ANA and dsDNA fluorescence image processing method, device and computer equipment

Abstract

The application relates to an ANA and dsDNA fluorescence image processing method, device and computer equipment. The method comprises the following steps: acquiring a fluorescence microscope shooting image; in response to a detection item selection operation on the fluorescence microscope shooting image, determining a detection item identifier associated with the fluorescence microscope shooting image; according to the detection item identifier associated with the fluorescence microscope shooting image, distributing the fluorescence microscope shooting image to a corresponding image detection pipeline, so that the image detection pipeline detects and processes the fluorescence microscope shooting image according to an image detection algorithm process matched with the detection item identifier, and obtains a structured detection result corresponding to the fluorescence microscope shooting image; and displaying the structured detection result corresponding to the fluorescence microscope shooting image. The method can improve the laboratory detection efficiency of an anti-nuclear antibody and an anti-double-stranded DNA antibody fluorescence image.

Description

Technical Field

[0001] This application relates to the fields of medical testing automation, digital pathological image analysis, medical artificial intelligence, and integrated hardware and software testing equipment, and in particular to an ANA and dsDNA fluorescence image processing method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Technology

[0002] In laboratory testing for autoimmune diseases, commonly used methods include ANA (Anti-Nuclear Antibody) indirect immunofluorescence assay and dsDNA (Crithidia luciliae) (CL) indirect immunofluorescence assay. However, traditional techniques typically rely on laboratory personnel to observe samples field-by-field under a fluorescence microscope and make judgments. This method often depends on human experience, is highly subjective, and has limited consistency among different personnel. Furthermore, a single image contains a large number of cellular targets, and different cell states and fluorescence intensities significantly affect the interpretation results. This approach is not conducive to improving the processing efficiency of ANA (Anti-Nuclear Antibody) and dsDNA (double-stranded DNA) immunofluorescence images, thus impacting laboratory testing efficiency.

[0003] Therefore, traditional techniques suffer from low laboratory detection efficiency for fluorescent images of antinuclear antibodies and anti-double-stranded DNA antibodies. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for processing ANA and dsDNA fluorescence images that can improve the laboratory detection efficiency of fluorescence images of antinuclear antibodies and anti-double-stranded DNA antibodies, in order to address the above-mentioned technical problems.

[0005] In a first aspect, this application provides a method for processing ANA and dsDNA fluorescence images, the method comprising:

[0006] Acquire fluorescence microscopy images to be processed; the fluorescence microscopy images are images obtained by taking pictures of the field of view of the sample to be tested through a fluorescence microscope; the sample to be tested is a sample prepared by indirect immunofluorescence method;

[0007] In response to the selection of detection items for the fluorescence microscope image, a detection item identifier associated with the fluorescence microscope image is determined; the detection item identifier is used to characterize whether the fluorescence microscope image is detected in ANA detection mode or dsDNA detection mode.

[0008] According to the detection item identifier associated with the fluorescence microscope image, the fluorescence microscope image is assigned to the corresponding image detection pipeline. The image detection pipeline performs detection processing on the fluorescence microscope image according to the image detection algorithm flow that matches the detection item identifier, and obtains the structured detection result corresponding to the fluorescence microscope image. The structured detection result includes the detection results corresponding to each of the multiple detection items that match the detection item identifier.

[0009] The structured detection results corresponding to the images captured by the fluorescence microscope are displayed on a display device.

[0010] In one embodiment, when the detection item identifier indicates that the fluorescence microscope image is detected using the ANA detection mode, the step of performing detection processing on the fluorescence microscope image according to the image detection algorithm flow matching the detection item identifier to obtain the structured detection result corresponding to the fluorescence microscope image includes:

[0011] The effectiveness of green fluorescence in the images captured by the fluorescence microscope is verified to obtain the green fluorescence effectiveness verification result; the green fluorescence effectiveness verification is used to verify whether the green fluorescence signal in the images captured by the fluorescence microscope meets the preset detection conditions.

[0012] If the green fluorescence validity verification result is passed, the fluorescence microscope image is cropped into multiple cell region cropped images according to the cropping strategy corresponding to the ANA detection mode.

[0013] The multiple cropped cell region images are input into a pre-trained ANA classification model to obtain the ANA classification model output; the ANA classification model output includes at least global titer classification output, cell type binary classification output, and titer regression output;

[0014] The ANA classification model output is subjected to ANA post-processing to obtain the ANA detection results of the fluorescence microscope image, which are used as the structured detection results. The ANA detection results include at least the positive / negative judgment results, positive category probability, karyotype results, titers corresponding to each karyotype, global titer probability distribution, detection statistics, and detection time information.

[0015] In one embodiment, cropping the fluorescence microscope image into multiple cell-cropped images according to the cropping strategy corresponding to the ANA detection mode includes:

[0016] The fluorescence microscope images are input into a pre-trained cell localization model to obtain cell localization results for the fluorescence microscope images. The cell localization results include the position information of ordinary cells in the fluorescence microscope images, and the position information of cells in the mitotic phase in the fluorescence microscope images. The cells in the mitotic phase are cells in the mitotic stage.

[0017] Based on the cell localization results, the images captured by the fluorescence microscope are cropped to obtain the multiple cropped images of the cell regions.

[0018] In one embodiment, the ANA post-processing of the ANA classification model output to obtain the ANA detection result of the fluorescence microscope image includes:

[0019] The global titer classification output is subjected to SoftMax calculation to obtain the titer probability distribution. The titer probability distribution is then mapped to a preset set of titer values ​​to obtain the global titer probability distribution.

[0020] The sigmoid function is calculated on the binary cell type output to obtain the cell type classification result; the cell type classification result is used to determine the positive or negative result.

[0021] The method further includes:

[0022] If the global titer is determined to be higher than the preset threshold and none of the cell types are positive, the cell type with the highest confidence level is selected as positive and used as the positive / negative judgment result.

[0023] The cell types include nuclear type group, cytoplasmic type group, and independent type group, and the method further includes:

[0024] Based on the cell type classification results, a mutual exclusion solution is performed on the karyotype classification probabilities of the karyotype group and the karyotype classification probabilities of the cytoplasmic group to obtain the winning type of the karyotype group and the winning type of the cytoplasmic group; wherein, the priority rule in the mutual exclusion solution is that when the probability of the lower priority type is significantly higher than that of the higher priority type, the winning result is allowed to be flipped.

[0025] Obtain the independent positive results of the independent genotype group, and extract the titer from the titer regression output of the target titer regression head in the ANA classification model based on the winning type of the karyotype group, the winning type of the cytoplasmic group, and the independent positive results, and determine the corresponding titer for each karyotype.

