Cell detection methods, devices, equipment and media

Abstract

This application discloses a cell detection method, apparatus, device, and medium, generally relating to the field of computer technology, and specifically to the field of visual recognition. The method includes: given that a conventional slice includes a target slice outline corresponding to a target organ, calculating cell outline data in the target slice outline using a first model, wherein the first model is a model trained with labeled cell outline data for recognizing cell outlines, and the cell outline data includes attribute parameters of the cells in the target slice outline; if the target attribute parameter of the cell outline data is greater than a preset threshold, determining that the target cell is implanted in the target organ, wherein the target attribute parameter includes at least one of the following: cell density of the first cell in the target slice outline, number of cell clusters corresponding to the first cell in the target slice outline; the target cell includes a target cell; and the cell size parameter of the first cell is a cell within a preset size threshold range.

Description

Technical Field

[0001] This disclosure generally relates to the field of computer technology, specifically to the field of visual recognition, and in particular to a cell detection method, apparatus, device, and medium. Background Technology

[0002] With the continuous development of medical technology, specific methods can be used to treat intractable diseases. For example, for different types of tumors, specific target cells can be designated according to the tumor type. These targeted cells are then used to treat the tumor. This is a precision treatment method that can accurately, quickly, and efficiently treat and even cure tumors. After the target cells are injected into the body, it is necessary to observe whether they have successfully entered the body.

[0003] In related technologies, it is common practice to collect routine pathological slides from patients, and then manually review these slides. After that, the characteristics of the cells in the routine pathological slides can be manually determined to determine whether the target cells have successfully entered the patient's body.

[0004] However, the above method requires manual review and examination of the routine slides, which usually takes a long time, resulting in low review efficiency and the presence of subjective human opinions. Therefore, the accuracy of the test results is somewhat biased. Summary of the Invention

[0005] In view of the above-mentioned defects or deficiencies in the existing technology, it is desirable to provide a cell detection method, device, equipment and medium that can solve the problem that manual review and slide reading are required, which usually takes a long time, resulting in low review efficiency and the presence of human subjective opinions, thus causing a certain deviation in the accuracy of the test results.

[0006] Firstly, a cell detection method is provided, the method comprising:

[0007] When it is determined that the conventional slice includes the target slice outline corresponding to the target organ, the cell outline data in the target slice outline is calculated using a first model. The first model is a model trained by labeled cell outline data for recognizing cell outlines. The cell outline data includes the attribute parameters of the cells in the target slice outline.

[0008] If the target attribute parameter of the cell contour data is greater than a preset threshold, it is determined that the target cell is implanted in the target organ. The target attribute parameter includes at least one of the following: the cell density of the first cell in the target slice contour, the number of cell clusters corresponding to the first cell in the target slice contour; the target cell includes a target cell; and the cell size parameter of the first cell is a cell within a preset size threshold range.

[0009] In this application, after determining that the target slice outline corresponding to the target organ is included in a conventional slice, a first model trained using labeled cell outline data is used to calculate cell outline data (including attribute parameters in the target slice outline) within the target slice outline. Then, if the target attribute parameters of the cell outline data (e.g., the cell density of the first cell in the target slice outline, the number of cell clusters corresponding to the first cell in the target slice outline) exceed a preset threshold, it is determined that the target cell (e.g., the target cell) has been implanted into the target organ. Thus, by using the trained first model to calculate cell outline data and obtain the target attribute parameters within that data, the implantation of the target cell into the target organ can be efficiently and accurately confirmed without manual identification of conventional slices, improving the efficiency and accuracy of conventional slice identification.

[0010] Secondly, a cell detection device is provided for use at a first node, the device comprising:

[0011] The calculation module is used to calculate cell contour data in the target slice contour using a first model when the target slice contour corresponding to the target organ is determined to be included in the conventional slice. The first model is a model trained by labeled cell contour data for recognizing cell contours. The cell contour data includes attribute parameters of cells in the target slice contour.

[0012] A determination module is used to determine that the target cell is implanted in the target organ when the target attribute parameter of the cell contour data calculated by the calculation module is greater than a preset threshold. The target attribute parameter includes at least one of the following: the cell density of the first cell in the target slice contour, the number of cell clusters corresponding to the first cell in the target slice contour; the target cell includes a target cell; and the cell size parameter of the first cell is a cell within a preset size threshold range.

[0013] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the method described in the first aspect.

[0014] Fourthly, a computer-readable storage medium is provided having a computer program stored thereon, characterized in that the program, when executed by a processor, implements the method described in the first aspect above.

[0015] Fifthly, a computer program product is provided, which includes instructions that, when executed by a processor, implement the method described in the first aspect above.

[0016] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0017] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0018] Figure 1 This is a schematic flowchart of the cell detection method provided in the embodiments of this application;

[0019] Figure 2 This is a schematic diagram of the structure of the cell detection device provided in the embodiments of this application;

[0020] Figure 3 A schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0021] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0022] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] The following is an explanation of the terms that appear in this application:

[0024] 1. UNeXt model

[0025] The UNeXt model is a kernel-based MLP architecture that uses a labeled MLP block to effectively label and project convolutional features, greatly reducing the amount of parameter computation and computational complexity. It is suitable for fast outer contour detection of biological organs.

