Image batch processing method and system

Abstract

The invention provides an image batch processing method and system, and relates to the technical field of image processing. The image processing plug-in selects a small number of to-be-processed images from a plurality of to-be-processed images corresponding to a target experiment as parameter adjusting images to be loaded into the image processing plug-in based on experiment configuration information set by a user on the cloud platform for the target experiment, and generates processing result previews of the parameter adjusting images under different image processing parameters; uploading image processing standard parameters confirmed by the user based on the processing result preview to the cloud platform; and the cloud platform performs batch processing on the plurality of to-be-processed images according to the acquired image processing standard parameters to obtain a batch processing result. In the image processing plug-in, only a small number of images are needed for parameter adjustment, and image preview effects under different parameters can be provided, so that unified parameters are determined in a WYSIWYG mode and uploaded to a cloud platform for multi-image batch processing, and the processing efficiency of large-batch images and the reliability of experimental results are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to an image batch processing method and system. BACKGROUND

[0002] In the field of biomedical image analysis, a variety of image processing tools (such as ImageJ, MATLAB, Fiji, CellProfiler, etc.) are widely used in cell detection, tissue segmentation, fluorescence signal recognition, etc. tasks, and have functions of image enhancement, target recognition, result statistics, etc.

[0003] However, these image processing tools have weak batch processing image capability, and can only process each image individually, which is low in processing efficiency. At the same time, due to the influence of human operation, the parameters set for different images of the same batch may be different, it is difficult to achieve unified analysis standard, and the reliability of experimental results is affected. SUMMARY

[0004] Therefore, the purpose of the present application is to provide an image batch processing method and system to improve the processing efficiency of a large number of images and improve the reliability of experimental results, and to cover the capabilities of visualization and comparative analysis.

[0005] In order to achieve the above purpose, the technical solutions adopted by the embodiments of the present application are as follows: In a first aspect, the present application provides an image batch processing method applied to an image processing plug-in and a cloud platform in communication with each other; the method comprises: The image processing plug-in selects a parameter adjustment image from a plurality of to-be-processed images corresponding to a target experiment and loads the parameter adjustment image into the image processing plug-in based on experiment configuration information set by a user on the cloud platform for the target experiment; the number of the parameter adjustment images is less than or equal to a preset value; The image processing plug-in generates a processing result preview image of the parameter adjustment image under different image processing parameters, and uploads image processing standard parameters confirmed by the user based on the processing result preview image to the cloud platform; The cloud platform performs batch processing on a plurality of to-be-processed images according to the obtained image processing standard parameters, and obtains a batch processing result.

[0006] In an optional implementation, the image processing plug-in generates a processing result preview image of the parameter adjustment image under different image processing parameters, and uploads image processing standard parameters confirmed by the user based on the processing result preview image to the cloud platform, comprising: The image processing plug-in generates preview images of processing results of the parameter-adjusting image under different channel separation parameters according to channel separation parameters set by a user, and uploads channel separation parameters corresponding to a processing result preview image selected by the user as image processing standard parameters to the cloud platform.

[0007] In an optional implementation, the image processing plug-in generates preview images of processing results of the parameter-adjusting image under different image processing parameters, and uploads image processing standard parameters confirmed by the user based on the preview images of processing results to the cloud platform, including: The image processing plug-in generates preview images of processing results of the parameter-adjusting image under different intensity threshold parameters according to intensity threshold parameters set by a user, and uploads intensity threshold parameters corresponding to a processing result preview image selected by the user as image processing standard parameters to the cloud platform.

[0008] In an optional implementation, the image processing plug-in generates preview images of processing results of the parameter-adjusting image under different image processing parameters, and uploads image processing standard parameters confirmed by the user based on the preview images of processing results to the cloud platform, including: The image processing plug-in generates preview images of processing results of the parameter-adjusting image under different HSB threshold parameters according to HSB threshold parameters set by a user, and uploads HSB threshold parameters corresponding to a processing result preview image selected by the user as image processing standard parameters to the cloud platform; the HSB threshold parameters include hue threshold parameters, saturation threshold parameters, and brightness threshold parameters.

[0009] In an optional implementation, the image processing plug-in generates preview images of processing results of the parameter-adjusting image under different image processing parameters, and uploads image processing standard parameters confirmed by the user based on the preview images of processing results to the cloud platform, including: The image processing plug-in generates preview images of processing results of the parameter-adjusting image under different cell segmentation parameters according to cell segmentation parameters set by a user, and uploads cell segmentation parameters corresponding to a processing result preview image selected by the user as image processing standard parameters to the cloud platform.

[0010] In an optional implementation, the cloud platform performs batch processing on a plurality of the to-be-processed images according to the obtained image processing standard parameters to obtain batch processing results, including: The cloud platform performs target detection on each of the to-be-processed images according to the obtained image processing standard parameters and recognition target configuration information set on the cloud platform, and performs index calculation according to a target detection result and index configuration information set on the cloud platform to obtain batch processing results.

[0011] In an optional embodiment, the cloud platform performs target detection on each of the to-be-processed images according to the acquired image processing standard parameters and recognition target configuration information set on the cloud platform, including: The cloud platform determines a chain dependency relationship between a plurality of to-be-detected targets according to the set recognition target configuration information, and performs target detection on each of the to-be-processed images according to the image processing standard parameters and the chain dependency relationship; the chain dependency relationship represents a detection order and a detection range of each of the to-be-detected targets.

[0012] In an optional embodiment, the cloud platform performs target detection on each of the to-be-processed images according to the acquired image processing standard parameters and recognition target configuration information set on the cloud platform, including: In a case where the to-be-processed images are non-fluorescent images, the cloud platform performs channel separation on a plurality of the to-be-processed images according to a channel separation parameter in the image processing standard parameters, and performs target detection on a gray-scale image of a specified channel of the to-be-processed images according to the recognition target configuration information set on the cloud platform and an intensity threshold parameter and / or a cell segmentation parameter in the image processing standard parameters; In a case where the to-be-processed images are fluorescent images, the cloud platform performs target detection on the to-be-processed images according to the set recognition target configuration information and an intensity threshold parameter and / or a cell segmentation parameter in the image processing standard parameters.

[0013] In an optional embodiment, the method further includes: The image processing plug-in loads an example image into the image processing plug-in based on the experimental configuration information, adjusts a color style of the parameter-adjusted image according to a color style of the example image, obtains a color style reference image, and uploads the color style reference image to the cloud platform; The cloud platform performs batch processing on a plurality of the to-be-processed images according to the acquired image processing standard parameters to obtain a batch processing result, including: In a case where the automatic balancing function is turned on, the cloud platform uniformly adjusts color styles of a plurality of the to-be-processed images according to the color style reference image, performs batch processing on the plurality of to-be-processed images with uniform color styles according to the acquired image processing standard parameters, and obtains a batch processing result.

[0014] In an optional embodiment, the method further includes: The image processing plug-in loads a pre-trained typing model into the image processing plug-in based on the experimental configuration information, and uploads the typing model to the cloud platform, so that the cloud platform performs organization recognition on the to-be-processed images by using the typing model.