[0026] In one embodiment, when the detection item identifier indicates that the fluorescence microscope image is detected using a dsDNA detection mode, the step of processing the fluorescence microscope image according to an image detection algorithm flow matching the detection item identifier to obtain the structured detection result corresponding to the fluorescence microscope image includes:

[0027] According to the cropping strategy corresponding to the dsDNA detection mode, the fluorescence microscope image is cropped into multiple ordinary cell region cropped images; the ordinary cell region cropped image is determined by a square sampling frame, and the side length of the sampling frame is a preset multiple of the long side length of the ordinary cell detection frame in the fluorescence microscope image;

[0028] The multiple cropped images of ordinary cell regions are input into a pre-trained dsDNA classification model to obtain the output of the dsDNA classification model; the output of the dsDNA classification model includes at least the original scores for positive and negative binary classification and the titer regression output;

[0029] The dsDNA classification model output is subjected to dsDNA post-processing to obtain the dsDNA detection results of the fluorescence microscope image, which are used as the structured detection results; the dsDNA detection results include at least the positive / negative judgment result, the positive category probability, the titer level, and the titer value.

[0030] In one embodiment, the step of performing dsDNA post-processing on the dsDNA classification model output to obtain the dsDNA detection results from the fluorescence microscope image includes:

[0031] The original scores of the binary classification were calculated using SoftMax to obtain the probabilities of the positive and negative categories.

[0032] If the probability of the positive category is greater than the probability of the negative category, the titer regression output is rounded to a preset level to obtain the titer level, and the titer regression output is mapped to a preset set of titer values ​​to obtain the titer value.

[0033] Secondly, this application also provides an ANA and dsDNA fluorescence image processing device, the device comprising:

[0034] The acquisition module is used to acquire fluorescence microscope images to be processed; the fluorescence microscope images are images obtained by capturing the field of view of the sample under test through a fluorescence microscope;

[0035] The selection module is used to determine the detection item identifier associated with the fluorescence microscope image in response to a detection item selection operation on the fluorescence microscope image; the detection item identifier is used to identify the detection mode used on the fluorescence microscope image; the detection mode includes one of ANA detection mode and dsDNA detection mode;

[0036] The detection module is used to assign the fluorescence microscope images to corresponding image detection pipelines according to the detection item identifiers associated with the fluorescence microscope images. The image detection pipelines then perform detection processing on the fluorescence microscope images according to the image detection algorithm flow that matches the detection item identifiers, thereby obtaining the structured detection results corresponding to the fluorescence microscope images. The structured detection results include the detection results corresponding to each of the multiple detection items that match the detection item identifiers.

[0037] The display module is used to display the structured detection results corresponding to the images captured by the fluorescence microscope via a display device.

[0038] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0039] Acquire fluorescence microscopy images to be processed; the fluorescence microscopy images are images obtained by taking pictures of the field of view of the sample to be tested through a fluorescence microscope; the sample to be tested is a sample prepared by indirect immunofluorescence method;

[0040] In response to the selection of detection items for the fluorescence microscope image, a detection item identifier associated with the fluorescence microscope image is determined; the detection item identifier is used to characterize whether the fluorescence microscope image is detected in ANA detection mode or dsDNA detection mode.

[0041] According to the detection item identifier associated with the fluorescence microscope image, the fluorescence microscope image is assigned to the corresponding image detection pipeline. The image detection pipeline performs detection processing on the fluorescence microscope image according to the image detection algorithm flow that matches the detection item identifier, and obtains the structured detection result corresponding to the fluorescence microscope image. The structured detection result includes the detection results corresponding to each of the multiple detection items that match the detection item identifier.

[0042] The structured detection results corresponding to the images captured by the fluorescence microscope are displayed on a display device.

[0043] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0044] Acquire fluorescence microscopy images to be processed; the fluorescence microscopy images are images obtained by taking pictures of the field of view of the sample to be tested through a fluorescence microscope; the sample to be tested is a sample prepared by indirect immunofluorescence method;

[0045] In response to the selection of detection items for the fluorescence microscope image, a detection item identifier associated with the fluorescence microscope image is determined; the detection item identifier is used to characterize whether the fluorescence microscope image is detected in ANA detection mode or dsDNA detection mode.

[0046] According to the detection item identifier associated with the fluorescence microscope image, the fluorescence microscope image is assigned to the corresponding image detection pipeline. The image detection pipeline performs detection processing on the fluorescence microscope image according to the image detection algorithm flow that matches the detection item identifier, and obtains the structured detection result corresponding to the fluorescence microscope image. The structured detection result includes the detection results corresponding to each of the multiple detection items that match the detection item identifier.

[0047] The structured detection results corresponding to the images captured by the fluorescence microscope are displayed on a display device.

[0048] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0049] Acquire fluorescence microscopy images to be processed; the fluorescence microscopy images are images obtained by taking pictures of the field of view of the sample to be tested through a fluorescence microscope; the sample to be tested is a sample prepared by indirect immunofluorescence method;

[0050] In response to the selection of detection items for the fluorescence microscope image, a detection item identifier associated with the fluorescence microscope image is determined; the detection item identifier is used to characterize whether the fluorescence microscope image is detected in ANA detection mode or dsDNA detection mode.

[0051] According to the detection item identifier associated with the fluorescence microscope image, the fluorescence microscope image is assigned to the corresponding image detection pipeline. The image detection pipeline performs detection processing on the fluorescence microscope image according to the image detection algorithm flow that matches the detection item identifier, and obtains the structured detection result corresponding to the fluorescence microscope image. The structured detection result includes the detection results corresponding to each of the multiple detection items that match the detection item identifier.

[0052] The structured detection results corresponding to the images captured by the fluorescence microscope are displayed on a display device.

[0053] The aforementioned ANA and dsDNA fluorescence image processing method, apparatus, computer device, computer-readable storage medium, and computer program product acquire a fluorescence microscope image to be processed. This image is obtained by capturing the field of view of the sample under test through a fluorescence microscope. The sample under test is prepared using indirect immunofluorescence. In response to a detection item selection operation on the fluorescence microscope image, a detection item identifier associated with the fluorescence microscope image is determined. This detection item identifier indicates whether the fluorescence microscope image is processed using an ANA detection mode or a dsDNA detection mode. The fluorescence microscope image is then assigned to a corresponding image detection pipeline according to the associated detection item identifier. This pipeline processes the fluorescence microscope image according to an image detection algorithm flow matching the detection item identifier, resulting in a structured detection result corresponding to the fluorescence microscope image. The structured detection result includes the detection results corresponding to multiple detection items matching the detection item identifier. Finally, the structured detection result corresponding to the fluorescence microscope image is displayed on a display device. Thus, by responding to a detection item selection operation on the fluorescence microscope image, ANA and dsDNA can be distinguished. The detection mode and dsDNA detection mode assign corresponding fluorescence microscope images to dedicated image detection pipelines and process them using matching algorithm flows. This accurately adapts to the differences in sample characteristics and interpretation logic between the two detection methods, avoiding detection errors caused by mixing algorithms. Simultaneously, it can automatically generate standardized structured detection results and display them through display devices. This achieves integrated, automated, and intelligent processing of ANA and dsDNA fluorescence image detection, effectively improving detection efficiency, result standardization, and interpretation accuracy. It integrates microscope image input, real-time processing, and result feedback into a single system, enabling compatibility with both ANA and dsDNA projects within a unified device. This reduces system deployment and maintenance complexity, improves platform scalability, and significantly shortens the time from image acquisition to interpretation, thereby improving the laboratory detection efficiency of antinuclear antibody and anti-double-stranded DNA antibody fluorescence images. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is an application environment diagram of an ANA and dsDNA fluorescence image processing method in one embodiment.