[0026] 2. U2Net model

[0027] U2Net is basically a U-Net composed of U-Nets, which is a two-level nested U structure.

[0028] U-Net is a typical segmentation architecture. It was originally proposed in 2015 for medical image segmentation, but it is not limited to medical images. In addition, it has evolved into many variants.

[0029] Figure 1 This is a schematic flowchart of a cell detection method provided in an embodiment of this application, as shown below. Figure 1 As shown, the method includes the following steps:

[0030] 301. Given that the conventional slice includes the target slice outline corresponding to the target organ, the cell outline data in the target slice outline is calculated using the first model.

[0031] In this embodiment of the application, the above-mentioned routine slides can be routine pathological slides.

[0032] Understandably, routine pathological sections are obtained by collecting lesion tissue of a certain size from a diseased individual and preparing them using histological methods. These sections can then be observed under a microscope to examine the lesions. The diseased individual is a living organism, which can be a human or any other organism.

[0033] In the embodiments of this application, the target organ can be any organ of an organism, such as the kidney, pancreas, female reproductive organs, male reproductive organs, thyroid gland, stomach, liver, bladder, uterus, prostate, etc.

[0034] In this embodiment of the application, the target slice outline may be a slice outline that includes the outline of the target organ.

[0035] In this embodiment, the target slice outline can be a part of a regular slice or the entirety of a regular slice; this embodiment does not limit this.

[0036] It is understandable that the aforementioned target slice outline is essentially the outline of the target organ. This target slice outline can include the complete outline of the target organ without being excessively larger than the outline of the target organ, so that the implantation status of the target cells in the target organ can be obtained subsequently through this target slice outline.

[0037] In one example, the target slice contour being part of a regular slice means that the regular slice may contain the target organ as well as physiological contours outside the target organ. In this case, it is necessary to obtain the target slice contour corresponding to the target organ from the regular slice. Generally, the target slice contour can be obtained using a second model, as detailed later, and will not be elaborated here.

[0038] In another example, the statement that the target slice outline can be the entirety of a regular slice means that the regular slice may exactly contain the target slice outline and meet the size requirements of the target slice outline. It should be noted that even when the target slice outline can be the entirety of a regular slice, it is still necessary to use the second model to obtain the target slice outline for confirmation. The specific method for obtaining this outline will be described later and will not be repeated here.

[0039] In the embodiments of this application, the first model is a model trained by labeled cell contour data and used to identify cell contours.

[0040] In the embodiments of this application, the first model mentioned above is a pre-set, trained model.

[0041] In the embodiments of this application, the first model described above can be used to calculate cell contour data in the target slice contour. It is understood that the first model can identify cell contours in the target slice contour and obtain cell contour data.

[0042] For example, the first model described above can first segment the target slice outline into multiple target sub-slice outlines labeled with order information, then obtain the cell sub-outline data in each target sub-slice outline, and then concatenate the cell sub-outline data from multiple target sub-slice outlines according to the previously labeled order information to finally obtain the cell outline data. The specific process of obtaining cell outline data by this first model will be described in detail later, and will not be repeated here.

[0043] In this embodiment of the application, the above-mentioned labeled cell contour data is manually reviewed and labeled in advance (e.g., reviewed and labeled by pathology experts) and used to train the first model, so that the first model can have the data to calculate the cell contour data in the target slice contour.

[0044] In one example, the first model mentioned above could be the u2net model.

[0045] In this embodiment of the application, the cell contour data includes the attribute parameters of the cells in the target slice contour.

[0046] In this embodiment of the application, cell contour data may include cell size information, cell contour information, cell density parameters, and the number of cell clusters.

[0047] For example, the cell cluster refers to all cells within a preset distance from the center point of any given cell to the center points of surrounding cells, divided into N cell clusters according to the preset number of cells in each cell cluster, where N is the number of cell clusters and is a positive integer. The number of cell clusters in the aforementioned cell clusters can be preset, that is, the number of cells in a cell cluster is the preset number of cells in a cell cluster.

[0048] In the embodiments of this application, the above-mentioned cell contour data can be used to confirm the number of cells.

[0049] 302. If the target attribute parameter of the above cell contour data is greater than a preset threshold, determine that the above target cells are implanted in the above target organ.

[0050] In this embodiment, the aforementioned preset threshold can be preset by the terminal or set by the user; this embodiment does not limit this.

[0051] In the embodiments of this application, the preset threshold may include at least one of the following: a preset density threshold and a preset cell cluster number threshold. For details, please refer to the following description, which will not be repeated here.

[0052] In this embodiment of the application, the target attribute parameter is at least one attribute parameter of the cell contour data.

[0053] For example, the target attribute parameters mentioned above include at least one of the following: the cell density of the first cell in the target slice outline, and the number of cell clusters in the target slice outline.

