Cell Segmentation Method, Device, Storage Medium and Program Product Based on Spatial Omics Sequencing

Abstract

The present invention discloses a cell segmentation method, device, storage medium and program product based on spatial omics sequencing. The method includes: acquiring a collected signal distribution image; performing an initial smoothing operation based on the signal distribution image to obtain an initial smoothed distribution image, performing an image segmentation operation based on the initial smoothed distribution image to obtain a current cell segmentation region set; adjusting a smoothing parameter corresponding to the smoothing operation, performing a smoothing operation and an image segmentation operation on a cell region image in the current first region set based on the adjusted smoothing parameter, and updating the current cell segmentation region set, returning to the previous screening step until a termination condition is met to obtain a target second region set; and obtaining each cell region based on the target second region set.

Description

Technical Field

[0001] The present invention relates to the field of gene technology, and particularly to a cell segmentation method and system, a computer device, a computer-readable storage medium, and a computer program product based on spatial omics sequencing. Background Art

[0002] Spatial transcriptome sequencing is a sequencing method that can read the base sequence information of RNA fragments while recording the spatial positions of these fragments. Currently, the spatial information that can be recorded by spatial omics sequencing is generally limited to two-dimensional space and is achieved by covering tissue sections onto sequencing chips. Probes capable of capturing RNA fragments are distributed on the surface of the sequencing chip, and the sequencer can read the base sequence information of the RNA fragments bound to the probes. Combining with the known positions of the probes, the spatial information of the base sequences can be obtained.

[0003] Currently, the bright spots in the signal distribution map in spatial omics sequencing do not always remain isolated and complete. Depending on the tissue section conditions and experimental conditions, the bright spots may appear blurred, crowded, with unclear boundaries, etc. Therefore, the methods in the prior art cannot accurately segment cells. For example, one limitation of spatial transcriptome sequencing is that it cannot determine which RNA fragments captured by the probes come from the same cell, which limits the ability to analyze cell specificity. In areas with dense cell signals, the RNA signals of different cells are mixed and it is difficult to form obvious boundaries. Therefore, the bright spots in the RNA signal distribution map do not always remain isolated and complete. Depending on the tissue section conditions and experimental conditions, the bright spots may appear blurred, crowded, with unclear boundaries, etc. Therefore, the methods in the prior art cannot accurately segment cells. Summary of the Invention

[0004] In order to solve the existing technical problems, the present invention provides a cell segmentation method and system, a computer device, a computer-readable storage medium, and a computer program product based on spatial omics sequencing, which can improve the accuracy of cell segmentation.

[0005] In a first aspect, a cell segmentation method based on spatial omics sequencing is provided, including: obtaining a collected signal distribution image, where the signal distribution image indicates the distribution characteristics of spatial omics signals of a tissue section; performing an initial smoothing operation based on the signal distribution image to obtain an initial smoothed distribution image, performing an image segmentation operation based on the initial smoothed distribution image to obtain a current set of cell segmentation regions; based on the current set of cell segmentation regions, screening to obtain cell region images to be further segmented to obtain a current first region set, screening cell region images that are not to be further segmented to obtain a current second region set; adjusting the smoothing parameter corresponding to the smoothing operation, based on the adjusted smoothing parameter, performing a smoothing operation and an image segmentation operation on the cell region images in the current first region set, and updating the current set of cell segmentation regions, returning to the previous screening step until a termination condition is met to obtain a target second region set; and obtaining each cell region based on the target second region set.

[0006] In a second aspect, a computer device is provided, including a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to execute the steps of the cell segmentation method based on spatial omics sequencing provided in the embodiments of the present application.

[0007] In a third aspect, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the processor is caused to execute the steps of the cell segmentation method based on spatial omics sequencing provided in the embodiments of the present application.

[0008] In a fourth aspect, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps of the cell segmentation method based on spatial omics sequencing provided in the embodiments of the present application are implemented.

[0009] The present application obtains a collected signal distribution image, performs an initial smoothing operation based on the initial signal distribution image to obtain an initial smoothed distribution image, performs an image segmentation operation based on the initial smoothed distribution image to obtain a current set of cell segmentation regions, based on the current set of cell segmentation regions, screens to obtain cell region images to be further segmented to obtain a current first region set, screens cell region images that are not to be further segmented to obtain a current second region set, and performs a smoothing operation and an image segmentation operation on the cell region images to be further segmented again until a termination condition is met to obtain a target second region set, thereby obtaining each cell region. By adjusting the smoothing parameter during multiple smoothing operations, the adjusted smoothing parameter is made more suitable for the current smoothing operation, and the cell region images in the dense region are further segmented into single cell region images to achieve better segmentation, thereby improving the accuracy of cell segmentation. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 Schematic diagram of bright spots presented in the RNA signal distribution image in one embodiment;

[0011] Figure 2 Schematic diagram of RNA signal distribution at different positions on the same tissue section in one embodiment;

[0012] Figure 3 Application environment diagram of the cell segmentation method based on spatial omics sequencing in one embodiment;

[0013] Figure 4 Flowchart of the cell segmentation method based on spatial omics sequencing in one embodiment;

[0014] Figure 5 Schematic diagram of performing multiple smoothing operations and image segmentation operations in one embodiment;

[0015] Figure 6 Schematic diagram of cells segmented by the cell segmentation method based on spatial transcriptome sequencing in one embodiment;

[0016] Figure 7 Flowchart of the cell segmentation method based on spatial omics sequencing in another embodiment;

[0017] Figure 8 Schematic diagram of the cell segmentation device based on spatial omics sequencing in one embodiment;

[0018] Figure 9 Schematic diagram of the structure of a computing device in one embodiment. Detailed implementation manners

[0019] The technical solution of the present invention will be further elaborated in detail below in conjunction with the accompanying drawings of the specification and specific embodiments.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this invention belongs. The terms used in the specification of this invention are only for the purpose of describing specific embodiments and are not intended to limit the scope of the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0021] In the following description, the expression "some embodiments" describes a subset of all possible embodiments. However, it should be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict.