[0056] Figure 2 This is a flowchart illustrating a fluorescence image processing method for ANA and dsDNA in one embodiment.

[0057] Figure 3 This is a schematic diagram of an ANA and dsDNA fluorescence image processing method in another embodiment;

[0058] Figure 4 This is a structural block diagram of an immunofluorescence image real-time interpretation device in one embodiment;

[0059] Figure 5 This is a structural block diagram of an ANA and dsDNA fluorescence image processing device in one embodiment;

[0060] Figure 6 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0062] The ANA and dsDNA fluorescence image processing method provided in this application can be applied to, for example... Figure 1In the application environment shown, in practical applications, terminal 102 can acquire fluorescence microscope images to be processed; the fluorescence microscope images are images obtained by capturing the field of view of the sample to be tested through fluorescence microscope 104; the sample to be tested is a sample prepared using indirect immunofluorescence; terminal 102 can determine the detection item identifier associated with the fluorescence microscope images in response to the detection item selection operation of the fluorescence microscope images; the detection item identifier is used to characterize whether the fluorescence microscope images are detected using ANA detection mode or dsDNA detection mode; terminal 102 can allocate the fluorescence microscope images to the corresponding image detection pipeline according to the detection item identifier associated with the fluorescence microscope images, so that the image detection pipeline performs detection processing on the fluorescence microscope images according to the image detection algorithm flow matching the detection item identifier, and obtains the structured detection results corresponding to the fluorescence microscope images; the structured detection results include the detection results corresponding to each of the multiple detection items matching the detection item identifier; terminal 102 can display the structured detection results corresponding to the fluorescence microscope images through a display device. The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc.

[0063] In one exemplary embodiment, such as Figure 2 As shown, a method for processing ANA and dsDNA fluorescence images is provided, which can be applied to... Figure 1 Taking terminal 102 as an example, the explanation includes the following steps S202 to S206. Wherein:

[0064] Step S202: Obtain the fluorescence microscope image to be processed.

[0065] Among them, the images captured by the fluorescence microscope are images obtained by taking pictures of the field of view of the sample under test through the fluorescence microscope.

[0066] The samples to be tested were prepared using indirect immunofluorescence methods. These indirect immunofluorescence methods include ANA indirect immunofluorescence and dsDNA Crithidia luciliae (CL, a matrix or substrate model for dsDNA indirect immunofluorescence detection) indirect immunofluorescence.

[0067] The sample to be tested can refer to an immunofluorescence detection slide that has undergone indirect immunofluorescence staining. In practical applications, under ANA detection mode, the sample to be tested can be a HEp-2 cell indirect immunofluorescence slide; under dsDNA detection mode, the sample to be tested can be a Crithidia luciliae short-hymened worm indirect immunofluorescence slide.

[0068] In practical implementation, a fluorescence microscope with image acquisition capabilities is used. The inspector places the sample under the microscope and then uses its camera module to capture images of the sample's field of view. These images are then transmitted to a terminal via the "POST / api / upload" interface. The terminal receives and processes these images. In practical applications, the terminal's image acquisition and caching module decodes, validates, names, and caches the images. The fluorescence microscope images can be single static images or frames extracted from a microscope video stream.

[0069] Step S204: In response to the detection item selection operation of the fluorescence microscope image, determine the detection item identifier associated with the fluorescence microscope image.

[0070] The detection item identifier is used to indicate whether the fluorescence microscope image was detected using the ANA detection mode or the dsDNA detection mode.

[0071] In practice, the terminal can display a detection item selection interface, through which the user can select whether to use ANA detection mode or dsDNA detection mode for the images captured by the fluorescence microscope. In other words, the detection item selection operation for the images captured by the fluorescence microscope is received through this detection item selection interface.

[0072] Then, in response to the selection of a detection item in the fluorescence microscope image, the terminal can determine the detection item identifier associated with the fluorescence microscope image. In practical applications, the terminal can bind the fluorescence microscope image with the detection item identifier of the selected detection item.

[0073] In practical applications, the communication between the fluorescence microscope and the terminal can be achieved through a network port, USB, data acquisition card, or wireless transmission interface.

[0074] Step S206: According to the detection item identifier associated with the fluorescence microscope image, the fluorescence microscope image is assigned to the corresponding image detection pipeline. The image detection pipeline performs detection processing on the fluorescence microscope image according to the image detection algorithm process that matches the detection item identifier, and obtains the structured detection result corresponding to the fluorescence microscope image.

[0075] The structured test results include the test results corresponding to each of the multiple test items that match the test item identifier.

[0076] In specific implementation, after determining the detection item identifier associated with the fluorescence microscope image, the terminal can allocate the fluorescence microscope image to the corresponding image detection pipeline according to the associated detection item identifier, realizing an online task access and concurrent execution mechanism, including:

[0077] 1. The terminal has a built-in server program that continuously listens for image upload requests;

[0078] 2. Each request must include at least an image file and a project identifier, which is used to distinguish between ANA testing mode and dsDNA testing mode;

[0079] 3. The main control module pre-initializes corresponding pipeline instances for different projects to avoid repeatedly loading models;

[0080] 4. When concurrent images arrive, image tasks can be sent to a thread pool or process pool for execution to improve throughput;

[0081] 5. After processing, return structured results, which include at least the classification results, image information, quality control information, positive and negative results, positive probability, and dsDNA titer value.

[0082] Then, the terminal can process the fluorescence microscope images through the image detection pipeline according to the image detection algorithm flow that matches the detection item identifier, and obtain the structured detection results corresponding to the fluorescence microscope images. In practical applications, the image detection algorithm flow corresponding to the ANA detection mode and the dsDNA detection mode will be further explained below, and will not be explained here.

[0083] Step S208: Display the structured detection results corresponding to the images captured by the fluorescence microscope via a display device.

[0084] In practice, the terminal can display the structured detection results corresponding to the images captured by the fluorescence microscope via a display device. In real-world applications, these structured detection results can be JSON, HL7 messages, PDF reports, database records, or printed documents. The display device can be a local display screen connected to the terminal, or it can be a browser page, a tablet terminal, a host computer client, or a hospital intranet terminal.