[0054] In the embodiments of this application, the target cells mentioned above include targeted cells, such as CAR-T cells.

[0055] For example, the aforementioned target cells can be target cells for different target organs, such as target cells for the kidneys, target cells for the pancreas, target cells for the female reproductive organs, target cells for the male reproductive organs, target cells for the thyroid gland, target cells for the stomach, target cells for the liver, target cells for the bladder, target cells for the uterus, and target cells for the prostate. This application embodiment does not limit this.

[0056] In this embodiment of the application, the cell size parameter of the first cell is a cell within a preset size threshold range.

[0057] For example, the preset size threshold can be a threshold range obtained based on the size parameters of the target cell. For instance, if the size of the target cell is 4nm×8nm×3nm, then the threshold range can be the threshold range for each size dimension, that is, the preset size threshold can be: 4nm(±0.2)×8nm(±0.4)×3nm(±0.15).

[0058] It is understandable that a regular slice consists of multiple cell types, while the target cell is only one type of cell. The target cell is similar in size to some cells in the regular slice. Therefore, when determining the target attribute parameter, if the target attribute parameter is the cell density of the first cell in the target slice outline, what is being determined is the overall density of the target cell and other cells with similar sizes to the target cell.

[0059] Furthermore, the first cell includes the target cell and other cells with similar sizes to the target cell. The cell size parameters of the other cells are within a preset size threshold range compared to the target cell.

[0060] Example 1: Assuming the target organ is the female reproductive organ, and the routine pathological section includes the outline of the female reproductive organ (i.e., the aforementioned target section outline), and the first model is the u2net model, then, assuming the routine pathological section confirms the inclusion of the section outline corresponding to the female reproductive organ, the u2net model is used to calculate the cell outline data of the section outline corresponding to the female reproductive organ. Subsequently, if the cell density of cells within a preset size threshold range and / or the number of cell clusters in the cell outline data are greater than a preset threshold, it can be determined that the target cells corresponding to the female reproductive organ have been implanted into the female reproductive organ.

[0061] In the method provided in this application embodiment, when a target slice outline corresponding to the target organ is determined in a conventional slice, a first model trained by annotating cell outline data for identifying cell outlines is used to calculate cell outline data in the target slice outline (the cell outline data includes attribute parameters in the target slice outline). Then, if the target attribute parameters of the cell outline data (e.g., the cell density of the first cell in the target slice outline, the number of cell clusters in the target slice outline) are greater than a preset threshold, it is determined that the target cell (e.g., the target cell) has been implanted into the target organ. In this way, the cell outline data can be calculated using the trained first model, and the target attribute parameters in the cell outline data can be obtained. Therefore, without manual identification of conventional slices, the implantation of target cells into the target organ can be efficiently and accurately confirmed, improving the efficiency and accuracy of conventional slice identification.

[0062] Optionally, as can be seen from the above, in determining the implantation of the target cells in the target organ, two target attribute parameters are compared with preset thresholds to determine the implantation of target cells in the target organ. The determination method of comparing the two target attribute parameters with preset thresholds will be explained below. It should be noted that in the actual process of determining the implantation of the target cells in the target organ, only one of the following two determination methods may be used, or both determination methods may be used simultaneously; this application embodiment does not limit this.

[0063] In another embodiment of this application, the first method of determination also provides a specific way to determine the implantation status of target cells when the target attribute parameter is the cell density of the first cell in the target slice outline. For example, when the target attribute parameter includes the cell density of the first cell in the target slice outline, the specific implementation of the aforementioned "determining that the target cell is implanted in the target organ when the target attribute parameter of the cell outline data is greater than a preset threshold" includes: using a preset cell morphology model to filter slice cells within the target slice outline range to obtain the number of first cells within the target slice outline range; determining the cell density of the first cell in the target slice outline based on the number of first cells within the target slice outline range; and determining that the target cell is implanted in the target organ when the cell density of the first cell in the target slice outline is greater than a preset density threshold.

[0064] For example, the above-mentioned preset cell morphology model is set according to the morphology of the target cell.

[0065] For example, the aforementioned preset cell morphology model may include the aspect ratio parameter and area parameter of the preset cell, which are not limited in this embodiment.

[0066] For example, the aforementioned preset cell morphology model can be preset by the terminal or customized by the user, and this application embodiment does not limit this.

[0067] As can be understood from the foregoing, the first cell refers to a cell whose size parameters are within a preset size threshold range. The cell contour data in the target slice outline essentially includes both cells within and outside this preset size threshold range. Therefore, by setting a preset cell morphology model, cells outside the preset size threshold range can be filtered out first. The remaining cells within the preset size threshold range are considered the first cells. By calculating the cell density of the first cells, it can be determined whether the target cells have been implanted in the target organ. Generally, if the cell density of the first cells is greater than a preset density threshold, it can be determined that the target cells have been implanted in the target organ.

[0068] For example, the sliced ​​cells mentioned above refer to all cells within the range of the target slice outline.