[0022] Spatial omics sequencing method is a sequencing method that can read the base sequence information of ribonucleic acid (RNA) fragments while recording the spatial positions of these fragments. Currently, the spatial information that can be recorded by spatial omics sequencing is generally limited to two-dimensional space, which is achieved by covering tissue sections onto sequencing chips. For example, based on the spatial transcriptomics sequencing method, probes that can capture RNA fragments are distributed on the surface of the sequencing chip, and a gene sequencer can read the base sequence information of the RNA fragments bound to the probes. By combining the known positions of the probes, the spatial information of the base sequences can be obtained.

[0023] One limitation of current spatial transcriptomics sequencing is that it cannot determine which RNA fragments captured by the probes come from the same cell, which limits the ability to analyze cell-to-cell specificity. To break this limitation, the data obtained from spatial transcriptomics sequencing often needs to be subjected to subsequent cell segmentation processing, that is, through data analysis methods, to identify which base sequences come from different cells. The mainstream cell segmentation methods are based on images, and there are mainly two sources of images: 1) The RNA signal distribution map formed by the signal intensity (probe density capturing RNA fragments) captured by the sequencing chip; 2) The staining signal distribution map obtained by separately staining the tissue sections used for sequencing. The characteristic of both types of pictures is that there are strong signal peaks at the corresponding positions of the cell nuclei in the tissue sections, which appear as isolated bright spots (or spots with a large difference from the background) in the pictures. As Figure 1 shown, cell segmentation is to identify the range of complete spots in the image and regard the RNA base sequences captured by the probes corresponding to this range as coming from the same cell, so as to integrate the sequenced base sequences on a cell-by-cell basis for subsequent analysis.

[0024] The difficulty in identifying bright spots (or other spots) based on images is that bright spots do not always remain isolated and complete. Depending on the tissue section conditions and experimental conditions, bright spots may appear blurred, crowded, with unclear boundaries, etc., or multiple different situations may occur in the same image, as Figure 2 shown, the RNA signal distribution states at different positions on the same tissue section are different.

[0025] In the prior art, the methods for identifying bright spots in images to segment cells mainly include: 1) the watershed algorithm; 2) clustering-based methods; 3) deep learning-based methods. The watershed algorithm is a classic algorithm used in the field of image segmentation, which can accurately and quickly identify the range of bright spots. However, it requires smooth signal transitions and obvious boundaries, and these two conditions are often not met in the signal distribution images of sample sections. Especially for RNA signal distribution maps, in the prior art, methods based on improving the watershed algorithm, such as preprocessing the image, etc., have not explored adaptive parameters and failed to adapt to variable sample conditions. The difficulty in the requirement for continuous signals lies in that the position of RNA signals comes from the position of probes, and the positions of probes are not continuous, resulting in isolated signal points rather than complete bright spots. The difficulty in the requirement for clear boundaries lies in that in the area of dense cells, the RNA signals of different cells are mixed, making it difficult to form obvious boundaries. In order to make the signals continuous, in the prior art, the signals are often smoothed to suppress random fluctuations and isolated signals. However, the smoothing process itself will further blur the boundaries, making it impossible to identify individual cells. Spatial transcriptome sequencing is a sequencing method that can read the base sequence information of RNA fragments while recording the spatial position of the fragments. Currently, the spatial information that can be recorded by spatial omics sequencing is generally limited to two-dimensional space, which is achieved by covering tissue sections on a sequencing chip. Probes that can capture RNA fragments are distributed on the surface of the sequencing chip, and the sequencer can read the base sequence information of the RNA fragments bound to the probes. Combining the known probe positions, the spatial information of the base sequence can be obtained.

[0026] One limitation of current spatial omics sequencing is that it cannot determine which RNA fragments captured by probes come from the same cell, which limits the ability to analyze cell-to-cell specificity. In the area of dense cells, the RNA signals of different cells are mixed, making it difficult to form obvious boundaries. Therefore, the bright spots in the RNA signal distribution map do not always remain isolated and complete. Depending on the tissue section and experimental conditions, the bright spots may be blurred, crowded, or have unclear boundaries. Therefore, the methods in the prior art cannot accurately segment cells.

[0027] A cell segmentation method based on spatial omics sequencing provided by an embodiment of the present application can adaptively adjust local parameters according to local conditions of the image to improve the recognition accuracy. The embodiment of the present application is based on the improvement of the watershed algorithm to make it more suitable for cell segmentation; automatically adjusts parameters, increases the degree of smoothing in sparse cell areas to suppress isolated random noise; in dense cell areas, reduces the degree of smoothing to retain the boundary morphology.