[0085] In the aforementioned ANA and dsDNA fluorescence image processing method, a fluorescence microscope image to be processed is acquired. This fluorescence microscope image is an image captured through the field of view of the sample to be tested using a fluorescence microscope. The sample to be tested is a sample prepared using indirect immunofluorescence. In response to a detection item selection operation on the fluorescence microscope image, a detection item identifier associated with the fluorescence microscope image is determined. The detection item identifier is used to characterize whether the fluorescence microscope image is processed using the ANA detection mode or the dsDNA detection mode. The fluorescence microscope image is then assigned to the corresponding image detection pipeline according to the detection item identifier. The image detection pipeline processes the fluorescence microscope image according to the image detection algorithm flow matching the detection item identifier, obtaining the structured detection result corresponding to the fluorescence microscope image. The structured detection result includes the detection results corresponding to multiple detection items matching the detection item identifier. Finally, the structured detection result corresponding to the fluorescence microscope image is displayed on a display device. Thus, by responding to the detection item selection operation on the fluorescence microscope image, the ANA detection mode and the dsDNA detection mode can be distinguished. The detection mode assigns corresponding fluorescence microscope images to a dedicated image detection pipeline and processes them using a matching algorithm. This accurately adapts to the differences in sample characteristics and interpretation logic between the two detection methods, avoiding detection errors caused by mixing algorithms. It also automatically generates standardized, structured detection results and displays them on a display device. This achieves integrated, automated, and intelligent processing of ANA and dsDNA fluorescence image detection, effectively improving detection efficiency, result standardization, and interpretation accuracy. It integrates microscope image input, real-time processing, and result feedback into a single system, enabling compatibility with both ANA and dsDNA tests within a unified device. This reduces system deployment and maintenance complexity, improves platform scalability, and significantly shortens the time from image acquisition to interpretation, thereby increasing the laboratory detection efficiency of antinuclear antibody and anti-double-stranded DNA antibody fluorescence images.

[0086] In an exemplary embodiment, when the detection item identifier represents the use of ANA detection mode for the fluorescence microscope image, the fluorescence microscope image is processed according to the image detection algorithm flow matching the detection item identifier to obtain the structured detection result corresponding to the fluorescence microscope image. This includes: verifying the effectiveness of green fluorescence in the fluorescence microscope image to obtain a green fluorescence effectiveness verification result; if the green fluorescence effectiveness verification result is successful, cropping the fluorescence microscope image into multiple cell region cropped images according to the cropping strategy corresponding to the ANA detection mode; inputting the multiple cell region cropped images into a pre-trained ANA classification model to obtain the ANA classification model output; and performing ANA post-processing on the ANA classification model output to obtain the ANA detection result of the fluorescence microscope image, which serves as the structured detection result.

[0087] Among them, the green fluorescence validity verification is used to verify whether the green fluorescence signal in the image captured by the fluorescence microscope meets the preset detection conditions.

[0088] The ANA classification model output includes at least global titer classification output, cellular type binary classification output, and titer regression output.

[0089] The ANA test results should include at least the positive / negative result, the probability of the positive category, the karyotype result, the titer corresponding to each karyotype, the global titer probability distribution, the test statistics, and the test time.

[0090] In specific implementation, when the detection item identifier represents the fluorescence microscope image using the ANA detection mode, the terminal performs detection processing on the fluorescence microscope image according to the image detection algorithm flow matching the detection item identifier to obtain the structured detection result corresponding to the fluorescence microscope image. This process specifically includes:

[0091] The terminal verifies the effectiveness of green fluorescence in images captured by a fluorescence microscope. This verifies whether the green fluorescence signal in the images meets preset detection conditions, i.e., whether the green fluorescence signal in the images is insufficient, thus obtaining the green fluorescence effectiveness verification result. In practical applications, the green fluorescence effectiveness verification result may include the following items:

[0092] Item 1. Has it passed verification?

[0093] Project 2. Reasons for failure;

[0094] Project 3. Information related to green fluorescence.

[0095] If the green fluorescence validity verification result is unsuccessful, it indicates that the green fluorescence signal in the image captured by the fluorescence microscope is insufficient. The terminal outputs a verification failure message and blocks the subsequent classification process, thereby preventing invalid images from entering the interpretation stage.

[0096] If the green fluorescence validity verification result is passed, it means that the green fluorescence signal in the fluorescence microscope image is sufficient. The terminal can then crop the fluorescence microscope image into multiple cell region cropped images according to the cropping strategy corresponding to the ANA detection mode.

[0097] Optionally, according to the cropping strategy corresponding to the ANA detection mode, the fluorescence microscope image is cropped into multiple cell cropping images, including: inputting the fluorescence microscope image into a pre-trained cell localization model to obtain the cell localization result of the fluorescence microscope image; and cropping the fluorescence microscope image according to the cell localization result to obtain multiple cell region cropping images.

[0098] The cell localization results include the location information of ordinary cells in fluorescence microscope images, and the location information of cells in the dividing phase in fluorescence microscope images.

[0099] Among them, cells in the mitotic phase are cells that are in the mitotic stage;

[0100] In practice, during the process of cropping the fluorescence microscope image into multiple cell cropping images according to the cropping strategy corresponding to the ANA detection mode, the terminal can input the fluorescence microscope image into a pre-trained cell localization model to obtain the cell localization result of the fluorescence microscope image.

[0101] Specifically, the terminal can perform dual-channel detection:

[0102] 1. Whole-cell detection, used to locate ordinary cells in an image; wherein, the terminal can input images captured by a fluorescence microscope into a pre-trained ordinary cell localization model to obtain the location information of ordinary cells in the fluorescence microscope images; wherein, the location information includes bounding box, confidence level, class number and class name.

[0103] 2. Cell Detection During Mitosis: This feature is used to locate cells in the mitotic phase. The terminal can input images captured by a fluorescence microscope into a pre-trained cell localization model to obtain the location information of cells in the fluorescence microscope images. This location information includes bounding boxes, confidence scores, class numbers, and class names.

[0104] In practical applications, the aforementioned cell localization models (i.e., ordinary cell localization models and mitotic cell localization models) can be ONNX (Open Neural Network Exchange, a cross-framework neural network model representation format) format target detection models. The terminal's main control processing module counts the total number of cells, the number of mitotic cells, and the time consumed for each detection.

[0105] Then, the terminal can crop the images captured by the fluorescence microscope according to the cell localization results to obtain multiple cropped cell region images. Specifically, the terminal can crop multiple cropped cell region images (i.e., multiple cell patches, local image blocks cropped from the original image) according to a preset strategy for the detected normal cells and dividing cells, and use them as multiple instances of a single image for subsequent input into the pre-trained ANA classification model.

[0106] Then, the terminal can input multiple cropped cell region images into the pre-trained ANA classification model to obtain the ANA classification model output.

[0107] The ANA classification model output includes at least the following:

[0108] ANA classification model output 1: Global titer classification output;

[0109] ANA classification model output 2: 14 cell types in binary classification;

[0110] The ANA classification model outputs 3:7 titer regression outputs.