[0069] For example, the method of determining the cell density of the first cell by the number of first cells described above can be to determine the cell density of the first cell by using the ratio of the number of first cells to the area of ​​the target slice outline.

[0070] Example 2: Combining with Example 1 above, when the target organ is the female reproductive organ, the first cell density can be calculated using the following formula 1:

[0071]

[0072] Density fam Quantity represents the cell density in the female reproductive organs (i.e., the cell density of the first cell mentioned above). cell The area represents the number of cells in the female reproductive organs (i.e., the number of cells mentioned in the first section). fam This refers to the area within the female reproductive organs (i.e., the outline range of the target slice mentioned above).

[0073] After that, Density fam Compared with a preset density threshold, in Density fam If the density exceeds the preset threshold, the target cells (i.e., the aforementioned target cells) are confirmed to be implanted into the female reproductive organs.

[0074] In this way, by filtering out the first cell within the outline of the target slice, the number of the first cell within the outline of the target slice is obtained, and the cell density of the first cell is obtained. Finally, the cell density of the first cell can be compared with a preset density threshold to determine whether the target cell has been implanted in the target organ. This allows for the determination of the implantation status of the target cell without the need for manual review of routine slices, thus improving the efficiency and accuracy of determining the implantation status of the target cell.

[0075] In another embodiment of this application, a second determination method is provided, specifically for determining the implantation status of target cells when the target attribute parameter is the number of cell clusters corresponding to the first cell. For example, when the target attribute parameter includes the number of cell clusters corresponding to the first cell in the target slice outline, the aforementioned "determining that the target cell is implanted in the target organ when the target attribute parameter of the cell outline data is greater than a preset threshold" includes: using a preset cell morphology model to filter slice cells within the target slice outline range to obtain the first cell within the target slice outline range; calculating the distance parameter between the center points of each first cell in the cell outline data; selecting second cells from the first cells whose distance parameter is less than a predetermined distance threshold; dividing the second cell into M cell clusters according to the number of cells in a preset cell cluster; and determining that the target cell is implanted in the target organ when the number of cell clusters M is greater than a preset cell cluster number threshold.

[0076] For example, the above-mentioned preset cell morphology model and the specific method of filtering the slice cells within the target slice outline range to obtain the first cell within the target slice outline range can be referred to the foregoing description, and will not be repeated here.

[0077] It should be noted that in practical applications, if both the first and second determination methods are used, the first method is generally used first, followed by the second; or the second method is used first, followed by the first. Regardless of the order of these two determination methods, if the first determination method "uses a preset cell morphology model to filter the slice cells within the outline range of the target slice", then subsequent determination methods do not need to perform this step.

[0078] For example, the aforementioned predetermined distance threshold can be user-defined or determined by the cell detection device itself; this application embodiment does not limit this.

[0079] For example, the number of cells in the preset cell cluster can be user-defined or determined by the cell detection device itself; this application embodiment does not limit this.

[0080] For example, the aforementioned preset cell cluster number threshold can be user-defined or determined by the cell detection device itself; this application embodiment does not limit this.

[0081] Understandably, in the above process, the distance between the center point of each first cell within the target slice outline and the center points of other first cells can be calculated firstly. When the distance between the center point of each first cell and the center points of other first cells around it is less than a predetermined distance threshold, it can be determined that they are cells in the same cell cluster, i.e., belonging to the second cells. Thus, the data information of the second cells in the same cell cluster is written into a list. Based on the number of data information entries for the second cells in the list, all second cells in the list can be divided into M cell clusters. Finally, the number of cell clusters is counted. When the number of cell clusters M is greater than a preset cell cluster number threshold, it can be determined that the target cell has been implanted in the target organ.

[0082] In this way, by filtering out the first cell within the outline of the target slice, and then obtaining the cell clusters composed of second cells that meet the requirements in the first cell, the number of cell clusters can be compared with the preset cell cluster number threshold to determine whether the target cells have been implanted in the target organ. This allows for the determination of the implantation status of the target cells without the need for manual review of routine slices, thus improving the efficiency and accuracy of determining the implantation status of the target cells.

[0083] In another embodiment of this application, a specific method for calculating cell contour data using the first model is also provided. For example, the specific implementation of "calculating cell contour data of the conventional slice using the first model" mentioned above includes: dividing the conventional slice into X conventional sub-slices using the first model, where X is a positive integer; calculating cell sub-contour data in the X conventional sub-slices using the first model; and stitching the X conventional sub-slices together according to the slice coordinate information using the first model and processing the cell sub-contour data in the X conventional sub-slices to obtain the cell contour data.

[0084] For example, each of the above X regular sub-slices includes slice data information and slice coordinate information.

[0085] Understandably, the size of a standard slice is very large relative to the observation of cell contour data. Therefore, it is necessary to divide the standard slice into X standard sub-slices, obtain the cell sub-contour data in each standard sub-slice, and then re-stitch them together to obtain the complete cell contour data, that is, to obtain the cell contour data of the target slice contour in the entire standard slice.