[0028] As Figure 3 shown, Figure 3It is an application environment diagram of a cell segmentation method based on spatial omics sequencing in an embodiment; the gene sequencer 10 communicates with the computer device 20, which is achieved by covering a tissue section on the sequencing chip of the gene sequencer 10. Probes capable of capturing RNA fragments are distributed on the surface of the sequencing chip of the gene sequencer 10. The gene sequencer 10 can read the base sequence information of the RNA fragments bound to the probes. Combining with the known probe positions, the spatial information of the base sequences can be obtained. The RNA signal distribution diagram is composed of the signal intensity distribution (the probe density of captured RNA fragments) captured by the sequencing chip in the gene sequencer 10. The computer device 20 acquires the RNA signal distribution image for processing and segments the cells. In an alternative implementation, the cell segmentation method based on spatial omics sequencing can also be implemented in the gene sequencer 10.

[0029] Please refer to Figure 4 , which is a flowchart of a cell segmentation method based on spatial omics sequencing provided by an embodiment of the present application. The cell segmentation method based on spatial omics sequencing is applied to a computer device. The cell segmentation method based on spatial omics sequencing includes the following steps:

[0030] S11. Obtain the collected signal distribution image, where the signal distribution image indicates the distribution characteristics of the spatial omics signals of the tissue section.

[0031] In this embodiment, the signal distribution image refers to an image that can reflect the distribution characteristics of the spatial omics signals of the tissue section in physical space. It can be an RNA signal distribution image captured by a sequencing chip or an image taken after staining the genetic material of the section. As Figure 1 and Figure 2 shown, there are strong signal peaks at the corresponding positions of the cell nuclei in the tissue section, which appear as an isolated bright spot (or a spot with a large difference from the background) in the picture. Subsequently, by identifying the range of the complete spots on the image and regarding the RNA base sequences captured by the probes corresponding to the range of the complete spots as originating from the same cell, the sequenced base sequences are integrated by cell unit. Among them, spatial omics can be understood as spatial multi-omics, and spatial omics includes but is not limited to one or more of the following combinations: transcriptomics, genomics, proteomics, metabolomics, lipidomics, glycomics, epigenomics, immunomics, microbiomics, metagenomics, pharmacomics, etc. Spatial omics signals include but are not limited to one or more of the following combinations: transcriptomics signals, genomics signals, proteomics signals, metabolomics signals, lipidomics signals, glycomics signals, epigenomics signals, immunomics signals, microbiomics signals, metagenomics signals, pharmacomics signals, etc.

[0032] S12. Based on the signal distribution image, perform an initial smoothing operation to obtain an initial smoothed distribution image. Based on the initial smoothed distribution image, perform an image segmentation operation to obtain the current set of cell segmentation regions.

[0033] In this embodiment, using a smoothing function, based on the signal distribution image, perform an initial smoothing operation to obtain an initial smoothed distribution image, where the initial smoothed distribution image is the image obtained after performing the first smoothing operation based on the signal distribution image. The smoothing function includes but is not limited to the Gaussian function, and the formula of the Gaussian function is as follows:

[0034]

[0035] The Gaussian function curve is a dotted line similar to an inverted V shape, with a convex middle and a smooth transition, and is relatively gentle on both sides. The parameter a is the magnification factor, which changes the height of the curve. For example, it can be set to 1; the parameter b is the horizontal position offset, which changes the horizontal position of the curve, and is set to 0 here; the smoothing parameter c affects the shape of the curve. The larger the value of the smoothing parameter c, the flatter the curve, and the smaller the value, the sharper the curve; e is the natural logarithm base. When using the Gaussian function for smoothing, the degree to which the pixel value at a position is affected by the surrounding pixel values is related to the shape of the Gaussian curve. When the smoothing parameter c is large, the curve is flat and the decline on both sides is slow, which means that the influence of this pixel by the surrounding pixels decreases slowly with the increase of the distance; on the contrary, the curve is sharp and the decline on both sides is rapid, which means that with the increase of the distance, the influence of pixels farther away on this pixel decreases sharply.

[0036] Through the smoothing operation, for each pixel on the image, according to the result of performing a certain operation based on the values of the neighboring pixels around this pixel, change the original value of this pixel. A typical smoothing process is Gaussian blur. The calculation formula for the value after performing Gaussian blur on the point with coordinates (x, y) in a two-dimensional space is as follows:

[0037]

[0038] where width is the length of the image, height is the width of the image, is the signal intensity of the pixel at coordinates (i, j), a is a constant, e is the constant natural logarithm base, c is the smoothing parameter, is the smoothed distribution image.

[0039] In an alternative implementation, the image segmentation operation can use the watershed algorithm to identify the signal peaks on the smoothed distribution image after the smoothing operation, such as Figure 1 and Figure 2 the bright spots and the distribution range of the bright spots shown, and extract the signal distribution within the range of each bright spot.

[0040] In an alternative implementation, for a signal distribution image containing multiple distribution conditions, it is also possible to pre-divide regions of different conditions, that is, to pre-segment to obtain the current cell segmentation region set after the first segmentation.

[0041] S13. Based on the current cell segmentation region set, screen to obtain the cell region images to be further segmented to obtain the current first region set, and screen the cell region images that are not to be further segmented to obtain the current second region set.

