central Processing Unit). The processor 51 may be an integrated circuit chip having the ability to process signals. The processor 51 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0073] See also Figure 6, which is a schematic diagram of the structure of the storage medium of the embodiment of the present application. The storage medium of the embodiment of the present application stores a program file 61 that can implement all the above methods, wherein the program file 61 can be stored in the above storage medium in the form of a software product, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (processor) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or terminal devices such as a computer, a server, a mobile phone, and a tablet.

[0074] The above description of the disclosed embodiments enables professionals and technicians in the field to implement or use the present application. Various modifications to these embodiments will be apparent to professionals and technicians in the field, and the general principles defined in this application can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown in the present application, but will conform to the widest range consistent with the principles and novel features disclosed in the present application.

Claims

1. An automated stem cell detection method, characterized in that: include: Acquire a cell image, generate a cell image training set, and use the initial cell label of the cell image as the initial training label of the cell image training set; Inputting the cell image training set into a deep learning model for a first round of model training, and outputting a first round of cell prediction results of the cell image training set through the deep learning model; updating the initial cell label of the cell image according to the cell prediction result, and performing cell tracking on the cell image according to the updated cell label to obtain a cell tracking result; The initial training labels of the cell image training set are updated according to the cell tracking results, and the updated cell image training set is input into the deep learning model for iterative training to obtain a trained cell detection model; Performing cell detection and tracking on the image of the cell to be detected according to the trained cell prediction model; The acquiring of cell images further comprises: Perform brightness, contrast, scaling, rotation, cropping, and mirror filling operations on each cell image in turn to obtain n enhanced images of each cell image; The initial cell markers of each cell image are scaled, rotated, cropped, and mirrored to generate cell markers for each enhanced image; The updating of the initial cell label of the cell image according to the cell prediction result is specifically as follows: Perform weighted summation of the cell prediction results of n enhanced images of each cell image; The weighted sum of each cell image is added to the initial cell label of the cell image to serve as a new cell label of each cell image.

2. The automated stem cell detection method according to claim 1, characterized in that: The step of using the initial cell label of the cell image as the initial training label of the cell image training set further comprises: Processing the fluorescent image corresponding to the cell image to obtain an initial cell label of the cell image; Or perform cell detection on the cell image based on a cell detector to obtain an initial cell label of the cell image.

3. The automated stem cell detection method according to claim 1, characterized in that: The deep learning model is a U-Net model, and the U-Net model uses binary cross entropy as a loss function.

4. The automated stem cell detection method according to claim 1, characterized in that: The cell tracking of the cell image according to the updated cell marker is specifically as follows: Calculate the area of ​​a cell marker in the tth frame and the t+1th frame respectively; Calculate the overlapping area of ​​the cell marker in the tth frame and the t+1th frame; It is determined whether the ratio of the overlapping area to the area of ​​the cell marker in the tth frame is greater than a set threshold. If it is greater than the set threshold, it is determined that the cell markers in the tth frame and the t+1th frame are the same cell, and the cell tracking result is obtained.

5. The automated stem cell detection method according to claim 4, characterized in that: The updating of the initial training labels of the cell image training set according to the cell tracking results further includes: Performing error tracking object detection on the cell tracking results, and eliminating cell labels of the detected error tracking objects, and generating training labels for the next round of model training; The specific steps of performing the error tracking object detection on the cell tracking result are as follows: determining whether the number of consecutive frames of the tracking object in the cell tracking result is greater than the set frame number α; if so, determining that the tracking object is a cell; otherwise, re-tracking the tracking object, and determining whether there are objects associated with the tracking object in the next consecutive β frames; if so, determining that the tracking object is a cell; if not, determining that the tracking object is an error tracking object, and clearing the cell marker of the error tracking object from the cell image.

6. An automated stem cell detection system for implementing the automated stem cell detection method of claim 1, characterized in that: include: Data acquisition module: used to acquire cell images, generate a cell image training set, and use the initial cell labels of the cell images as initial training labels of the cell image training set; Model training module: used for inputting the cell image training set into the deep learning model for a first round of model training, and outputting the first round of cell prediction results of the cell image training set through the deep learning model; A cell tracking module: used for updating the initial cell mark of the cell image according to the cell prediction result, and performing cell tracking on the cell image according to the updated cell mark to obtain a cell tracking result; Data update module: used to update the initial training labels of the cell image training set according to the cell tracking results, and input the updated cell image training set into the deep learning model for iterative training to obtain a trained cell detection model, and perform cell detection and tracking on the cell image to be detected according to the trained cell prediction model.

7. A terminal, characterized in that: The terminal includes a processor and a memory coupled to the processor, wherein: The memory stores program instructions for implementing the automated stem cell detection method according to any one of claims 1 to 5; The processor is used to execute the program instructions stored in the memory to control automated stem cell detection.

8. A storage medium, characterized in that: The method stores program instructions executable by a processor, wherein the program instructions are used to execute the automated stem cell detection method according to any one of claims 1 to 5.

Images