[0111] Among them, the 14 cell types can be divided into at least the following three groups:

[0112] 1. Karyotypes: homogeneous, dense fine-granular, granular, nuclear membrane, and polymorphic;

[0113] 2. Cytoplasmic type group: cytoplasmic fibrous type, cytoplasmic granular type, cytoplasmic reticular / mitochondrial-like type, cytoplasmic polar / Golgi-like type;

[0114] 3. Independent types: centromere type, nuclear point type, nucleolar type, cytoplasmic rod ring type, nuclear type during cell mitosis.

[0115] Finally, the terminal can perform ANA post-processing on the ANA classification model output to obtain the ANA detection results of the fluorescence microscope images, which can be used as the structured detection results.

[0116] Optionally, ANA post-processing is performed on the ANA classification model output to obtain the ANA detection results of the fluorescence microscope images, including: performing SoftMax calculation on the global titer classification output to obtain the titer probability distribution, mapping the titer probability distribution to a preset set of titer values ​​to obtain the global titer probability distribution; performing sigmoid calculation on the cytotype binary classification output to obtain the cytotype classification result; the cytotype classification result is used to determine the positive or negative judgment result.

[0117] In practical implementation, the terminal can perform SoftMax calculation on the global titer classification output. Specifically, the terminal can input the global titer classification output into the SoftMax function (a function that maps multiple scores to a probability distribution) to obtain the titer probability distribution. This titer probability distribution is then mapped to a preset set of titer values ​​to obtain the global titer probability distribution. The titer value set includes 0, 80, 160, 320, 640, and 1280.

[0118] Additionally, the terminal can perform sigmoid calculations on the binary cytotype output to obtain the cytotype classification result; this result is used to determine the positive or negative result. For example, the cytotype classification result may include: homogeneous type (average probability 0.85); speckled type (average probability 0.12); and nucleolar type (average probability 0.03). The terminal can then output the positive or negative result based on the cytotype classification result.

[0119] Optionally, the above method further includes: when it is determined that the global titer is higher than a preset threshold and all cell types are not positive, selecting the cell type with the highest positive confidence as positive and using it as the positive / negative judgment result.

[0120] When the terminal determines that the global titer is not lower than the preset threshold but none of the cell types have reached the positive threshold, the terminal triggers the fallback logic, forcibly selecting the cell type with the highest positive confidence as the positive result, in order to improve the conservatism and consistency of the overall interpretation.

[0121] Optionally, the cellular types include the nuclear type group, the cytoplasmic type group, and the independent type group.

[0122] The above method also includes: performing a mutually exclusive solution on the karyotype group and the cytoplasmic group based on the cytotype classification results to obtain the winning type of the karyotype group and the winning type of the cytoplasmic group; obtaining the independent positive results of the independent group; and extracting the titer from the titer regression output of the target titer regression head in the ANA classification model based on the winning type of the karyotype group, the winning type of the cytoplasmic group, and the independent positive results to determine the corresponding titer for each karyotype.

[0123] In the mutual exclusion solution, the priority rule is that when the probability of a lower priority type is significantly higher than that of a higher priority type, the winning result can be flipped.

[0124] In practice, the terminal can perform a mutually exclusive solution for the karyotype classification probabilities of the karyotype group and the karyotype classification probabilities of the cytoplasmic group based on the cytotype classification results, to obtain the winning karyotype type of the karyotype group and the winning cytoplasmic type of the cytoplasmic group. For example, the terminal can statistically calculate the karyotype classification probabilities of the karyotype group and the karyotype group based on the cytotype classification results; for instance, the karyotype group includes homogeneous, speckled, and nucleolar types, where the classification probability of the homogeneous type is 0.85; the classification probability of the speckled type is 0.12; and the karyotype of the nucleolar type is 0.03. The cytoplasmic morphology group includes mitochondrial M2 type, ribosomal P type, and cytoplasmic background. The classification probability of mitochondrial M2 type is 0.08; ribosomal P type is 0.05; and cytoplasmic background is 0.87. It can be seen that the homogeneous type in the karyotype group has the highest classification probability, far exceeding the threshold. Spotted and nucleolar types are mutually excluded, and the homogeneous type wins in the karyotype group. The cytoplasmic background type has the highest classification probability, far exceeding the threshold, indicating the lack of specific cytoplasmic fluorescence. Mitochondrial M2 type and ribosomal P type are mutually excluded, and the cytoplasmic background type wins in the cytoplasmic morphology group, i.e., cytoplasmic negative. This achieves the goal that in the structured detection results, the karyotype only shows the homogeneous type, and the cytoplasmic type only shows the negative, ensuring that only one clear result from each group is retained, conforming to clinical reporting standards.

[0125] Then, the terminal can obtain independent positive results for the independent group. The independent positive result of the independent group is a special fluorescence pattern in the ANA test that does not belong to either the karyotype group or the cytoplasmic group, and has a separate positive judgment result.

[0126] The terminal extracts the titer from the titer regression output of the target titer regression head in the ANA classification model based on the karyotype winning type, cytoplasmic winning type, and independent positive results. It then determines the titer corresponding to each karyotype and outputs the final type-titer control result. The target titer regression head can include a shared titer regression head or a dedicated titer regression head; both shared and dedicated titer regression heads are titer prediction modules in the ANA classification model, used to output continuous predicted titer values.

[0127] Among them, a shared titer regression head can refer to a titer prediction module that is used in general modes (such as for homogeneous, speckled, mitochondrial M2 types, etc.); a dedicated titer regression head can refer to a titer prediction module that is designed separately for special modes (such as for centromere type, nucleolar type, Jo-1 type, nuclear membrane type, etc.).

[0128] Finally, the ANA test results output by the terminal in ANA test mode include:

[0129] Test results 1. Positive / negative interpretation results;

[0130] Result 2. Positive probability;

[0131] Result 3. One or more karyotype results;

[0132] 4. Test results: titers corresponding to each karyotype;

[0133] Detection Result 5. Global Titer Probability Distribution;

[0134] 6. Detection Results. Detection Statistics and Time Consumption Information.

[0135] The technical solution of this embodiment verifies the effectiveness of green fluorescence in images captured by a fluorescence microscope, thereby selecting fluorescence microscope images whose signals meet preset detection conditions and ensuring the reliability of subsequent detection. Then, a cropping strategy specific to the ANA detection mode is used to obtain standard cell region cropped images. Combined with a pre-trained ANA classification model, multi-dimensional outputs are achieved, including global titer classification, binary cell type classification, and titer regression. Finally, ANA post-processing yields complete ANA detection results covering positive / negative judgment, karyotype, titer, and other dimensions. This achieves automated and standardized processing of the entire ANA fluorescence image detection process, effectively improving the accuracy and completeness of results under the ANA detection mode.

[0136] In an exemplary embodiment, when the detection item identifier characterizes the fluorescence microscope image using the dsDNA detection mode, the fluorescence microscope image is processed according to the image detection algorithm flow matching the detection item identifier to obtain the structured detection result corresponding to the fluorescence microscope image. This includes: cropping the fluorescence microscope image into multiple ordinary cell region cropped images according to the cropping strategy corresponding to the dsDNA detection mode; inputting the multiple ordinary cell region cropped images into a pre-trained dsDNA classification model to obtain the dsDNA classification model output; and performing dsDNA post-processing on the dsDNA classification model output to obtain the dsDNA detection result of the fluorescence microscope image, which serves as the structured detection result.