[0042] In this embodiment, the current first region set stores the cell region images that need to be further segmented. The cell region images that need to be further segmented indicate that there are multiple bright spots in the cell region images and need to be further segmented. The current second region set stores the cell region images that do not need to be further segmented. That is, a cell region image in the current second region set is the distribution range of the region where a bright spot is located.

[0043] S14. Adjust the smoothing parameter corresponding to the smoothing operation. Based on the adjusted smoothing parameter, perform a smoothing operation and an image segmentation operation on the cell region images in the current first region set, and update the current cell segmentation region set, then return to the previous screening step until the termination condition is met to obtain the target second region set.

[0044] Perform a smoothing operation and an image segmentation operation on the cell region images to be further segmented again, and update the current cell segmentation region set, that is, return to the previous screening step, that is, return to execute S13, and then execute S14, and repeat the execution in a loop until the termination condition is met. The termination condition includes but is not limited to that the current cell segmentation region set is empty, and the number of times of performing the smoothing operation and the image segmentation operation is greater than the preset number of times. The current second region set after the termination condition is met is used as the target second region set.

[0045] As Figure 5 shown, Figure 5 is a schematic diagram of performing multiple smoothing operations and image segmentation operations in an embodiment. The dots represent signal points whose signal intensity values exceed the preset signal intensity value. The solid circles represent the bright spots recognized in the initial segmentation, that is, the signal dense regions. The dashed circles represent the bright spot regions recognized in the second segmentation. After performing the second smoothing operation and image segmentation operation, the larger cell region images can be further finely segmented, such as Figure 5 the solid line region where the two dashed line regions are located in

[0046] In this embodiment, since the smoothing parameter corresponding to the initial smoothing operation is more suitable for the cell sparse region but not for the cell dense region, if the smoothing parameter corresponding to the initial smoothing operation is used, the latter cannot be effectively segmented, and multiple cells are recognized as a whole, such as Figure 5The large solid circle at the lower right corner as shown. To further separate this part of the cells in the large solid circle, it is necessary to adjust the smoothing parameter to re-smooth and segment the area where the dashed box is located. By adjusting the smoothing parameter during multiple smoothing operations, the cell area image in the dense area can be further subdivided.

[0047] S15. Based on the target second region set, obtain each cell region.

[0048] In this embodiment, an image of a cell region in the target second region set is the distribution range of the area where a bright spot is located. As Figure 6 shown, Figure 6 is a schematic diagram of cells segmented by a cell segmentation method based on spatial transcriptome sequencing in an embodiment, Figure 6 the area enclosed by the white curve in is regarded as the coverage range of a single cell.

[0049] In the above embodiment, obtain the acquired signal distribution image. Based on the initial signal distribution image, perform an initial smoothing operation to obtain an initial smoothed distribution image. Based on the initial smoothed distribution image, perform an image segmentation operation to obtain the current cell segmentation region set. Based on the current cell segmentation region set, screen the cell region images to be further segmented to obtain the current first region set. Screen the cell region images that are not to be further segmented to obtain the current second region set. Perform the smoothing operation and the image segmentation operation again on the cell region images to be further segmented until the termination condition is met to obtain the target second region set, thereby obtaining each cell region. By adjusting the smoothing parameter during multiple smoothing operations to make the adjusted smoothing parameter more suitable for the current smoothing operation, the cell region image in the dense area is further subdivided to obtain a single cell region image, achieving better segmentation, and thus improving the accuracy of cell segmentation.

[0050] In some embodiments, the performing a smoothing operation based on the signal distribution image to obtain an initial smoothed distribution image includes:

[0051] Obtain a filtering threshold, and based on the filtering threshold, filter the signal distribution image to obtain a filtered distribution image;

[0052] Based on the filtered distribution image, perform an initial smoothing operation to obtain an initial smoothed distribution image.

[0053] Optionally, the obtaining a filtering threshold and filtering the signal distribution image based on the filtering threshold to obtain a filtered distribution image includes at least one of the following:

[0054] Obtain a preset filtering threshold, and set the signal intensity of the pixel points in the signal distribution image that are less than the filtering threshold to a preset intensity value to obtain the filtered distribution image;

[0055] Calculate the filtering threshold based on the signal intensity corresponding to the pixel points in the signal distribution image, and set the signal intensity of the pixel points in the signal distribution image that are less than the filtering threshold to a preset intensity value to obtain the filtered distribution image;

[0056] Obtain the signal intensity values of the neighboring pixel points in the neighboring regions of each pixel point in the signal distribution image. Based on the signal intensity values of the respective neighboring pixel points corresponding to each pixel point, calculate the average signal intensity value corresponding to each pixel point. Based on the average signal intensity value corresponding to each pixel point, determine the filtering threshold corresponding to each pixel point, and set the signal intensity of the pixel points that are less than the filtering threshold corresponding to the pixel point to a preset intensity value to obtain the filtered distribution image.

[0057] In this embodiment, based on a given preset filtering threshold, if the signal intensity at a certain position in the signal distribution image is lower than the preset filtering threshold, its intensity is set to 0. If the signal intensity at a certain position in the signal distribution image is greater than or equal to the preset filtering threshold, the signal intensity at that position remains unchanged. For pixel points at different positions, their thresholds can be the same or different. When setting a filtering threshold for a pixel point at a position, the average signal intensity of the signal intensities corresponding to the pixel points within a circular region around the pixel point can be calculated, and a preset ratio of the average signal intensity is used as the filtering threshold for the pixel point. In an alternative implementation, the radius of the circular region is preferably 10 times the preset radius, and the preset ratio is preferably 100%. For the preset filtering threshold, preferably 100% of the average signal intensity in the signal distribution image. When necessary, the preset filtering threshold can be set to 0, that is, no filtering is performed.