[0015] Secondly, the present invention provides an image batch processing system, including image processing plugins and a cloud platform that communicate with each other; The image processing plugin is used to select parameter-tuning images from multiple images to be processed corresponding to the target experiment based on the experimental configuration information set by the user for the target experiment on the cloud platform and load them into the image processing plugin, wherein the number of parameter-tuning images is less than or equal to a preset value. The image processing plugin is also used to generate preview images of the processed results of the parameter-tuned image under different image processing parameters, and upload the image processing standard parameters confirmed by the user based on the preview images of the processed results to the cloud platform. The cloud platform is used to perform batch processing on multiple images to be processed according to the acquired image processing standard parameters, so as to obtain batch processing results.

[0016] The image batch processing method and system provided in this invention involve an image processing plugin that, based on experimental configuration information set by the user on a cloud platform for a target experiment, selects parameter-adjusting images from multiple images to be processed corresponding to the target experiment and loads them into the plugin, with the number of parameter-adjusting images being less than or equal to a preset value. The image processing plugin generates preview images of the processed images under different image processing parameters and uploads the image processing standard parameters confirmed by the user based on the preview images to the cloud platform. The cloud platform performs batch processing on multiple images to be processed according to the acquired image processing standard parameters to obtain batch processing results. The image processing plugin requires only a small number of images for parameter adjustment and can provide image preview effects under different parameters, thereby determining unified parameters in a WYSIWYG manner and uploading them to the cloud platform for batch processing of multiple images, improving the processing efficiency of large batches of images and the reliability of experimental results.

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This diagram illustrates one component of the image batch processing system provided in an embodiment of the present invention. Figure 2A flowchart diagram of the image batch processing method provided by the embodiment of the present application is shown. Figure 3 A schematic diagram of a new project page is shown. Figure 4 A schematic diagram of a configuration page of experimental basic information in a new experiment is shown. Figure 5 A schematic diagram of a configuration page of color and staining components of a fluorescent KI67 experiment is shown. Figure 6 A schematic diagram of a configuration page of color and staining components of a group of islet quantitative experiment is shown. Figure 7 A schematic diagram of a configuration page of image information is shown. Figure 8 A schematic diagram of a configuration page of merging rules is shown. Figure 9 A schematic diagram of a configuration page of target type as cells is shown. Figure 10 A schematic diagram of a target information configuration page is shown. Figure 11 A schematic diagram of a configuration page of target type as tissues and using a channel graph (gray scale graph) to identify tissues is shown. Figure 12 A schematic diagram of a configuration page of target type as tissues and using a merged graph HSB to identify tissues is shown. Figure 13 A schematic diagram of a configuration page of target type as tissues and using a merged graph classification to identify tissues is shown. Figure 14 A schematic diagram of a configuration page of target type as cells in tissues is shown. Figure 15 A schematic diagram of a configuration page of target type as RNAscope is shown. Figure 16 A schematic diagram of a configuration page of counting KI67 cell nuclei is shown. Figure 17 A schematic diagram of a configuration page of counting tumor area is shown. Figure 18 A schematic diagram of a configuration page of counting KI67 positive rate is shown. Figure 19 A schematic diagram of obtaining a color style reference graph by using a Color Reinhard Tool module is shown. Figure 20 A schematic diagram of loading a parameter-adjusted image and selecting a color by using a Color Deconvolution module is shown. Figure 21 A separation effect diagram of three channels based on the color selected by a user is shown. Figure 22 The user adjusts the intensity threshold parameters of the three channel separation figures in Figure 21 The separation effect figure after adjustment is shown; Figure 23 The schematic diagram of obtaining intensity threshold parameters by Threshhold Classifier module is shown; Figure 24 The schematic diagram of obtaining HSB threshold parameters by Color Threshhold Classifier module is shown; Figure 25 The schematic diagram of uploading the model by Weka Classifier Upload module is shown; Figure 26 The schematic diagram of obtaining cell segmentation parameters by Cellpose3.0 module is shown; Figure 27 The schematic diagram of obtaining ruler line segments by ZJ Ruler module and uploading is shown; Figure 28 The schematic diagram of the configuration page with the target name as cell nucleus is shown; Figure 29 The schematic diagram of the configuration page with the target name as B cell nucleus is shown; Figure 30 The schematic diagram of the configuration page with the target name as KI67 is shown; Figure 31 The schematic diagram of the picture upload page is shown; Figure 32 The schematic diagram of the experimental result page is shown; Figure 33 The interface schematic diagram of the recognition effect is shown; Figure 34 The configuration schematic diagram of the experimental summary BI page is shown; Figure 35 The schematic diagram of the table display page of the visualization chart is shown; Figure 36 The schematic diagram of the chart display page of the table data is shown.

[0020] Icon: 10-image batch processing system; 100-image processing plug-in; 200-cloud platform. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations.

[0022] Therefore, the following detailed description of the embodiments of the application provided in the drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of the application. Based upon these embodiments of the application, those skilled in the art will be able to obtain all other embodiments of the application without making creative efforts, which fall within the scope of the application.

[0023] It should be noted that the relational terms herein such as first and second and the like are used solely to distinguish one from another entity or action, without necessarily requiring or implying any such actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element preceded by "comprises... a" does not, without more constraints, foreclose the existence of additional identical elements in the process, method, article, or apparatus that comprises the recited element.

[0024] In order to solve the problems of weak batch processing image capability of existing image processing tools, no unified analysis standard, and poor reliability of experimental results, embodiments of the application provide an image batch processing method and system. Only a small number of images are needed in the image processing plug-in for parameter adjustment, and image preview effects under different parameters can be provided, and then the seen is the obtained in a what-you-see-is-what-you-get manner to determine the unified parameters uploaded to the cloud platform for batch processing of multiple images, thereby improving the processing efficiency of a large number of images and the reliability of experimental results.

[0025] In the following, the embodiments of the application will be described in detail with reference to the accompanying drawings.

[0026] Referring to Figure 1 is a schematic diagram of one composition of the image batch processing system 10 provided by the embodiments of the application. The image batch processing system 10 includes an image processing plug-in 100 and a cloud platform 200 in communication with each other. The image processing plug-in 100 can be arranged in an image processing tool (such as ImageJ), and the image processing tool can be installed in an electronic device such as a notebook computer, a PC (Personal Computer), and the like for use by a user. The cloud platform 200 can be arranged in a cloud server, and the user can also log in to the cloud platform 200 through a webpage on the electronic device and perform related configuration of image batch processing on the cloud platform 200.

[0027] Referring to Figure 2This is a schematic flowchart of an image batch processing method provided in an embodiment of the present invention. This image batch processing method is applied to... Figure 1 The image processing plugin 100 and cloud platform 200 are shown. It should be noted that the image batch processing method of this invention does not rely on... Figure 2 The specific order described below is a limitation. It should be understood that in other embodiments, the order of some steps in the image batch processing method of the present invention can be omitted according to actual needs. The following will describe... Figure 2 The specific process shown will be explained in detail.

[0028] Step S201: Based on the experimental configuration information set by the user for the target experiment on the cloud platform, the image processing plugin selects a parameter-tuning image from multiple images to be processed corresponding to the target experiment and loads it into the image processing plugin; the number of parameter-tuning images is less than or equal to a preset value.