[0137] The cropped image of the ordinary cell region is determined by a square sampling frame, the side length of which is a preset multiple of the long side length of the ordinary cell detection frame in the fluorescence microscope image.

[0138] The output of the dsDNA classification model includes at least the original scores for positive and negative binary classification and the titer regression output.

[0139] The dsDNA test results include at least the positive / negative result, the probability of a positive category, the titer level, and the titer value.

[0140] In specific implementation, when the detection item identifier characterizes the fluorescence microscope image using the dsDNA detection mode, the terminal performs detection processing on the fluorescence microscope image according to the image detection algorithm flow matching the detection item identifier to obtain the structured detection result corresponding to the fluorescence microscope image. This process specifically includes:

[0141] The terminal can crop the fluorescence microscope image into multiple ordinary cell region cropped images according to the cropping strategy corresponding to the dsDNA detection mode. This includes: the terminal inputting the fluorescence microscope image into a pre-trained ordinary cell localization model to obtain the cell localization result of the fluorescence microscope image; and cropping the fluorescence microscope image according to the cell localization result to obtain multiple cell region cropped images.

[0142] The cell localization results include the positional information of ordinary cells in the fluorescence microscope images. In practical applications, under dsDNA detection mode, the terminal can reuse the pre-trained ordinary cell localization model mentioned above. The terminal can input the fluorescence microscope images into the pre-trained ordinary cell localization model to obtain the positional information of ordinary cells in the fluorescence microscope images; the positional information includes bounding boxes, confidence scores, class numbers, and class names.

[0143] Then, for each detected ordinary cell, the terminal can calculate the center point, width, height, and long side length of the detection frame. The terminal can then determine a sampling frame based on the long side length of the detection frame. This sampling frame can be a square, with its side length being a preset multiple of the long side length of the detection frame; in practical applications, this preset multiple can be set to 3. This achieves the construction of a square sampling frame three times the long side length of the detection frame. The terminal can then crop the fluorescence microscope image into multiple cropped images of the ordinary cell region based on this square sampling frame, thus preserving the target subject while retaining necessary background and fluorescence context information around the target.

[0144] Optionally, multiple cropped images (patches) of ordinary cell regions can be filtered. The filtering steps are as follows:

[0145] 1. Sort by detection confidence level from highest to lowest;

[0146] 2. Select the first N cell regions, preferably N=12;

[0147] 3. When the number of detected cells is less than N, use deterministic patch filling or random background patching to supplement the fixed number;

[0148] 4. Zero-padding is applied to sampling regions that extend beyond the image boundaries;

[0149] 5. Scale all patches to a fixed size, preferably 224×224 image size.

[0150] Then, after obtaining multiple cropped images of ordinary cell regions, the terminal can input these images into a pre-trained dsDNA classification model to obtain the dsDNA classification model output; wherein, the dsDNA classification model output includes at least two tensors:

[0151] 1. Logits for binary classification (i.e., the raw scores, unnormalized scores, of binary classification).

[0152] 2. Titer regression output.

[0153] Finally, the terminal can perform dsDNA post-processing on the output of the dsDNA classification model to obtain the dsDNA detection results from the fluorescence microscope images.

[0154] Optionally, dsDNA post-processing is performed on the output of the dsDNA classification model to obtain the dsDNA detection results of the fluorescence microscope images, including: performing SoftMax calculation on the original scores of positive and negative binary classification to obtain the probability of the positive class and the probability of the negative class; when the probability of the positive class is greater than the probability of the negative class, the titer regression output is rounded to a preset level to obtain the titer level, and the titer regression output is mapped to a preset set of titer values ​​to obtain the titer value.

[0155] In practice, after obtaining the positive and negative binary classification logits output by the dsDNA classification model, the terminal can perform SoftMax calculation on the original scores of the positive and negative binary classifications. That is, the positive and negative binary classification logits are input into the SoftMax function to obtain the probability of the positive class and the probability of the negative class.

[0156] When the probability of a positive category is greater than the probability of a negative category, the terminal determines it as positive, limits and rounds the titer regression output to a preset level to obtain the titer level, and maps the titer regression output to one of the preset titer value sets to obtain the titer value. The preset titer value set includes 10, 20, 40, 80, and 160.

[0157] If the probability of a positive class is less than or equal to the probability of a negative class, the terminal determines it as negative and outputs a titer of 0.

[0158] Finally, the dsDNA detection results output by the terminal in dsDNA detection mode include:

[0159] Test Result 1: Positive / Negative Interpretation;

[0160] Test Result 2: Positive Probability;

[0161] Test result 3: Titer grade;

[0162] Test result 4: titer value.

[0163] The technical solution of this embodiment uses a square sampling frame with a side length that is a preset multiple of the long side length of a common cell detection frame to crop the cell region image. This accurately adapts to the cell morphology and recognition requirements of dsDNA detection, ensuring that the cropped region is standardized and uniform. The cropped cell region image is then input into a pre-trained dsDNA classification model, and SoftMax is used to calculate the positive and negative probabilities and perform post-processing on the titer regression output. This achieves accurate determination of dsDNA positive and negative, positive probability, titer level, and titer value, realizing the standardization and automation of the entire dsDNA fluorescence image detection process, and significantly improving the accuracy and standardization of dsDNA detection results.

[0164] In another embodiment, such as Figure 3 As shown, a method for processing ANA and dsDNA fluorescence images is provided, which can be applied to... Figure 1 Taking the terminal in the example, the explanation includes the following steps:

[0165] Step S310: Obtain the fluorescence microscope image to be processed; the fluorescence microscope image is an image obtained by taking a picture of the field of view of the sample to be tested through a fluorescence microscope; the sample to be tested is a sample prepared by indirect immunofluorescence method;

[0166] Step S320: In response to the detection item selection operation of the fluorescence microscope image, determine the detection item identifier associated with the fluorescence microscope image; the detection item identifier is used to characterize whether the fluorescence microscope image is detected in ANA detection mode or dsDNA detection mode.

[0167] Step S330: Assign the fluorescence microscope images to the corresponding image detection pipeline according to the detection item identifier associated with the fluorescence microscope images.

[0168] In the ANA detection mode, the steps include: Step S3301, verifying the effectiveness of green fluorescence on the fluorescence microscope image to obtain a green fluorescence effectiveness verification result; Step S3302, if the green fluorescence effectiveness verification result is successful, cropping the fluorescence microscope image into multiple cell region cropped images according to the cropping strategy corresponding to the ANA detection mode; Step S3303, inputting the multiple cell region cropped images into a pre-trained ANA classification model to obtain the ANA classification model output; Step S3304, performing ANA post-processing on the ANA classification model output to obtain the ANA detection result of the fluorescence microscope image, which is used as the structured detection result.