[0058] In the above embodiment, before performing the smoothing operation, the signal distribution image is first filtered, so that noise can be effectively filtered, the image quality before segmentation can be improved, and thus the accuracy of cell segmentation can be improved.

[0059] In some embodiments, the smoothing parameters corresponding to the adjustment of the smoothing operation include:

[0060] Calculate the smoothing parameter value for each smoothing operation, where the smoothing parameter value corresponding to the current smoothing operation currently being executed is less than the smoothing parameter value corresponding to the previous smoothing operation of the current smoothing operation.

[0061] Optionally, the smoothing parameter value corresponding to the current smoothing operation is between 1 / 10 of the smoothing parameter value corresponding to the previous smoothing operation and 9 / 10 of the smoothing parameter value corresponding to the previous smoothing operation.

[0062] In this embodiment, a hierarchical parameter adjustment strategy is adopted. Specifically, in the initial smoothing operation, a relatively high smoothing parameter is used, which will result in relatively ideal bright spots in areas where the tissue section cells are sparse, and single-cell regions are identified. In areas where cells are dense, due to the use of a relatively high smoothing parameter, the boundaries of the bright spots will be more difficult to identify, resulting in multiple cells being recognized as a single overall region. Therefore, before the secondary segmentation, the cell-dense regions will be re-smoothed with a reduced smoothing parameter. That is, an automatic parameter adjustment is adopted to increase the smoothing degree in cell-sparse regions to suppress isolated random noise; in cell-dense regions, the smoothing degree is reduced to preserve the boundary morphology. In an alternative implementation, the value of the smoothing parameter corresponding to the current smoothing operation is between 1 / 4 of the value of the smoothing parameter corresponding to the previous smoothing operation and 3 / 4 of the value of the smoothing parameter corresponding to the previous smoothing operation. Generally, when reducing the smoothing parameter, the value of the smoothing parameter will be reduced to a certain proportion of the value of the smoothing parameter corresponding to the previous smoothing operation. The regions obtained from the re-segmentation will continue to be screened according to a preset screening upper limit and a preset screening lower limit, and the cell region images to be further segmented are selected for three or more times of segmentation until the finally obtained indicators meet the expectations or the number of segmentation times reaches the preset value.

[0063] In the above embodiment, a hierarchical parameter adjustment strategy is adopted. During the process of performing multiple smoothing operations, the smoothing parameter is gradually reduced to re-smooth the cell-dense regions, so that the smoothing parameter is more suitable for the segmentation of single-cell regions in cell-dense regions. The smoothing degree is increased in cell-sparse regions to suppress isolated random noise; in cell-dense regions, the smoothing degree is reduced to preserve the boundary morphology, thereby retaining more details of the cell-dense regions, so that the image segmentation operation can identify the cell boundaries therein, thereby improving the accuracy of cell segmentation.

[0064] In some embodiments, screening the cell region images to be further segmented based on the current cell segmentation region set to obtain the current first region set, and screening the cell region images not to be further segmented to obtain the current second region set includes:

[0065] Putting the cell region images in the current cell segmentation region set that are higher than the preset screening upper limit into the current first region set;

[0066] Putting the cell region images in the current cell segmentation region set that are lower than the preset screening upper limit and higher than the preset screening lower limit into the current second region set.

[0067] In this embodiment, the preset screening upper limit includes, but is not limited to, the upper limit of the preset cell area and the upper limit of the total number of preset cellular total RNAs. The preset screening lower limit includes, but is not limited to, the lower limit of the preset cell area and the lower limit of the total number of preset cellular total RNAs. Different preset screening upper limits and preset screening lower limits correspond to different cell types. Parameter setting controls can be provided through the user interface, and options for the preset screening upper limits and preset screening lower limits corresponding to different cell types can be set through the parameter setting controls, or the user can customize and input the preset screening upper limits and preset screening lower limits. Based on the user interface, a trigger operation on the parameter setting control is obtained, and the target upper limit and target lower limit corresponding to the trigger operation are obtained. The target upper limit is used as the preset screening upper limit, and the target lower limit is used as the preset screening lower limit. Therefore, the preset screening upper limit and preset screening lower limit can be adjusted according to the cell category, thereby improving the accuracy of cell segmentation.

[0068] The cell region image that is lower than or equal to the preset screening upper limit and higher than or equal to the preset screening lower limit is the image that conforms to the single cell region image, and this part is the region that will no longer be segmented. The cell region image that is higher than the preset screening upper limit is the cell dense region image, which is the cell region image that needs to be continuously segmented. The cell region image that is lower than the preset screening lower limit is the noise region image, and this part of the image is discarded and no further processing is performed.

[0069] In the above embodiment, the single cell regions and the dense cell regions that need to be continuously segmented in the current cell segmentation region set are screened through the preset screening upper limit and preset screening lower limit, so as to continuously segment the dense cell regions until single cell regions are obtained, thereby improving the accuracy of cell segmentation.