[0029] In this embodiment, after logging into the cloud platform, the user needs to create a project and an experiment. For example... Figure 3 On the "Create New Topic" page, users can enter a topic name to create a new topic. Under a newly created topic, multiple experiments can be created, each corresponding to a specific image processing task. For example, if the newly created experiment is a target experiment, the corresponding experimental configuration information for the target experiment can be set, including configuring basic experimental information, configuring colors and staining components, configuring color and subtyping categories, and configuring member permissions, etc.

[0030] In this embodiment, the basic experimental information mainly includes the experiment name, advanced customization, configuration ruler, AI tissue typing, automatic balancing, staining scheme, number of channels, model used, typing model, and typing category. The experiment name is the unique identifier of the experiment; advanced customization is a field visible only in the specified version; the image ruler function (configuration ruler) can be enabled or disabled. When enabled, the corresponding experiment must be selected in the ZJ Ruler plugin module to draw the ruler and upload it to the platform. AI tissue typing can be enabled or disabled. If enabled, the model used, typing model, and typing type need to be configured according to the experimental situation. The corresponding experiment can be selected and uploaded in the Weka Classifier Upload plugin; typing categories include two-category, three-category, and four-category. The staining scheme includes fluorescent and non-fluorescent staining methods, which cannot be modified after user submission. Fluorescent staining cannot be mixed with other types; the number of channels is determined based on the staining situation of the uploaded image; the automatic balancing function can be enabled or disabled. If enabled, a reference image will be uploaded, and the color style of subsequent uploaded images will be automatically adjusted based on the reference image. Figure 4 The image shown is a schematic diagram of the configuration page for basic experimental information in a newly created experiment.

[0031] In this embodiment, configuring color and coloring components mainly involves configuring the number of image channels, the color of each channel, and the coloring components. The color selection is displayed according to the coloring scheme.

[0032] For fluorescence experiments (using fluorescence as the staining scheme), seven color channels are supported: Red, Green, Blue, Yellow, Cyan, Magenta, and Gray. It should be noted that the channel creation order should match the channel order of the original fluorescence grayscale image. For example, in the KI67 fluorescence experiment, the filenames should be sorted as follows: Cell nucleus: Sample_W0002F0001T0001Z001C1; B cytoplasm: Sample_W0002F0001T0001Z001C2; KI67: Sample_W0002F0001T0001Z001C3; The channel sorting is as follows: Channel 1: Cell nucleus; Channel 2: B cytoplasm; Channel 3: KI67.

[0033] like Figure 5 The image shown is a schematic diagram of the configuration page for the colors and staining components in the fluorescent KI67 experiment.

[0034] For histochemical experiments (using a non-fluorescent staining protocol), the channel colors vary depending on the staining protocol used; please refer to the specific staining protocol for details. For example... Figure 6 The image shown is a schematic diagram of the color and staining component configuration page for the histochemical quantification experiment of pancreatic islets.

[0035] In this embodiment, configuring colors and classification categories mainly involves configuring the channel colors and categories corresponding to the classification model (such as the Weka model). It supports 4 channels, with 5 colors available for each channel (e.g., red, green, blue, yellow, white). Each channel corresponds to a classification category name (e.g., "pancreas," "islet," etc.). It should be noted that the channel order must be consistent with the training order of the classification model.

[0036] After the user configures the basic experimental information, colors and staining components, color and classification categories, and member permissions for the target experiment on the cloud platform, the experimental configuration information of the target experiment can be synchronized to the image processing plugin through the cloud platform. That is, the user can view the newly created experimental configuration information of the target experiment through the image processing plugin, and then select a small number of images (such as 1, 2, 3, etc.) from the multiple images to be processed corresponding to the target experiment as parameter tuning images and load them into the image processing plugin. The number of parameter tuning images can be determined according to actual needs, and this embodiment does not impose any restrictions on this.

[0037] In step S202, the image processing plug-in generates preview images of processing results of the parameter-adjusted image under different image processing parameters, and uploads image processing standard parameters confirmed by the user based on the preview images of the processing results to the cloud platform.

[0038] In this embodiment, the user can adjust the image processing parameters based on the operation interface provided by the image processing plug-in, the image processing plug-in generates corresponding preview images of processing results according to the image processing parameters adjusted by the user for the user to view, the user selects the image processing parameters corresponding to the preview image of the processing result with better effect and uploads the image processing parameters to the cloud platform as image processing standard parameters. It can be understood that the adjustment of the image processing parameters by the user can be directly reflected in the preview image of the processing result, and the user can more easily determine the image processing parameters with better effect to upload to the cloud platform for unified batch processing of multiple images.

[0039] In step S203, the cloud platform performs batch processing on the multiple images to be processed according to the obtained image processing standard parameters, and obtains a batch processing result.

[0040] In this embodiment, after the cloud platform obtains the image processing standard parameters uploaded by the image processing plug-in, the cloud platform performs batch processing on the multiple images to be processed based on the image processing standard parameters and the image information, merging rules, identification targets, indicators and other information configured by the user on the cloud platform, and obtains a batch processing result.

[0041] It can be seen that the image batch processing method and system provided in the embodiments of the present application, the image processing plug-in selects a parameter-adjusted image from multiple images to be processed corresponding to a target experiment based on the experiment configuration information set by the user on the cloud platform for the target experiment, loads the parameter-adjusted image into the image processing plug-in, and the number of the parameter-adjusted images is less than or equal to a preset value; the image processing plug-in generates preview images of processing results of the parameter-adjusted image under different image processing parameters, and uploads image processing standard parameters confirmed by the user based on the preview images of the processing results to the cloud platform; the cloud platform performs batch processing on the multiple images to be processed according to the obtained image processing standard parameters, and obtains a batch processing result. Only a small number of images are required in the image processing plug-in for parameter adjustment, and the image processing plug-in can provide image preview effects under different parameters, and then determine the unified parameters in a what-you-see-is-what-you-get manner to upload to the cloud platform for batch processing of multiple images, thereby improving the processing efficiency of a large number of images and the reliability of the experimental results.

[0042] In an implementation, after the user creates a target experiment on the cloud platform, the user can perform "advanced setting" operation for the target experiment, mainly including image information, merging rules, identification target and indicator configuration.

[0043] Image information configuration is mainly to configure the attribute fields filled in when uploading images (such as "group", "sample", etc.), which supports subsequent filtering and analysis in BI pages. As shown in Figure 7 , it is a schematic diagram of the configuration page of image information.

[0044] Merge rule configuration is applicable to fluorescence images. When uploading image files, automatic merging is performed according to the same characters in the file name. Merge method: set the same characters from the Xth to Yth position to determine the same group of images. For example: Sample_W0014F0001T0001Z001C1 and Sample_W0014F0001T0001Z001C2, if the first 26 characters are the same, they can be merged. As shown in Figure 8 , it is a schematic diagram of the configuration page of merge rule.

[0045] Target recognition configuration is used to determine the object type and rules of AI recognition. In this embodiment, four target types are supported: cells, tissues, cells in tissues, and RNAscope.