[0169] In the dsDNA detection mode, the steps include: Step S3305, cropping the fluorescence microscope image into multiple cropped images of ordinary cell regions according to the cropping strategy corresponding to the dsDNA detection mode; Step S3306, inputting the multiple cropped images of ordinary cell regions into a pre-trained dsDNA classification model to obtain the dsDNA classification model output; Step S3307, performing dsDNA post-processing on the dsDNA classification model output to obtain the dsDNA detection result of the fluorescence microscope image, which is used as the structured detection result.

[0170] Step S340: Display the structured detection results corresponding to the images captured by the fluorescence microscope via a display device.

[0171] It should be noted that the specific limitations of the above steps can be found in the above description of the specific limitations of an ANA and dsDNA fluorescence image processing method, which will not be repeated here.

[0172] For the convenience of those skilled in the art, Figure 4A real-time immunofluorescence image interpretation device is provided, comprising: a microscope image input module, an image acquisition and caching module, a main control processing module, an algorithm inference module, a result generation and display module, a communication interface module, a data storage module, and a power supply and heat dissipation module. The microscope image input module is connected to the image acquisition and caching module and is used to receive real-time image frames or single images transmitted from the microscope camera. The image acquisition and caching module is connected to the main control processing module and is used to send image data and item identifiers to the processing queue. The main control processing module is connected to the algorithm inference module and issues inference commands in ANA mode or dsDNA mode to the algorithm inference module. The algorithm inference module is connected to the data storage module and is used to write raw images, intermediate inference results, structured results, and logs. The main control processing module is connected to the result generation and display module and is used to output analysis pages, alarm information, and result reports. The main control processing module is connected to the communication interface module and is used to provide services to external systems via HTTP interface, LAN messages, or local IPC.

[0173] The system comprises the following modules: a microscope image input module for receiving image data from a fluorescence microscope or its camera, accessible via USB, Gigabit Ethernet, direct camera connection, acquisition card, serial bus, or LAN transmission; an image acquisition and caching module for decoding, caching, numbering, and queuing input images to support real-time processing; a main control processing module, which can be an industrial PC, embedded computing unit, CPU+GPU / CPU+NPU computing board, or server motherboard, for task scheduling, project switching, result integration, and peripheral control; an algorithm inference module for performing inference calculations for cell detection models, ANA classification models, and dsDNA classification models, preferably deployed using the ONNX runtime to support multi-threaded real-time processing; a result generation and display module for outputting positive / negative results, karyotype, titer, confidence level, target quantity, anomaly alerts, and visualization overlays; a communication interface module for sending analysis results to a laboratory information system, host computer, browser frontend, microscope control software, or third-party clients; and a data storage module for storing raw images, detection frames, classification results, logs, and traceability information. The power supply and heat dissipation modules are used to ensure the stability of power supply and the safety of temperature control during long-term operation of the equipment.

[0174] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0175] Based on the same inventive concept, this application also provides an ANA and dsDNA fluorescence image processing apparatus for implementing the aforementioned ANA and dsDNA fluorescence image processing method. The solution provided by this apparatus is similar to the solution described in the above method; therefore, the specific limitations in one or more ANA and dsDNA fluorescence image processing apparatus embodiments provided below can be found in the limitations of the ANA and dsDNA fluorescence image processing method described above, and will not be repeated here.

[0176] In one exemplary embodiment, such as Figure 5 As shown, an ANA and dsDNA fluorescence image processing device is provided, comprising:

[0177] The acquisition module 510 is used to acquire the fluorescence microscope image to be processed; the fluorescence microscope image is an image obtained by taking a picture of the field of view of the sample to be tested through the fluorescence microscope;

[0178] Selection module 520 is configured to determine a detection item identifier associated with the fluorescence microscope image in response to a detection item selection operation on the fluorescence microscope image; the detection item identifier is used to identify the detection mode used on the fluorescence microscope image; the detection mode includes one of ANA detection mode and dsDNA detection mode;

[0179] The detection module 530 is used to assign the fluorescence microscope images to corresponding image detection pipelines according to the detection item identifiers associated with the fluorescence microscope images. The image detection pipelines then perform detection processing on the fluorescence microscope images according to the image detection algorithm flow that matches the detection item identifiers, thereby obtaining structured detection results corresponding to the fluorescence microscope images. The structured detection results include the detection results corresponding to each of the multiple detection items that match the detection item identifiers.

[0180] Display module 540 is used to display the structured detection results corresponding to the images captured by the fluorescence microscope via a display device.

[0181] In one embodiment, when the detection item identifier indicates that the fluorescence microscope image is detected using the ANA detection mode, the detection module 530 is used to perform green fluorescence validity verification on the fluorescence microscope image to obtain a green fluorescence validity verification result. The green fluorescence validity verification is used to verify whether the green fluorescence signal in the fluorescence microscope image meets the preset detection conditions. If the green fluorescence validity verification result is successful, the fluorescence microscope image is cropped into multiple cell region cropped images according to the cropping strategy corresponding to the ANA detection mode. The multiple cell region cropped images are input into a pre-trained ANA classification model to obtain the ANA classification model output. The ANA classification model output includes at least global titer classification output, cell type binary classification output, and titer regression output. The ANA classification model output is post-processed to obtain the ANA detection result of the fluorescence microscope image, which is used as the structured detection result. The ANA detection result includes at least positive / negative judgment result, positive category probability, karyotype result, titer corresponding to each karyotype, global titer probability distribution, detection statistics, and detection time information.

[0182] In one embodiment, the detection module 530 is used to input the fluorescence microscope image into a pre-trained cell localization model to obtain the cell localization result of the fluorescence microscope image; the cell localization result includes the position information of ordinary cells in the fluorescence microscope image, and the position information of mitotic cells in the fluorescence microscope image; the mitotic cells are cells in the mitotic stage; according to the cell localization result, the fluorescence microscope image is cropped to obtain the multiple cropped cell region images.

[0183] In one embodiment, the detection module 530 is used to perform SoftMax calculation on the global titer classification output to obtain a titer probability distribution, and to map the titer probability distribution to a preset set of titer values ​​to obtain the global titer probability distribution; to perform sigmoid calculation on the cell type binary classification output to obtain a cell type classification result; the cell type classification result is used to determine the positive or negative judgment result;

[0184] The device is also used to select the cell type with the highest confidence level as positive when the global titer is determined to be higher than a preset threshold and all cell types are not positive, and use this as the positive / negative judgment result.

[0185] The cytotypes include karyotype, cytoplasmic, and independent types. The device is further configured to perform a mutual exclusion solution on the karyotype and cytoplasmic types based on the cytotype classification results, to obtain the winning karyotype type of the karyotype group and the winning cytoplasmic type of the cytoplasmic type group; wherein, the priority rule in the mutual exclusion solution is that the winning result is allowed to be flipped when the probability of the low priority type is significantly higher than that of the high priority type; obtain the independent type positive results of the independent type group, and extract the titer from the titer regression output of the target titer regression head in the ANA classification model based on the winning karyotype, the winning cytoplasmic type, and the independent type positive results, to determine the titer corresponding to each karyotype.