[0070] In some embodiments, performing an initial smoothing operation on the signal distribution image to obtain an initial smoothed distribution image includes:

[0071] Performing an initial smoothing operation on the signal distribution image and the initial smoothing parameter value to obtain an initial smoothed distribution image, where the initial smoothing parameter value is the minimum value that satisfies the inequality, and the inequality is:

[0072]

[0073] where represents the average cell radius, represents the smoothing parameter.

[0074] In this embodiment, the inequality is derived based on the Gaussian function, and the average cell radius can be configured on the user interface. Since different cell types have different average cell radii, different average cell radii corresponding to different cell types can be configured on the user interface, so as to obtain a smoothing parameter suitable for this cell type. In this way, during the smoothing operation, appropriate smoothing processing is applied to the conditions of different regions on the signal distribution image respectively to ensure that the image segmentation can identify the bright spots on the image as accurately as possible.

[0075] In the above embodiment, in combination with the cell type, the average cell radius suitable for the cell type is obtained, so as to obtain a more accurate initial smoothing parameter. In this way, during the smoothing operation, appropriate smoothing processing is applied to the conditions of different regions on the signal distribution image respectively to ensure that the image segmentation can identify the bright spots on the image as accurately as possible.

[0076] In some embodiments, obtaining each cell region based on the target second region set includes:

[0077] Obtain a plurality of target region images in the target second region set; for any one of the target region images, obtain each probe corresponding to each pixel point in the target region image, and label the probes corresponding to each pixel point in the target region image as the same cell.

[0078] In this embodiment, a probe is a small single-stranded DNA or RNA fragment (about 20 to 500 bp) used to detect its complementary nucleic acid sequence. The probe is used in cell or tissue samples to visualize the location and distribution of RNA by complementary pairing with the target RNA sequence. These probes are usually labeled with multiple fluorescent dyes so that the spatial distribution of RNA can be captured by an image acquisition device. In image analysis, each pixel point represents a specific position in the image, and the signal intensity corresponding to each pixel point can reflect the RNA distribution at that position. In spatial omics sequencing, these probes are designed to specifically identify and capture target mRNAs. In the signal distribution image, each position corresponds to a pixel point, and each pixel point within the same bright spot region corresponds to at least one probe position, and the corresponding at least one probe position is the same cell. The distribution region of the same bright spot can represent the same cell region. When the tissue section is covered on the sequencing chip, the probe positions corresponding to the positions where cells exist on the section correspond to the positions where the bright spots are located in the signal distribution image. There are no bright spots in the signal distribution image corresponding to the probe positions where there are no cells on the section. Since the target second region set contains single cell region images, the probe positions corresponding to each pixel point in a single cell region image are the same cell region.

[0079] In the above embodiments, after obtaining a single-cell region by segmentation, the probe positions are obtained according to the positions of the pixel points in the single-cell region, so that the probe positions corresponding to the pixel points in the image of the same cell region are marked as the same cell, thereby improving the accuracy of cell segmentation.

[0080] As Figure 7 shown, Figure 7 FIG. is a flowchart of a cell segmentation method based on spatial omics sequencing in another embodiment, and the flowchart includes:

[0081] S71. Obtain the acquired signal distribution image.

[0082] S72. Perform signal filtering on the signal distribution image to obtain a filtered distribution image.

[0083] S73. Perform a smoothing operation on the filtered distribution image to obtain an initial smoothed distribution image, and based on the initial smoothed distribution image, perform an image segmentation operation to obtain a current cell segmentation region set.

[0084] S74. Based on the current cell segmentation region set, screen out the cell region images to be further segmented to obtain a current first region set, and screen out the cell region images that are not to be further segmented to obtain a current second region set.

[0085] In this embodiment, according to a preset screening upper limit and a preset screening lower limit, S 0 the elements (regions) are screened. Discard the regions with any index lower than the preset screening lower limit, and transfer the regions with any index higher than the preset screening upper limit to another set S 1 , that is, the current first region set. At this time, S 0 will only retain the regions with all indexes between the preset screening upper limit and the preset screening lower limit, and transfer the S 0 elements to another set S , that is, the current second region set, and clear S 0 .

[0086] S75. Adjust the smoothing parameter corresponding to the smoothing operation, and based on the adjusted smoothing parameter, perform a smoothing operation and an image segmentation operation on the cell region images in the current first region set to obtain a current segmentation result, and update the current cell segmentation region set based on the current segmentation result.

[0087] For each element (region) in S 1 obtained in S74, the signal distribution of each region is re-smoothed. At this time, the smoothing parameterc is set to a certain proportion of the smoothing parameter corresponding to the previous smoothing operation, and the preferred range of the proportion is from 1 / 10 to 9 / 10, more preferably from 1 / 4 to 3 / 4, and the most preferred value is 1 / 2. After the smoothing process is completed, perform an image segmentation operation on each region separately, and store the recognized regions into S 0 . Clear S 1 .

[0088] S76. Determine whether the current condition meets the termination condition.

[0089] If the current condition does not meet the termination condition, return to execute S74. If the current condition meets the termination condition, that is, until S 1 contains no elements or the repetition times reach the specified upper limit, execute S77.

[0090] S77. Obtain each cell region based on the target second region set.

[0091] On the other hand, the present application provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the cell segmentation method based on spatial omics sequencing described in any embodiment of the present application.

[0092] Among them, in the computer program product, an optional implementation form of the program module architecture of the computer program for implementing each step of the cell segmentation method based on spatial omics sequencing can be a cell segmentation device based on spatial omics sequencing.