[0086] For example, the process of cutting regular slices, calculating the X cell sub-contour data in regular sub-slices, and finally stitching together X regular sub-slices is all completed by the first model.

[0087] For example, the size of the aforementioned conventional sub-slices is preset. Specifically, it can be predicted by the cell detection device or set by the user; this application embodiment does not limit this.

[0088] Understandably, this first model was trained using labeled cell contour data before the cell detection method was implemented.

[0089] The aforementioned labeled cell contour data may include the following data: several fully cut first regular sub-slices and the slice coordinate information of each first regular sub-slice; fully labeled first cell sub-contour data in each of the several first regular sub-slices; a first regular slice stitched together based on the slice coordinate information of each first regular sub-slice; and first cell contour data obtained after processing the first cell sub-contours of the several first regular sub-slices. That is, during the training process using the first model, all data from the complete processing process when using the first model will be provided as training data. The first model will be iteratively trained repeatedly, ultimately obtaining a first model capable of autonomously cutting, stitching, and processing cell sub-contour data to obtain cell contour data.

[0090] For example, in the process of dividing a regular slice into X regular sub-slices, the cutting step size is smaller than the size of the X regular sub-slices. For instance, the cutting step size is 3 / 4 of the size of the X regular sub-slices.

[0091] Understandably, when the cutting step size is less than the dimensions of the aforementioned X regular sub-slices, it can be guaranteed that the cell outline of each cell in the target slice outline within the regular slice is complete and not missed. It is foreseeable that if the cutting is strictly performed according to the dimensions of X regular sub-slices, some cell outlines may be located at the edges of the regular sub-slices, resulting in some or all of the edge cell outlines in the regular sub-slices not being fully acquired. Therefore, compensating for image cutting with a size less than X regular sub-slices ensures the completeness of the final acquired cell outline data.

[0092] Example 3: Combining Examples 1 and 2 above, when the regular slice is a regular slice including female reproductive organs (i.e., the target organ mentioned above) and the first model is the u2net model, the regular slice is divided into X regular sub-slices (i.e., the X regular sub-slices mentioned above) by setting the cutting step size according to the size of the regular sub-slice, i.e., 3 / 4 of the size of the regular sub-slice, and the slice coordinate information of each regular sub-slice in the regular slice is marked for the X regular sub-slices.

[0093] Subsequently, the cell sub-contour data of each of the X regular sub-slices is obtained using the u2net model. Finally, the X regular sub-slices are stitched together according to their respective slice coordinate information using the u2net model, and the cell sub-contour data of the X regular sub-slices is processed using the u2net model to obtain cell contour data.

[0094] It should be noted that when the above-mentioned regular slice is divided into X regular sub-slices, the aforementioned target attribute parameters include the cell density of the first cell in the outline of the target slice. That is, in the first determination method, the process of obtaining the first cell density in Example 2 can be described as follows in Example 4.

[0095] Example 4: Combining with Example 1 above, when the target organ is the female reproductive organ, the first cell density in each routine subsection can be calculated using the following formula 1:

[0096]

[0097] Density fam Quantity represents the cell density in a regular subslice (i.e., the cell density of the first cell mentioned above). cell The area represents the number of cells in a regular subsection (i.e., the first cell number mentioned above). fam This refers to the area within a regular sub-slice (i.e., the range of the target slice outline mentioned above).

[0098] Subsequently, based on the preset cell density scoring method, each routine subsection was divided into high-density routine subsections, medium-density routine subsections, low-density routine subsections, and normal-density routine subsections.

[0099] Finally, for all regular subsections, the density score corresponding to the final first cell density is calculated according to Formula 2 below, i.e., the scoring formula.

[0100]

[0101] Where Score is the cell density index score, Hn is the number of high-density patches, Mn is the number of medium-density patches, Ln is the number of low-density patches, and Nn is the number of normal-density patches.

[0102] If the cell density index score is greater than 90%, the target cells (i.e., the aforementioned target cells) are confirmed to be implanted into the female reproductive organs.

[0103] Thus, the first model can be used to calculate the cell contour data of the target organ corresponding to the target slice in the conventional slice, so as to use the cell contour data to subsequently determine the implantation status of the target cells in the target organ, thereby improving the efficiency and accuracy of determining the implantation status of the target cells.

[0104] In another embodiment of this application, a specific method for processing conventional slices before calculating cell contour data is also provided. For example, before the aforementioned "calculating cell contour data of the conventional slice data information using the first model," the specific implementation includes: using a second model to obtain the outer contour data from the conventional slice data information, and determining the organ corresponding to the outer contour data as the target organ.

[0105] For example, the second model mentioned above is a model trained by annotated outer contour data and used to identify the outer contour data of the target organ.

[0106] In one example, the second model mentioned above could be the UNeXt model.

[0107] Furthermore, when the second model mentioned above can be the UNeXt model, since it is a multilayer perceptron (MLP) architecture based on convolutional kernels, it uses a labeled MLP block to effectively label and project convolutional features, and greatly reduces the amount of parameter computation and computational complexity, thereby enabling fast outer contour detection of target organs.