[0046] When the target type is configured as cells, it is suitable for identifying single cell contours and is used to count the number, area, intensity, etc. of various cells. The configuration contents include: target name, target type, target information configuration (color selection, priority, display or not), detected target, position scaling, detected channel, average intensity, area ( ), circularity, edge ROI. Among them, the target name can be nucleus, tumor cell, etc.; the target type is fixed as "cell", indicating that the current target is recognized based on cell contour features; the color selection can be selected by dragging the bar, and the color used to mark the cells in the image can be selected; the priority is an integer, and when multiple targets overlap, the larger the value, the higher the priority; whether to display can be selected to display or hide, indicating whether to present the cell in the recognition image; the detected target can be selected by the drop-down method, indicating the detection range of the target name, and the data comes from the staining components marked by the model or other existing target names, and the * indicates the first-level target configured in the plug-in; the detection position can be selected on the boundary line or inside the boundary line, mainly to identify the position characteristics of the cell, and the detection range; the position scaling is a number (such as 0.1, 0.0, -0.1), which adjusts the detection range, and a positive value expands and a negative value shrinks, for example, -0.1 indicates shrinking and 0.1 indicates expanding; the detected channel mainly sets the content in the detection range of the target (such as B cell channel); the average intensity is an intensity filtering threshold range, for example, set to 60-255, indicating that less than 60 is not involved in the statistics; the area ( ) the area of the detected target (before zooming), limit the minimum / maximum area of the identified target; the roundness of the detected target (before zooming), limit the shape feature of the identified target, the closer to 1, the more round; whether the edge ROI excludes the image four around boundary incomplete target; as shown in Figure 9 , it is a configuration page schematic diagram of the target type being cells, the user clicks the gear on the side of the target name on the page, and the target information configuration page shown in Figure 10 will pop up on the page, so that the user can configure the color selection, priority and whether to display on the target information configuration page. It should be noted that in the subsequent configuration pages of the target type being tissues, cells in tissues and RNAscope ( Figure 11~Figure 15 ), the user can configure the target information through the gear on the side of the target name in the popped-up target information configuration page, so the subsequent parts related to target information configuration will not be described in detail, and Figure 10 can be referred to. Tissue is to identify the regional features in the image (such as tumor area, necrosis area, fibrous area, etc.), and tissue identification can be configured based on three types of input images, which are channel graph, merged graph HSB and merged graph type (using the type result graph of the type model). Among them, the channel graph is to use the image gray value threshold to identify the tissue, and the configuration contents include: target name, target information configuration (color selection, priority, whether to display), target type, input image, detected channel, threshold, Pass, Watershed, area ( ), roundness, edge ROI, detection range. When configuring the target name, the tissue area name such as “fibrous area” can be customized; the color selection of the target information configuration can be selected by dragging the bar, and the color used to mark the cells in the image; the priority is an integer, when multiple targets overlap, the smaller the value, the higher the priority; whether to display can be selected to display or hide, indicating whether to present the tissue in the identification graph; the target type is fixed as “tissue”, the input image is the channel graph, the detected channel is used to specify the used channel (from the experiment configuration); the threshold is used to set the identification threshold of the pixel value, the Pass check indicates that the selected one meets the threshold condition, and the Pass unchecked indicates that the selected one does not meet the threshold condition; the area ( ) limits the minimum / maximum area of the identified target; the roundness detects the roundness of the target, limits the shape feature of the identified target, the closer to 1, the more round; whether the edge ROI excludes the image four around boundary incomplete target; the data is derived from the image processing plug-in upload. As shown in Figure 11 , it is a configuration page schematic diagram of the target type being tissues and using the channel graph (gray scale graph) to identify the tissues.

[0047] The merged image HSB is screened by hue (H), saturation (S) and brightness (B) setting area color range, and the configured contents include: target name, target information configuration (color selection, priority, whether to display), target type, input image, H (hue) range, S (saturation) range and B (brightness) range, Pass, Watershed, area, circularity. When configuring the target name, the organization area name can be customized, such as "inflammatory fibrous area"; the target type is fixed as "tissue", indicating that the current configuration is used to identify the tissue area, the input image is the merged image HSB, the H (hue) range is used to set the color recognition interval (such as 30~70 representing orange yellow green), the S (saturation) range is used to filter low saturation area, and the recognition accuracy of pure color block is improved, and the B (brightness) range is used to filter extremely dark area, and the clear area recognition effect is improved. Other fields (such as target information configuration, Pass, Watershed, area, circularity, etc.) are the same as those of the channel image; for example Figure 12 As shown in the configuration page schematic diagram of the target type being tissue and the tissue being identified by the merged image HSB, it should be noted that Figure 12 In the above and all Passes in the cloud platform and plug-ins, the meaning of Pass check indicates that the selected one meets the threshold value condition, and the selected one does not meet the threshold value condition.

[0048] The input image uses the merged image type, which is suitable for the image result after using the type model (such as Weka model), and is directly based on the tissue type identification. The configured contents include: target name, target information configuration (color selection, priority, whether to display), target type, input image, type category, Watershed, area, circularity, and edge ROI. The target name can use a custom name, such as "islet"; the target type is fixed as "tissue", indicating that the current configuration is used to identify the tissue area; the input image is the merged image type; the type category needs to match a type category (such as "islet", "fibrosis") configured in the experiment editing. Other fields (such as target information configuration, Watershed, area, circularity, etc.) are the same as those of the channel image; for example Figure 13 As shown in the configuration page schematic diagram of the target type being tissue and the tissue being identified by the merged image type.

[0049] It can be understood that the input image of the tissue recognition selection is the merged image type, so that the tissue target is obtained by the model operation, and the gray image and the merged image HSB are based on the traditional image processing method to identify the tissue target.

[0050] When the target type is configured as intra-tissue cells, it is suitable to identify a specific cell population within the identified tissue region, and realize compound identification; for example, to count the KI67 positive cells in the tumor region. The configuration contents include: target name, target information configuration (color selection, priority, display or not), target type, detection target, and detection range. The target name can use a custom name, and the target type is fixed as "intra-tissue cells", indicating that this identification is a double condition screening of tissue + cells; the color selection of the target information configuration can be selected by dragging the bar, which is used to visualize the color of the target; the priority is an integer, indicating the display priority when overlapping with other targets; whether to display can be selected to display or hide, indicating whether to display in the image; the detection target is used to specify the cell type to be identified (such as Ki67 cells); and the detection range is used to limit the tissue region to which the identified target belongs (such as "tumor"). As shown in Figure 14 , it is a configuration page diagram of the target type as intra-tissue cells.

[0051] When the target type is configured as RNAscope, the RNAscope target is used to identify the RNA probe signal point in the image, and the unit is particle, which is suitable for expression analysis. The configuration contents include: target name, target information configuration (color selection, priority, display or not), target type, detection target, probe channel, probe number, Prominence, and threshold. The target name can use a custom name, such as PPIB fluorescent signal; the target type is fixed as "RNAscope", which is used to identify RNA fluorescent probe particles; the color selection of the target information configuration can be selected by dragging the bar, which is used to mark the color of the RNA particle in the image; the priority is an integer, indicating the display priority when overlapping with other targets; whether to display can be selected to display or hide, indicating whether to display the particle result in the image; the detection target is selected by the drop-down method, and the data is derived from the existing index name; the probe channel is selected by the drop-down method, and the data is derived from the channel in the experiment editing; the probe number is used to set the minimum number of statistics after target identification; Prominence is used to set the particle identification significance, and the smaller the value, the more fluorescent points are detected, and threshold is used to filter the probes in a certain intensity range. As shown in Figure 15 , it is a configuration page diagram of the target type as RNAscope.