[0093] Please refer to Figure 8 , an embodiment of the present application provides a cell segmentation device based on spatial omics sequencing, including: an acquisition module 81, configured to acquire a collected signal distribution image, where the signal distribution image indicates the distribution characteristics of the spatial omics signals of a tissue section; a smoothing and segmentation module 82, configured to perform an initial smoothing operation based on the signal distribution image to obtain an initial smoothed distribution image, and perform an image segmentation operation based on the initial smoothed distribution image to obtain a current cell segmentation region set; a screening module 83, configured to screen, based on the current cell segmentation region set, cell region images to be continuously segmented to obtain a current first region set, and screen cell region images that are not continuously segmented to obtain a current second region set; the smoothing and segmentation module 82 is further configured to adjust the smoothing parameter corresponding to the smoothing operation, and perform a smoothing operation and an image segmentation operation on the cell region images in the current first region set based on the adjusted smoothing parameter, and update the current cell segmentation region set, and return to the previous screening step until the termination condition is met to obtain a target second region set; a determination module 84, configured to obtain each cell region based on the target second region set.

[0094] Optionally, the smoothing and segmentation module 82 is further configured to:

[0095] Obtain a filtering threshold, and based on the filtering threshold, filter the signal distribution image to obtain a filtered distribution image;

[0096] Based on the filtered distribution image, perform an initial smoothing operation to obtain an initial smoothed distribution image.

[0097] Optionally, the smoothing and segmentation module 82 is further configured to:

[0098] Obtain a preset filtering threshold, and set the signal intensity of the pixel points in the signal distribution image that are less than the filtering threshold to a preset intensity value to obtain the filtered distribution image;

[0099] Calculate the filtering threshold based on the signal intensity corresponding to the pixel points in the signal distribution image, and set the signal intensity of the pixel points in the signal distribution image that are less than the filtering threshold to a preset intensity value to obtain the filtered distribution image;

[0100] Obtain the signal intensity values of the neighboring pixel points in the neighboring regions of each pixel point in the signal distribution image, calculate the average signal intensity value corresponding to each pixel point based on the signal intensity values of the respective neighboring pixel points corresponding to each pixel point, determine the filtering threshold corresponding to each pixel point based on the average signal intensity value corresponding to each pixel point, and set the signal intensity of the pixel points that are less than the filtering threshold corresponding to the pixel point to a preset intensity value to obtain the filtered distribution image.

[0101] Optionally, the smoothing and segmentation module 82 is further configured to:

[0102] Calculate the smoothing parameter value for each smoothing operation, where the smoothing parameter value corresponding to the current smoothing operation currently being performed is less than the smoothing parameter value corresponding to the previous smoothing operation of the current smoothing operation.

[0103] Optionally, the smoothing parameter value corresponding to the current smoothing operation is between 1 / 10 of the smoothing parameter value corresponding to the previous smoothing operation and 9 / 10 of the smoothing parameter value corresponding to the previous smoothing operation.

[0104] Optionally, the smoothing and segmentation module 82 is further configured to:

[0105] Put the cell region images in the current cell segmentation region set that are higher than the preset screening upper limit into the current first region set;

[0106] Put the cell region images in the current cell segmentation region set that are lower than the preset screening upper limit and higher than the preset screening lower limit into the current second region set.

[0107] Optionally, the smoothing and segmentation module 82 is further configured to:

[0108] Based on the signal distribution image and the initial smoothing parameter value, perform an initial smoothing operation to obtain an initial smoothed distribution image, where the initial smoothing parameter value is the minimum value that satisfies the inequality:

[0109]

[0110] where represents the average cell radius, represents the smoothing parameter, is the natural base of the exponential function.

[0111] Optionally, the determination module 84 is further configured to:

[0112] Obtain a plurality of target region images in the target second region set; for any one of the target region images, obtain each probe corresponding to each pixel point in the target region image, and label each probe corresponding to each pixel point in the target region image as the same cell.

[0113] Those skilled in the art can understand that Figure 8 the structure of the cell segmentation device based on spatial omics sequencing in

[0114] does not limit the cell segmentation device based on spatial omics sequencing, and each of the modules can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in or independent of the controller in the computing device in hardware form, or can be stored in the memory of the computing device in software form, so that the controller can call and execute the operations corresponding to each of the above modules. In other embodiments, the cell segmentation device based on spatial omics sequencing may include more or fewer modules than shown in the figure.

[0114] Please refer to Figure 9 , on the other hand, an embodiment of the present application further provides a computing device 20, including a memory 3011 and a processor 3012. The memory 3011 stores a computer program. When the computer program is executed by the processor, the processor 3012 executes the steps of the cell segmentation method based on spatial omics sequencing provided in any one of the above embodiments of the present application. The computing device may include a desktop computer, a laptop computer, a tablet computer, a handheld computer, a smart speaker, a server, etc., a mobile phone (for example, a smart phone, a wireless phone, etc.), a wearable device (for example, a pair of smart glasses or a smart watch), or a similar device.