[0108] It is understandable that the second model described above is a pre-trained model. Specifically, the training data for the second model may include: labeled outer contour data of the organ to be identified. After iteratively training the second model using the labeled outer contour data of the organ to be identified, the second model can be used to obtain the outer contour data from the data information of the above-mentioned conventional slices and can determine that the organ corresponding to the outer contour data is the target organ.

[0109] In another embodiment of this application, a specific method for processing slice data in conventional slices before calculating cell contour data is also provided. For example, before the aforementioned "calculating cell contour data in the target slice contour using the first model", the specific implementation includes: preprocessing the conventional slices to obtain the data information of the conventional slices.

[0110] For example, the above preprocessing is used to remove image noise from the above conventional slices.

[0111] In one example, the preprocessing described above can be Gaussian filtering. Specifically, Gaussian filtering can be performed using the following formula 2.

[0112]

[0113] Where G is the Gaussian weight, x and y are the coordinate values ​​with the center point of the regular slice as the origin, and σ is the standard deviation.

[0114] For example, the preprocessing described above may further include at least one of the following: reading the conventional slices according to a preset slice level, converting the conventional slices into a predetermined format, and performing image enhancement on the conventional slices.

[0115] In one example, reading the above-mentioned regular slice according to the preset slice level can be: reading the above-mentioned regular slice according to the preset slice level of 10×.

[0116] In one example, converting the above-mentioned regular slices to a predetermined format can be done by converting regular slices to RGB format.

[0117] In one example, image enhancement of a regular slice can be achieved by using gamma transform in OpenCV to enhance image contrast. Specifically, image enhancement can be performed according to Formula 3 below.

[0118] O(r,c)=I(r,c) γ Formula 3

[0119] The output image is denoted as O, I represents the input image (i.e., a regular slice), r and c represent the grayscale value of the r-th row and c-th column, and γ is an adjustable parameter.

[0120] Understandably, before reading the aforementioned conventional slices, the conventional slices are obtained by scanning the specified tissue region (e.g., the tissue region where the target organ is located in the human body) with a scanner, resulting in multi-level, high-resolution conventional slices.

[0121] In this way, by preprocessing the conventional slides, the data in the conventional slides can be more easily obtained in subsequent steps, thereby enabling accurate and rapid determination of the implantation status of target cells in the target organ in the conventional slides, and improving the efficiency and accuracy of determining the implantation status of target cells.

[0122] Figure 2 This is a block diagram of a cell detection device according to an embodiment of this application, which can be deployed on a requesting node (e.g., the second node described above). Reference Figure 2 The device includes a calculation module 601 and a determination module 602.

[0123] The calculation module is used to calculate cell contour data in the target slice contour using a first model when the target slice contour corresponding to the target organ is determined to be included in the conventional slice. The first model is a model trained by labeled cell contour data for recognizing cell contours. The cell contour data includes attribute parameters of cells in the target slice contour.

[0124] A determination module is used to determine that the target cell is implanted in the target organ when the target attribute parameter of the cell contour data calculated by the calculation module is greater than a preset threshold. The target attribute parameter includes at least one of the following: the cell density of the first cell in the target slice contour, the number of cell clusters corresponding to the first cell in the target slice contour; the target cell includes a target cell; and the cell size parameter of the first cell is a cell within a preset size threshold range.

[0125] In one possible embodiment, when the target attribute parameter includes the cell density of the first cell in the target slice outline, the device further includes a filtering module 603, which is used to filter slice cells within the target slice outline range using a preset cell morphology model to obtain the number of the first cell within the target slice outline range; the determining module 602 is specifically used to determine the cell density of the first cell in the target slice outline based on the number of the first cell within the target slice outline range filtered by the filtering module 603; the determining module 602 is specifically used to determine that the target cell is implanted in the target organ when the cell density of the first cell in the target slice outline is greater than a preset density threshold.

[0126] In one possible embodiment, when the target attribute parameter includes the number of cell clusters corresponding to the first cell in the target slice outline, the device further includes a filtering module 603, a screening module 604, and an execution module 605. The filtering module 603 is used to filter slice cells within the target slice outline range using a preset cell morphology model to obtain the first cell within the target slice outline range. The calculation module 601 is specifically used to calculate the distance parameter between the center points of each first cell in the cell outline data. The screening module 604 is used to screen out second cells in the first cells whose distance parameter calculated by the calculation module 601 is less than a predetermined distance threshold. The execution module is used to divide the second cells screened by the screening module 604 into M cell clusters according to the number of cells in a preset cell cluster. The determination module 602 is specifically used to determine that the target cell is implanted in the target organ when the number of cell clusters M is greater than a preset cell cluster number threshold.

[0127] In one possible embodiment, the device further includes an execution module 605, which is configured to use the first model to divide the regular slice into X regular sub-slices, each of the X regular sub-slices including slice data information and slice coordinate information, where X is a positive integer; the calculation module 601 is specifically configured to use the first model to calculate cell sub-contour data in the X regular sub-slices divided by the execution module 605; the calculation module is specifically configured to use the first model to stitch the X regular sub-slices according to the slice coordinate information and process the cell sub-contour data in the X regular sub-slices to obtain the cell contour data.