[0052] Data index configuration is mainly to configure the experimental indexes required for statistical analysis, which can be derived from different target objects (cells, tissues, intracellular cells, RNAscope), and calculated and analyzed by various ways. The basic configuration fields of the index include index name and display format. The index name is used to identify the data item, such as "nucleus number", "KI67 positive rate" and the like, which will be displayed in the result page and the export file. The display format is used to control the numerical precision, which supports integer, 1-4 decimal places (suitable for proportion, optical density, etc.) and percentage.

[0053] In the present embodiment, the cloud platform supports extracting data indexes from the following three source objects: 1. Source object: cell (such as cell, intracellular cell, RNAscope). In addition to the aforementioned basic configuration fields, it also includes statistical dimensions, calculation type and selection of specific objects, etc. The number dimension calculation type supports summation, and the selection of specific objects is the object of target type "cell". The area dimension calculation type supports summation, average or whole image proportion, and the selection of specific objects is the object of target type "cell". The intensity dimension calculation type supports summation or average, and the selection of specific objects is the object of target type "cell" and the measurement channel is a certain channel image. The optical density dimension calculation type supports summation or average, and the selection of specific objects is the object of target type "cell" and the measurement channel is a certain channel image. For example, as shown in the configuration page diagram for counting KI67 cell nucleus number, the index name is KI67 cell nucleus number, the source object is cell, the statistical dimension is number, the calculation type is summation, and the calculation specific object is KI67. Figure 16

[0054] 2. Source object: tissue. Tissue target is generally used for area, average intensity, optical density and other quantitative analysis. In addition to the aforementioned basic configuration fields, it also includes statistical method, calculation logic and selection of label, etc. The number dimension calculation type supports summation, and the selection of specific objects is the object of target type "tissue". The area dimension calculation type supports summation, average or whole image proportion, and the selection of specific objects is the object of target type "tissue". The intensity dimension calculation type supports summation or average, and the selection of specific objects is the object of target type "tissue" and the measurement channel is a certain channel image. The optical density dimension calculation type supports summation or average, and the selection of specific objects is the object of target type "tissue" and the measurement channel is a certain channel image. The correlation selection specific object is the object of target type "tissue", the correlation channel 1 is the selected channel image 1, the correlation channel 2 is the selected channel image 2, and the Pearson correlation coefficient of channel image 1 and channel image 2 is calculated within the selected target range. For example, as shown in the configuration page diagram for calculating the average intensity of tissue, the index name is average intensity, the source object is tissue, the statistical dimension is area, the calculation type is average, and the calculation specific object is tissue. Figure 17 ​As shown in the figure, it is a schematic diagram of a configuration page for statistical tumor area, the index name is tumor area, the source object is tissue, the statistical method is area, the calculation logic is summation, and the label is tumor area.

[0055] 3. Source object: custom (calculated by other indexes) that is, a new index is calculated by combining and calculating existing indexes. In addition to the aforementioned basic configuration fields, it also includes coefficient, index, operator symbol, coefficient, and calculated index. The common positive rate calculation coefficient is generally defaulted to 1. If the H-score score needs to be calculated, the coefficient can be adjusted. The index and the calculated index refer to two indexes selected from the existing index list to participate in the operation. For example, Figure 18 As shown in the figure, it is a configuration page schematic diagram for statistical KI67 positive rate; the index name is KI67 positive rate, and the calculation logic is KI67 cell nucleus number / B cell nucleus number.

[0056] Based on the related content of the identification target configuration on the foregoing cloud platform, it can be known that some parameters in the identification target configuration come from the image processing plug-in. In the embodiment, the image processing plug-in mainly includes five functional modules of Color Reinhard Tool (image color equalization), Color Split IHC (image color separation IHC), Cellpose3.0 (cell segmentation), Tissue Classifier (tissue level classifier), and ZJ Ruler (ZJ ruler).

[0057] In the embodiment, the Color Reinhard Tool module mainly adjusts the color style of a subsequent uploaded image according to an example image to realize unified color style conversion, which is suitable for image standardization and comparative analysis.

[0058] In one implementation, the image batch processing method provided by the embodiment of the present application can further include: the image processing plug-in loads the example image into the image processing plug-in based on the experimental configuration information, adjusts the color style of the parameter-adjusted image according to the color style of the example image to obtain a color style reference image, and uploads the color style reference image to the cloud platform. In the step S203, the cloud platform performs batch processing on the plurality of to-be-processed images according to the obtained image processing standard parameters to obtain a batch processing result, specifically including: the cloud platform adjusts the color styles of the plurality of to-be-processed images uniformly according to the color style reference image in the case that the automatic equalization function is started, and performs batch processing on the plurality of to-be-processed images with uniform color styles according to the obtained image processing standard parameters to obtain a batch processing result.

[0059] For example, Figure 19As shown, after the user configures the experimental configuration information corresponding to the target experiment on the cloud platform, the Color Reinhard Tool module of the image processing plug-in can select the corresponding topic and target experiment, and load the sample image (Select Sample IMG) and the parameter-adjusted image (Select Test IMG) corresponding to the target experiment saved locally, execute the Reinhard algorithm to match the color, and finally upload the obtained color style reference image to the cloud platform through "Upload" for batch processing. When the user starts the automatic balancing function on the cloud platform, the cloud platform will uniformly adjust the color styles of multiple to-be-processed images according to the uploaded color style reference image, that is, standardize the color styles of multiple to-be-processed images through the color style reference image, and reduce batch differences; then, the multiple to-be-processed images with uniform color styles are processed in batches according to the image information, merging rules, identification targets, index data and the like configured by the user, and batch processing results are obtained.

[0060] In the present embodiment, the Color Split IHC module includes two sub-modules, namely, a Color Deconvolution module and an NMF Tool module. The Color Deconvolution module supports dye standard color / staining pickup, performs a deconvolution operation based on a color space, and the user can select an image staining component and manually adjust parameters to realize channel separation. The NMF Tool module is an application of an NMF (Non-negative Matrix Factorization) algorithm, and a preset value is used as an initial value for NMF iteration. After clicking NMF Deconvolution, the color deconvolution vector after iteration is obtained after a certain period of training, and higher-precision color separation deconvolution is realized. The NMF Tool also supports staining pickup.

[0061] In one embodiment, the image processing plug-in generates preview images of processing results of the parameter-adjusted image under different image processing parameters in the step S202, and uploads the image processing standard parameters confirmed by the user based on the preview images of processing results to the cloud platform, specifically including: the image processing plug-in generates preview images of processing results of the parameter-adjusted image under different channel separation parameters according to the channel separation parameters set by the user, and uploads the channel separation parameters corresponding to the preview image of processing results selected by the user as the image processing standard parameters to the cloud platform.