[0115] Among them, the processor 3012 is the control center, which connects all parts of the entire computing device through various interfaces and lines. By running or executing software programs and / or modules stored in the memory 3011, and by calling the data stored in the memory 3011, it executes various functions of the computing device and processes data. Optionally, the processor 3012 may include one or more processing cores; the processor 3012 includes, but is not limited to, one or more of the following combinations: Central Processing Unit (CPU), Graphics Processing Unit (GPU), Field-Programmable Gate Array (FPGA), etc. Preferably, the processor 3012 may integrate an application processor and a modem processor. Among them, the application processor mainly processes the operating system, user interfaces, and application programs, etc., and the modem processor mainly processes wireless communication. It can be understood that the above-mentioned modem processor may not be integrated into the processor 3012 either.

[0116] The memory 3011 can be used to store software programs and modules. The processor 3012 executes various functional applications and data processing by running the software programs and modules stored in the memory 3011. The memory 3011 mainly includes a program storage area and a data storage area. Among them, the program storage area can store the operating system, application programs required for at least one function (such as the sound playback function, the image playback function, etc.); the data storage area can store the data created according to the use of the computing device. In addition, the memory 3011 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices. Correspondingly, the memory 3011 may also include a memory controller to provide the processor 3012 with access to the memory 3011.

[0117] In another aspect of the embodiments of the present application, there is also provided a storage medium storing a computer program, which, when executed by a processor, causes the processor to execute the steps of the cell segmentation method based on spatial omics sequencing provided in any one of the above embodiments of the present application.

[0118] Those of ordinary skill in the art can understand that all or part of the processes in the methods provided in the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0119] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. The protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A cell segmentation method based on spatial omics sequencing, characterized in that: include: Acquiring a collected signal distribution image, wherein the signal distribution image indicates distribution characteristics of spatial omics signals of the tissue section; Based on the signal distribution image and the initial smoothing parameter value, an initial smoothing operation is performed to obtain an initial smoothed distribution image, and based on the initial smoothed distribution image, an image segmentation operation is performed to obtain a current cell segmentation region set, wherein the initial smoothing parameter value is a minimum value that satisfies the inequality, wherein the inequality is: in represents the average cell radius, represents the smoothing parameter, is the natural base of the exponential function; Based on the current cell segmentation region set, filter out the cell region images to be further segmented to obtain the current first region set, and filter out the cell region images not to be further segmented to obtain the current second region set; Adjusting the smoothing parameters corresponding to the smoothing operation, performing the smoothing operation and the image segmentation operation on the cell region image in the current first region set based on the adjusted smoothing parameters, and updating the current cell segmentation region set, returning to the previous screening step until the termination condition is met, and obtaining the target second region set, wherein the smoothing parameters corresponding to the smoothing operation include: calculating the smoothing parameter value of each smoothing operation, wherein the smoothing parameter value corresponding to the current smoothing operation currently being performed is less than the smoothing parameter value corresponding to the previous smoothing operation of the current smoothing operation; Based on the target second region set, each cell region is obtained.

2. The cell segmentation method based on spatial omics sequencing according to claim 1, characterized in that: The performing a smoothing operation based on the signal distribution image to obtain an initial smoothed distribution image comprises: Acquire a filtering threshold, and filter the signal distribution image based on the filtering threshold to obtain a filtered distribution image; Based on the filtered distribution image, an initial smoothing operation is performed to obtain an initial smoothed distribution image.

3. The cell segmentation method based on spatial omics sequencing according to claim 2, characterized in that: The obtaining of the filtering threshold, filtering the signal distribution image based on the filtering threshold to obtain the filtered distribution image comprises at least one of the following: Obtaining a preset filtering threshold, setting the signal intensity of the pixel points in the signal distribution image that are less than the filtering threshold to a preset intensity value, and obtaining the filtered distribution image; Calculating the filtering threshold based on the signal strength corresponding to the pixel points in the signal distribution image, setting the signal strength of the pixel points in the signal distribution image that are less than the filtering threshold to a preset strength value, and obtaining the filtered distribution image; Obtain the signal strength value of each neighboring pixel point in the neighborhood of each pixel point in the signal distribution image, calculate the average signal strength value corresponding to each pixel point based on the signal strength value of each neighboring pixel point corresponding to each pixel point, determine the filtering threshold corresponding to each pixel point based on the average signal strength value corresponding to each pixel point, set the signal strength of the pixel point that is less than the filtering threshold corresponding to the pixel point to a preset strength value, and obtain the filtered distribution image.

4. The cell segmentation method based on spatial omics sequencing according to claim 1, characterized in that: The smoothing parameter value corresponding to the current smoothing operation is between 1 / 10 and 9 / 10 of the smoothing parameter value corresponding to the previous smoothing operation.

5. The cell segmentation method based on spatial omics sequencing according to claim 1, characterized in that: The method of screening the cell region images to be further segmented based on the current cell segmentation region set to obtain the current first region set, and screening the cell region images not to be further segmented to obtain the current second region set comprises: Putting cell region images above a preset screening upper limit in the current cell segmentation region set into the current first region set; The cell region images in the current cell segmentation region set that are lower than a preset screening upper limit and higher than a preset screening lower limit are placed into the current second region set.

6. The cell segmentation method based on spatial omics sequencing according to claim 1, characterized in that: The obtaining of each cell region based on the target second region set includes: Acquire multiple target area images in the target second area set; for any of the target area images, acquire each probe corresponding to each pixel point in the target area image, and mark each probe corresponding to each pixel point in the target area image as the same cell.

7. A computing device, characterized in that The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 6.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 6.

9. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images