[0128] In one possible embodiment, the determining module 602 is further configured to use a second model to obtain the outer contour data in the data information of the conventional slice, and determine the organ corresponding to the outer contour data as the target organ; the second model is a model trained by annotating the outer contour data to identify the outer contour data of the target organ in the conventional slice.

[0129] In one possible embodiment, the apparatus further includes an execution module 605, which is configured to preprocess the conventional slice to obtain data information of the conventional slice; wherein the preprocessing is used to remove image noise from the conventional slice.

[0130] The cell detection device provided in this application, when determining that a target slice outline corresponding to the target organ is included in a conventional slice, calculates cell outline data (including attribute parameters in the target slice outline) in the target slice outline using a first model trained on labeled cell outline data for identifying cell outlines. Then, if the target attribute parameters of the cell outline data (e.g., the cell density of the first cell in the target slice outline, the number of cell clusters corresponding to the first cell in the target slice outline) are greater than a preset threshold, it is determined that the target cell (e.g., a target cell) has been implanted into the target organ. Thus, by using the trained first model to calculate cell outline data and obtain the target attribute parameters in the cell outline data, the implantation of the target cell into the target organ can be efficiently and accurately confirmed without manual identification of conventional slices, improving the efficiency and accuracy of conventional slice identification.

[0131] It should be understood that the units described in the cell detection device correspond to the various steps in the method described in the accompanying drawings. Therefore, the operations and features described above for the method are also applicable to the cell detection device, the resource access device, and the units contained therein, and will not be repeated here. The cell detection device and the resource access device can be pre-implemented in a browser or other secure application on a computer device, or they can be loaded into a browser or other secure application on a computer device through download or other means. The corresponding units in the cell detection device and the resource access device can cooperate with the units in the computer device to implement the solutions of the embodiments of this application.

[0132] The division of modules or units mentioned in the detailed description above is not mandatory. In fact, according to the embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0133] It should be noted that for details not disclosed in the cell detection device and resource access device of the embodiments of this application, please refer to the details disclosed in the above embodiments of this application, which will not be repeated here.

[0134] The following is for reference. Figure 3 , Figure 3 A schematic diagram of a computer device suitable for implementing embodiments of this application is shown. For example... Figure 3 As shown, the computer system 1700 includes a central processing unit (CPU) 1701, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 1702 or programs loaded from storage section 1708 into random access memory (RAM) 1703. RAM 1703 also stores various programs and data required for the system's operating instructions. CPU 1701, ROM 1702, and RAM 1703 are interconnected via bus 1704. Input / output (I / O) interface 1705 is also connected to bus 1704.

[0135] The following components are connected to I / O interface 1705: an input section 1706 including a keyboard, mouse, etc.; an output section 1707 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1708 including a hard disk, etc.; and a communication section 1709 including a network interface card such as a LAN card, modem, etc. The communication section 1709 performs communication processing via a network such as the Internet. A drive 1710 is also connected to I / O interface 1705 as needed. Removable media 1711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 1710 as needed so that computer programs read from them can be installed into storage section 1708 as needed.

[0136] Specifically, according to embodiments of this application, the flowchart above refers to... Figure 1 The described process can be implemented as a computer software program. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. In such an embodiment, the computer program contains program code for performing the methods shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via communication section 1709, and / or installed from removable medium 1711. When the computer program is executed by central processing unit (CPU) 1701, it performs the functions defined in the system of this application.

[0137] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0138] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operational instructions of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two connected blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified functions or operational instructions, or using a combination of dedicated hardware and computer instructions.

[0139] The units or modules described in the embodiments of this application can be implemented in software or hardware. The described units or modules can also be housed in a processor; for example, a processor can be described as including a first receiving module, a second receiving module, and a transmitting module. The names of these units or modules do not, in certain circumstances, constitute a limitation on the unit or module itself.

[0140] In another aspect, this application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments, or may exist independently and not assembled into the electronic device. The aforementioned computer-readable storage medium stores one or more programs that, when used by one or more processors, execute the cell detection method described in this application.