[0062] As Figure 20As shown, assuming the user selects the corresponding topic and target experiment based on the Color Deconvolution module, and loads two locally saved hyperparameter-tuned images (Open Images) corresponding to the target experiment for A / B testing, this is used to observe the separation effect of the same set of channel separation parameters (including color unconvolution parameters and intensity threshold parameters) in the two hyperparameter-tuned images, thereby determining the optimal color unconvolution parameters and intensity threshold parameters. The user can select some colors (such as the color of collagen fibers, background color) on the two hyperparameter-tuned images, and by clicking Color Deconvolution, can obtain... Figure 21 The diagram shows the separation effect of the three channels. Users can... Figure 21 The intensity threshold parameter of each channel separation map can be further adjusted to obtain... Figure 22 The separation effect diagrams of the three channels shown correspond to intensity threshold parameters of [0~205], [0~245], and [0~252], respectively. The user clicks "Upload Config" to upload the color uncoupling parameters and intensity threshold parameters to the cloud platform. The cloud platform will then batch process multiple images corresponding to the target experiment according to these color uncoupling parameters and intensity threshold parameters.

[0063] In this embodiment, the Threshhold Classifier module in the Tissue Classifier is mainly used to adjust the intensity threshold of the channel separation image, thereby improving the recognition accuracy of each channel separation image.

[0064] In one implementation, step S202 above involves the image processing plugin generating preview images of the processed results of the parameter-tuned image under different image processing parameters, and uploading the image processing standard parameters confirmed by the user based on the preview images to the cloud platform. Specifically, the image processing plugin generates preview images of the processed results of the parameter-tuned image under different intensity threshold parameters according to the intensity threshold parameters set by the user, and uploads the intensity threshold parameters corresponding to the user-selected preview images of the processed results as image processing standard parameters to the cloud platform.

[0065] like Figure 23As shown, users can select the corresponding research topic and target experiment using the Threshold Classifier module of the image processing plugin, and load locally saved hyperparameter-tuned images (Select 8-bit Images) corresponding to the target experiment. By adjusting the intensity threshold parameters (i.e., Threshold Min and Threshold Max), the ThresholdClassifier module will display preview images of the processed hyperparameter-tuned images under different intensity threshold parameters. Users can then select the preview image with the best processing result and upload the corresponding intensity threshold parameters (e.g., Min is 30, Max is 255) to the cloud platform via Upload. The cloud platform will then batch process multiple images corresponding to the target experiment according to these intensity threshold parameters.

[0066] In this embodiment, the Color ThreshholdClassifier module in the Tissue Classifier is mainly used to adjust the hue (H), saturation (S), and brightness (B) threshold parameters of the image and filter the color range of the region.

[0067] In one embodiment, step S202 above involves the image processing plugin generating preview images of the processed results of the parameter-tuned image under different image processing parameters, and uploading the image processing standard parameters confirmed by the user based on the preview images to the cloud platform. Specifically, the image processing plugin generates preview images of the processed results of the parameter-tuned image under different HSB threshold parameters according to the HSB threshold parameters set by the user, and uploads the HSB threshold parameters corresponding to the user-selected preview images as image processing standard parameters to the cloud platform. The HSB threshold parameters include hue threshold parameters, saturation threshold parameters, and brightness threshold parameters.

[0068] like Figure 24 As shown, users can select the corresponding research topic and target experiment using the Color Threshhold Classifier module of the image processing plugin, and load the locally saved hyperparameter-tuned images corresponding to the target experiment (Select ColorImages). By adjusting the HSB threshold parameters (i.e., hue (H) threshold parameter, saturation (S) threshold parameter, and brightness (B) threshold parameter), the module will display preview images of the processed hyperparameter-tuned images under different HSB threshold parameters. Users can then select the preview image with the best processing result and upload the corresponding HSB threshold parameters (e.g., H: 150~185 (pass unchecked), S: 0~255 (pass checked), B: 30~255 (pass checked)) to the cloud platform via Upload. The cloud platform will then batch process multiple images corresponding to the target experiment according to these HSB threshold parameters.

[0069] In this embodiment, the Weka Classifier Upload module in Tissue Classifier allows users to upload their locally trained Weka models to the cloud platform. The cloud platform will automatically configure the tissue image fields, seamlessly connecting with the genotyping task in the experimental settings, and supporting subsequent AI recognition calls.

[0070] In one embodiment, the image batch processing method provided by the present invention may further include: the image processing plugin loads a pre-trained fractal model into the image processing plugin based on experimental configuration information, and uploads the fractal model to the cloud platform so that the cloud platform can use the fractal model to perform tissue recognition on the image to be processed.

[0071] like Figure 25 As shown, in the Weka Classifier Upload module of the image processing plugin, the user selects the corresponding topic and target experiment, obtains the locally trained Weka model (classification model) through Select Model File, and uploads the Weka model to the cloud platform through Upload. The cloud platform then associates the Weka model with the target experiment. After the user uploads multiple images to be processed corresponding to the target experiment in batches, the user calls the Weka model to perform tissue recognition on the images to be processed and generates a classification result map, with each color corresponding to a tissue category.

[0072] In this embodiment, the Cellpose 3.0 module integrates an enhanced Cellpose algorithm, which supports cell segmentation of different target sizes and structural types, is suitable for various contour detection tasks, and has more stable recognition and strong adaptability.

[0073] In one embodiment, step S202 above involves the image processing plugin generating preview images of the processed results of the parameter-tuned image under different image processing parameters, and uploading the image processing standard parameters confirmed by the user based on the preview images to the cloud platform. Specifically, the image processing plugin generates preview images of the processed results of the parameter-tuned image under different cell segmentation parameters according to the cell segmentation parameters set by the user, and uploads the cell segmentation parameters corresponding to the preview images of the processed results selected by the user as image processing standard parameters to the cloud platform.

[0074] like Figure 26As shown, the user selects the corresponding research topic and target experiment using the Cellpose 3.0 image processing module, and loads the locally saved hyperparameter-tuned image corresponding to the target experiment. The user selects the segmentation model to use (e.g., zhijing_if_nuclei) and adjusts parameters such as diameter, cellprob range, flow range, radius, sharpening, smoothing, and image grayscale value inversion. Clicking the "RUN" button, the Cellpose 3.0 module performs cell segmentation on the hyperparameter-tuned image, obtaining the corresponding processing result preview image. Assuming that the user-set cellprob range and flow range each contain four parameter values, the Cellpose 3.0 module will automatically generate 16 sets of parameter combinations based on the user's settings, resulting in 16 processing result preview images. The user selects the cell segmentation parameters corresponding to the best-performing processing result preview image, such as diameter, cellprob (cell probability threshold), flow (flow range threshold), radius, and the model name of the segmentation model used, and uploads it to the cloud platform via Upload. Subsequently, the cloud platform performs batch processing on multiple images corresponding to the target experiment according to the cell segmentation parameters.

[0075] In this embodiment, the ZJ Ruler module allows users to draw ruler segments and input the actual length and unit (such as μm, mm) of the ruler segments. The actual length, unit, and pixel length of the ruler segments drawn by the user are uploaded to the cloud platform. Based on the uploaded data, the cloud platform determines the conversion relationship between image pixel values ​​and actual length units. Based on this conversion relationship, all pixel-based measurements can be automatically converted into real physical units, thereby achieving accurate measurement of physical quantities such as area, length, and diameter.

[0076] like Figure 27 As shown, in the ZJ Ruler module of the image processing plugin, the user selects the corresponding topic and target experiment, and loads the locally saved ruler image via Select Image. The user draws ruler segments on the ruler image and inputs the actual length and unit of the ruler segments (e.g., 100μm). The image is then uploaded to the cloud platform via Upload. The cloud platform determines the conversion relationship between the image pixel values ​​and the actual length unit, which is used in subsequent image analysis to convert the pixel units to real physical units. It should be noted that the ruler image can be the image to be processed from the target experiment, or it can be an image captured using the same equipment and shooting parameters as the image to be processed from the target experiment.

[0077] In one embodiment, step S203 above involves batch processing multiple images to be processed based on the acquired image processing standard parameters to obtain batch processing results. Specifically, this includes: the cloud platform performing target detection on each image to be processed based on the acquired image processing standard parameters and the target recognition configuration information set on the cloud platform, and calculating indicators based on the target detection results and the indicator configuration information set on the cloud platform to obtain batch processing results.

[0078] In this embodiment, the target configuration information determines the corresponding type and rules of AI recognition, while the indicator configuration information includes experimental indicators and calculation logic required for experimental statistical analysis. The cloud platform performs target detection on each image to be processed based on the image processing standard parameters and the target configuration information pre-configured by the user, obtains the target detection results, and then performs indicator calculation based on the target detection results and indicator configuration information to finally obtain the batch processing results.

[0079] In one implementation, the above steps involve the cloud platform performing target detection on each image to be processed based on the acquired image processing standard parameters and the target recognition configuration information set on the cloud platform. Specifically, the cloud platform determines the chain dependency relationship between multiple targets to be detected based on the set target recognition configuration information, and performs target detection on each image to be processed based on the image processing standard parameters and the chain dependency relationship. The chain dependency relationship represents the detection order and detection range of each target to be detected.

[0080] In this embodiment, the chain dependency relationship can be understood as a hierarchical identification logic achieved by screening multi-level targets layer by layer. The detection of the next target depends on the identification result of the previous target, forming a "serial" analysis. It should be noted that the intracellular / RNAscope is essentially also a chain screening.

[0081] Taking the KI67 experiment as an example, the experimental images include three separate channel images: 1. cell nucleus, 2. B cell cytoplasm, and 3. KI67-positive cells. The experimental objectives include counting the total number of KI67-positive cells and calculating the proportion of KI67-positive cells in B cells. To achieve these experimental objectives, it is necessary to identify the cell nucleus, B cells, and KI67-positive cells sequentially.

[0082] like Figure 28 The diagram shows a configuration page for a target named "cell nucleus". The target to be detected is "*cell nucleus", indicating that the cell nucleus is the original segmentation target obtained based on the segmentation model. In this embodiment, it is referred to as a first-level target. The detection position is within the boundary line, and the position scaling of 0.0 indicates that the detection is performed within the boundary line of the target segmented by the model. The average intensity range is 0~255, indicating that cell nuclei that meet this condition will be marked as "cell nucleus".

[0083] likeFigure 29 As shown in the configuration page of the target name B cell, the detection target is "nucleus" (without *), indicating that no segmentation model is needed, but based on the recognition result of the primary target, it is detected whether there is B cytoplasm outside the confirmed nucleus boundary. If it is surrounded by cytoplasm, it is determined as B cell, which is called secondary target in this embodiment. The position scaling is 0.1, indicating that it is expanded by 0.1 times.

[0084] As shown in the configuration page of the target name B cell, the detection target is "nucleus" (without *), indicating that no segmentation model is needed, but based on the recognition result of the primary target, it is detected whether there is B cytoplasm outside the confirmed nucleus boundary. If it is surrounded by cytoplasm, it is determined as B cell, which is called secondary target in this embodiment. The position scaling is 0.1, indicating that it is expanded by 0.1 times. Figure 30 As shown in the configuration page of the target name KI67, the detection target is "B cell" (without *), indicating that no segmentation model is needed, but based on the recognition result of the secondary target, it is detected whether there is KI67 positive signal inside the B cell boundary. If there is, it is marked as "KI67", which is called tertiary target in this embodiment.

[0085] It can be understood that in this embodiment, the target detection of the to-be-processed image is performed in the order of nucleus (primary target) → B cell (secondary target) → KI67 positive cell (tertiary target), and the detection of B cell is based on the detected nucleus, and the detection of KI67 positive cell is based on the detected B cell. In this way, the detection order and range of nucleus, B cell and KI67 positive cell can be determined, forming the above-mentioned chain dependence relationship.

[0086] In one embodiment, the cloud platform performs target detection on each to-be-processed image according to the obtained image processing standard parameters and the recognition target configuration information set on the cloud platform, specifically including: in the case that the to-be-processed image is a non-fluorescent image, the cloud platform performs channel separation on a plurality of to-be-processed images according to the channel separation parameters in the image processing standard parameters, and performs target detection on the gray image of the specified channel of the to-be-processed image according to the recognition target configuration information set on the cloud platform and the intensity threshold parameter and / or cell segmentation parameter in the image processing standard parameters; in the case that the to-be-processed image is a fluorescent image, the cloud platform performs target detection on the to-be-processed image according to the set recognition target configuration information and the intensity threshold parameter and / or cell segmentation parameter in the image processing standard parameters.

[0087] In this embodiment, the image to be processed in the target experiment may be a fluorescent image or a non-fluorescent image. The difference between fluorescent and non-fluorescent images is that fluorescent images are grayscale images of each channel, and the original grayscale data is directly read during processing without color separation; while non-fluorescent images are single RGB color images, and color deconvolution is required before processing to obtain grayscale images of different channels. Therefore, when the image to be processed is a non-fluorescent image, the cloud platform needs to first perform channel separation on multiple images to be processed according to the channel separation parameters uploaded by the image processing plugin to obtain grayscale images of multiple channels; then, based on the target recognition configuration information and the intensity threshold parameters and / or cell segmentation parameters uploaded by the image processing plugin, target detection is performed on the grayscale images of the specified channels of the image to be processed. When the image to be processed is a fluorescent image, the cloud platform does not need to perform channel separation on the image to be processed, and directly performs target detection on the image to be processed based on the target recognition configuration information and the intensity threshold parameters and / or cell segmentation parameters uploaded by the image processing plugin.

[0088] In one implementation, users can access the services provided by the cloud platform. Figure 31 The image upload page shown allows for the batch upload of multiple images corresponding to the target experiment. When uploading images, users need to select the corresponding verification group. The verification group is the management unit for image verification in the experiment, and users can view the progress of manual verification for each group on the verification progress page. If the user sets the "group" and "sample" attribute fields as required in the image information configuration, the user also needs to fill in these attribute fields when uploading images. Custom attribute fields such as "group" and "sample" can be used as statistical dimensions for the experiment in BI analysis. After the cloud platform completes the batch processing of multiple images, the batch processing results can be viewed through... Figure 32 The experiment results page is shown below. This page allows users to view the recognition status and results of uploaded images, and also supports common management operations such as image search, group filtering, batch recognition and verification, result download, and attribute modification. Figure 33 The image shown is a schematic diagram of the recognition effect, which includes functions such as outlining, filling, adjusting line width, adjusting target color and transparency, switching on and off different targets, switching channels, downloading effect images, displaying custom indicators, and manual verification tools.