[0141] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method of detecting cells, characterized by, The method comprises the following steps: In a case where it is determined that the regular slice includes a target slice contour corresponding to a target organ, cell contour data in the target slice contour is calculated by using a first model, the first model being a model trained by labeling cell contour data and used for identifying cell contours, and the cell contour data including attribute parameters of cells in the target slice contour; the first model is used to first segment the target slice contour into a plurality of target sub-slice contours labeled with sequence information, then obtain cell sub-contour data in each target sub-slice contour, and then splice the cell sub-contour data in the plurality of target sub-slice contours according to the sequence information labeled in advance to finally obtain the cell contour data; the cell contour data in the target slice contour includes cells meeting a preset size threshold range and cells not meeting the preset size threshold range; In a case where a target attribute parameter of the cell contour data is greater than a preset threshold, it is determined that target cells are implanted in the target organ, and the target attribute parameter includes a cell density of first cells in the target slice contour; The target cells include target cells; and a cell size parameter of the first cells is a cell in a preset size threshold range; In a case where the target attribute parameter includes the cell density of the first cells in the target slice contour, the determination that the target cells are implanted in the target organ in the case where the target attribute parameter of the cell contour data is greater than the preset threshold comprises: A preset cell morphology model is used to filter slice cells in a range of the target slice contour to obtain a number of first cells in the range of the target slice contour; the preset cell morphology model includes a length-width ratio parameter and an area parameter of a preset cell; The cell density of the first cells in the target slice contour is determined according to the number of the first cells in the range of the target slice contour; In a case where the cell density of the first cells in the target slice contour is greater than a preset density threshold, it is determined that the target cells are implanted in the target organ. The method comprises the following steps: The first model is used to segment the regular slice into X regular sub-slices, each of the X regular sub-slices including slice data information and slice coordinate information, and X being a positive integer; The first model is used to calculate cell sub-contour data in the X regular sub-slices; The first model is used to splice the X regular sub-slices according to the slice coordinate information and process the cell sub-contour data in the X regular sub-slices to obtain the cell contour data.

2. The method of claim 1, wherein, In a case where the target attribute parameter includes a number of cell clusters corresponding to the first cells in the target slice contour, the determination that the target cells are implanted in the target organ in the case where the target attribute parameter of the cell contour data is greater than the preset threshold comprises: A preset cell morphology model is used to filter slice cells in a range of the target slice contour to obtain a number of first cells in the range of the target slice contour; the preset cell morphology model includes a length-width ratio parameter and an area parameter of a preset cell. calculate a distance parameter between the center point positions of each first cell in the cell contour data; screen out second cells in the first cells with the distance parameter less than a predetermined distance threshold value; divide the second cells into M cell clusters according to the number of cells in a preset cell cluster; determine that the target cells are implanted in the target organ when the number of cell clusters M is greater than a preset cell cluster number threshold value.

3. The method of claim 1, wherein, Before the cell contour data in the target slice contour is calculated by using the first model, the method further comprises: obtain outer contour data in the data information of the conventional slice by using a second model, and determine that the organ corresponding to the outer contour data is a target organ; the second model is a model trained by labeling the outer contour data and used to identify the outer contour data of the target organ in the conventional slice.

4. The method of claim 1, wherein, Before the cell contour data in the target slice contour is calculated by using the first model, the method further comprises: preprocess the conventional slice to obtain data information of the conventional slice; wherein the preprocessing is used to remove image noise of the conventional slice.

5. A cell detection device, characterized by comprises: a calculation module, configured to calculate cell contour data in a target slice contour by using a first model when it is determined that the conventional slice comprises the target slice contour corresponding to a target organ, the first model being a model trained by labeling the cell contour data and used to identify the cell contour, the cell contour data comprising attribute parameters of cells in the target slice contour; the first model first divides the target slice contour into a plurality of target sub-slice contours labeled with sequence information, then obtains cell sub-contour data in each target sub-slice contour, and finally splices the cell sub-contour data in the plurality of target sub-slice contours according to the sequence information labeled in advance to obtain the cell contour data; the cell contour data in the target slice contour comprises cells meeting a preset size threshold range and cells not meeting the preset size threshold range; a determination module, configured to determine that target cells are implanted in the target organ when a target attribute parameter of the cell contour data calculated by the calculation module is greater than a preset threshold value, the target attribute parameter comprising a cell density of first cells in the target slice contour; the target cells comprise target cells; and the cell size parameter of the first cells is a cell size parameter of cells in a preset size threshold range; wherein when the target attribute parameter comprises the cell density of the first cells in the target slice contour, the determination that the target cells are implanted in the target organ when the target attribute parameter of the cell contour data calculated by the calculation module is greater than the preset threshold value comprises: filtering slice cells within the target slice contour range by using a preset cell morphology model to obtain a number of first cells within the target slice contour range; the preset cell morphology model comprises a length-width ratio parameter and an area parameter of a preset cell; determining the cell density of the first cells in the target slice contour according to the number of first cells within the target slice contour range; and In a case where a cell density of the first cell in the target slice profile is greater than a preset density threshold, it is determined that the target cell is implanted in the target organ; The method comprises the following steps: segmenting the regular slice into X regular sub-slices by using the first model, each of the X regular sub-slices comprising slice data information and slice coordinate information, X being a positive integer; calculating cell sub-profile data in the X regular sub-slices by using the first model; splicing the X regular sub-slices and processing the cell sub-profile data in the X regular sub-slices according to the slice coordinate information to obtain the cell profile data by using the first model.

6. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the method of any one of claims 1-4.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that The program is executed by the processor to implement the method of any one of claims 1-4.

8. A computer program product comprising instructions therein, wherein the computer program product is characterised in that, The instructions are executed by the processor to implement the method of any one of claims 1-4.

Images