[0089] In one implementation, the cloud platform also provides a BI (Business Intelligence) analysis interface for statistical analysis and visualization of experimental data. For example... Figure 34 The image shown is a configuration diagram of the BI page for summarizing experiments. The configuration content mainly includes name, color, and attributes (derived from the image attribute field in "Advanced Settings"); for example... Figure 35 The image shown is a table display page for visual charts. Figure 36For the chart display page of table data, box plot and scatter plot are supported, and different indexes (single / multiple selection) can be selected for display according to requirements; it should be noted that all experimental recognition results in the embodiment can be exported as an Excel file, and export of data according to attribute dimensions such as verification groups, groups, and samples is also supported.

[0090] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can also be implemented by other manners. The apparatus embodiments described above are only schematic, for example, the flowcharts and block diagrams in the drawings show the possible implementation architectures, functions and operations of the apparatus, method and computer program product according to the embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment or a part of code, which contains one or more executable instructions for implementing the specified logic function. It should also be noted that, in some alternative implementation manners, the functions noted in the blocks can also occur in different order from that noted in the drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and sometimes they can also 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 the combination of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system for executing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0091] In addition, the functional modules in the various embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0092] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0093] The above merely provides the preferred embodiments of the present application, and is not used to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the principles and technical scope of the present application shall fall into the scope of the present application.

Claims

1. A method for batch image processing, characterized in that, An image processing plugin and cloud platform for mutual communication; the method includes: The image processing plugin selects parameter-tuning images from multiple images to be processed corresponding to the target experiment based on the experiment configuration information set by the user on the cloud platform for the target experiment and loads them into the image processing plugin; the number of parameter-tuning images is less than or equal to a preset value; The image processing plugin generates preview images of the processed results of the parameter-tuned image under different image processing parameters, and uploads the image processing standard parameters confirmed by the user based on the preview images of the processed results to the cloud platform. The cloud platform performs batch processing on multiple images to be processed based on the acquired image processing standard parameters to obtain batch processing results.

2. The image batch processing method according to claim 1, characterized in that, The image processing plugin generates preview images of the processed images under different image processing parameters, and uploads the standard image processing parameters confirmed by the user based on the preview images to the cloud platform, including: The image processing plugin generates preview images of the processed images under different channel separation parameters based on the channel separation parameters set by the user, and uploads the channel separation parameters corresponding to the user-selected processing result preview images as standard image processing parameters to the cloud platform.

3. The image batch processing method according to claim 1, characterized in that, The image processing plugin generates preview images of the processed images under different image processing parameters, and uploads the standard image processing parameters confirmed by the user based on the preview images to the cloud platform, including: The image processing plugin generates preview images of the processed images under different intensity threshold parameters based on the intensity threshold parameters set by the user, and uploads the intensity threshold parameters corresponding to the user-selected processing result preview images as standard image processing parameters to the cloud platform.

4. The image batch processing method according to claim 1, characterized in that, The image processing plugin generates preview images of the processed images under different image processing parameters, and uploads the standard image processing parameters confirmed by the user based on the preview images to the cloud platform, including: The image processing plugin generates preview images of the processed images under different HSB threshold parameters based on the HSB threshold parameters set by the user, and uploads the HSB threshold parameters corresponding to the user-selected preview images as standard image processing parameters to the cloud platform; the HSB threshold parameters include hue threshold parameters, saturation threshold parameters, and brightness threshold parameters.

5. The image batch processing method according to claim 1, characterized in that, The image processing plugin generates preview images of the processed images under different image processing parameters, and uploads the standard image processing parameters confirmed by the user based on the preview images to the cloud platform, including: The image processing plugin generates preview images of the processed results under different cell segmentation parameters based on the cell segmentation parameters set by the user, and uploads the cell segmentation parameters corresponding to the user-selected preview images as standard image processing parameters to the cloud platform.

6. The image batch processing method according to any one of claims 1-5, characterized in that, The cloud platform performs batch processing on multiple images to be processed based on the acquired image processing standard parameters to obtain batch processing results, including: The cloud platform performs target detection on each image to be processed based on the acquired image processing standard parameters and the target recognition configuration information set on the cloud platform, and calculates indicators based on the target detection results and the indicator configuration information set on the cloud platform to obtain batch processing results.

7. The image batch processing method according to claim 6, characterized in that, The cloud platform performs target detection on each image to be processed based on the acquired image processing standard parameters and the target recognition configuration information set on the cloud platform, including: The cloud platform determines the chain dependency relationship between multiple targets to be detected based on the set target recognition configuration information, and performs target detection on each of the images to be processed according to the image processing standard parameters and the chain dependency relationship; the chain dependency relationship represents the detection order and detection range of each target to be detected.

8. The image batch processing method according to claim 6, characterized in that, The cloud platform performs target detection on each image to be processed based on the acquired image processing standard parameters and the target recognition configuration information set on the cloud platform, including: When the image to be processed is a non-fluorescent image, the cloud platform performs channel separation on multiple images to be processed according to the channel separation parameters in the image processing standard parameters, and performs target detection on the grayscale image of a specified channel of the image to be processed according to the target recognition configuration information set on the cloud platform and the intensity threshold parameters and / or cell segmentation parameters in the image processing standard parameters; When the image to be processed is a fluorescent image, the cloud platform performs target detection on the image to be processed according to the set target recognition configuration information and the intensity threshold parameter and / or cell segmentation parameter in the image processing standard parameters.

9. The image batch processing method according to any one of claims 1-5, characterized in that, The method further includes: The image processing plugin loads the example image into the image processing plugin based on the experimental configuration information, adjusts the color style of the parameter-tuning image according to the color style of the example image to obtain a color style reference image, and uploads the color style reference image to the cloud platform; The cloud platform performs batch processing on multiple images to be processed based on the acquired image processing standard parameters, obtaining batch processing results, including: When the automatic balancing function is enabled, the cloud platform adjusts the color style of multiple images to be processed uniformly according to the color style reference map, and performs batch processing on multiple images to be processed with the same color style according to the obtained image processing standard parameters to obtain batch processing results.

10. The image batch processing method according to any one of claims 1-5, characterized in that, The method further includes: The image processing plugin loads the pre-trained fractal model into itself based on the experimental configuration information, and uploads the fractal model to the cloud platform so that the cloud platform can use the fractal model to perform tissue recognition on the image to be processed.

11. An image batch processing system, characterized in that, This includes interconnected image processing plugins and cloud platforms; The image processing plugin is used to select parameter-tuning images from multiple images to be processed corresponding to the target experiment and load them into the image processing plugin based on the experiment configuration information set by the user for the target experiment on the cloud platform; the number of parameter-tuning images is less than or equal to a preset value. The image processing plugin is also used to generate preview images of the processed results of the parameter-tuned image under different image processing parameters, and upload the image processing standard parameters confirmed by the user based on the preview images of the processed results to the cloud platform. The cloud platform is used to perform batch processing on multiple images to be processed according to the acquired image processing standard parameters, so as to obtain batch processing results.

Images