[0186] In one embodiment, when the detection item identifier indicates that the fluorescence microscope image is captured in dsDNA detection mode, the detection module 530 is used to crop the fluorescence microscope image into multiple cropped images of ordinary cell regions according to the cropping strategy corresponding to the dsDNA detection mode; the cropped images of ordinary cell regions are defined by square sampling frames, the side length of which is a preset multiple of the long side length of the ordinary cell detection frame in the fluorescence microscope image; the multiple cropped images of ordinary cell regions are input into a pre-trained dsDNA classification model to obtain the dsDNA classification model output; the dsDNA classification model output includes at least the original scores for positive and negative binary classification and the titer regression output; the dsDNA classification model output is subjected to dsDNA post-processing to obtain the dsDNA detection result of the fluorescence microscope image, which is used as the structured detection result; the dsDNA detection result includes at least the positive / negative judgment result, the positive category probability, the titer level, and the titer value.

[0187] In one embodiment, the detection module 530 is used to perform SoftMax calculation on the original scores of the positive and negative binary classification to obtain the positive class probability and the negative class probability; if the positive class probability is greater than the negative class probability, the titer regression output is rounded to a preset level to obtain the titer level, and the titer regression output is mapped to a preset titer value set to obtain the titer value.

[0188] Each module in the aforementioned ANA and dsDNA fluorescence image processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0189] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 6 As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements an ANA and dsDNA fluorescence image processing method. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0190] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0191] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the ANA and dsDNA fluorescence image processing method described above. The steps of the ANA and dsDNA fluorescence image processing method described here may be steps from the ANA and dsDNA fluorescence image processing methods of the various embodiments described above.

[0192] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, causes the processor to perform the steps of the ANA and dsDNA fluorescence image processing method described above. The steps of the ANA and dsDNA fluorescence image processing method described here may be steps from the ANA and dsDNA fluorescence image processing methods of the various embodiments described above.

[0193] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, causes the processor to perform the steps of the ANA and dsDNA fluorescence image processing method described above. The steps of the ANA and dsDNA fluorescence image processing method described here may be steps from the ANA and dsDNA fluorescence image processing methods of the various embodiments described above.

[0194] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0195] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic resistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0196] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0197] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for processing ANA and dsDNA fluorescence images, characterized in that, The method includes: Acquire fluorescence microscopy images to be processed; the fluorescence microscopy images are images obtained by taking pictures of the field of view of the sample to be tested through a fluorescence microscope; the sample to be tested is a sample prepared by indirect immunofluorescence method; In response to the selection of detection items for the fluorescence microscope image, a detection item identifier associated with the fluorescence microscope image is determined; the detection item identifier is used to characterize whether the fluorescence microscope image is detected in ANA detection mode or dsDNA detection mode. According to the detection item identifier associated with the fluorescence microscope image, the fluorescence microscope image is assigned to the corresponding image detection pipeline. The image detection pipeline performs detection processing on the fluorescence microscope image according to the image detection algorithm flow that matches the detection item identifier, and obtains the structured detection result corresponding to the fluorescence microscope image. The structured detection result includes the detection results corresponding to each of the multiple detection items that match the detection item identifier. The structured detection results corresponding to the images captured by the fluorescence microscope are displayed on a display device.

2. The method according to claim 1, characterized in that, When the detection item identifier indicates that the fluorescence microscope image is detected using the ANA detection mode, the step of processing the fluorescence microscope image according to the image detection algorithm flow matching the detection item identifier to obtain the structured detection result corresponding to the fluorescence microscope image includes: The effectiveness of green fluorescence in the images captured by the fluorescence microscope is verified to obtain the green fluorescence effectiveness verification result; the green fluorescence effectiveness verification is used to verify whether the green fluorescence signal in the images captured by the fluorescence microscope meets the preset detection conditions. If the green fluorescence validity verification result is passed, the fluorescence microscope image is cropped into multiple cell region cropped images according to the cropping strategy corresponding to the ANA detection mode. The multiple cropped cell region images are input into a pre-trained ANA classification model to obtain the ANA classification model output; the ANA classification model output includes at least global titer classification output, cell type binary classification output, and titer regression output; The ANA classification model output is subjected to ANA post-processing to obtain the ANA detection results of the fluorescence microscope image, which are used as the structured detection results. The ANA detection results include at least the positive / negative judgment results, positive category probability, karyotype results, titers corresponding to each karyotype, global titer probability distribution, detection statistics, and detection time information.

3. The method according to claim 2, characterized in that, The step of cropping the fluorescence microscope image into multiple cell-cropped images according to the cropping strategy corresponding to the ANA detection mode includes: The fluorescence microscope images are input into a pre-trained cell localization model to obtain cell localization results for the fluorescence microscope images. The cell localization results include the position information of ordinary cells in the fluorescence microscope images, and the position information of cells in the mitotic phase in the fluorescence microscope images. The cells in the mitotic phase are cells in the mitotic stage. Based on the cell localization results, the images captured by the fluorescence microscope are cropped to obtain the multiple cropped images of the cell regions.

4. The method according to claim 2, characterized in that, The step of performing ANA post-processing on the output of the ANA classification model to obtain the ANA detection results of the fluorescence microscope image includes: The global titer classification output is subjected to SoftMax calculation to obtain the titer probability distribution. The titer probability distribution is then mapped to a preset set of titer values ​​to obtain the global titer probability distribution. The sigmoid function is calculated on the binary cell type output to obtain the cell type classification result; the cell type classification result is used to determine the positive or negative result. The method further includes: If the global titer is determined to be higher than the preset threshold and none of the cell types are positive, the cell type with the highest confidence level is selected as positive and used as the positive / negative judgment result. The cell types include nuclear type group, cytoplasmic type group, and independent type group, and the method further includes: Based on the cell type classification results, a mutual exclusion solution is performed on the karyotype group and the cytoplasmic group to obtain the winning type of the karyotype group and the winning type of the cytoplasmic group; wherein, the priority rule in the mutual exclusion solution is that when the probability of the low priority type is significantly higher than that of the high priority type, the winning result is allowed to be flipped; Obtain the independent positive results of the independent genotype group, and extract the titer from the titer regression output of the target titer regression head in the ANA classification model based on the winning type of the karyotype group, the winning type of the cytoplasmic group, and the independent positive results, and determine the corresponding titer for each karyotype.

5. The method according to claim 1, characterized in that, When the detection item identifier indicates that the fluorescence microscope image is detected using the dsDNA detection mode, the step of processing the fluorescence microscope image according to the image detection algorithm flow matching the detection item identifier to obtain the structured detection result corresponding to the fluorescence microscope image includes: According to the cropping strategy corresponding to the dsDNA detection mode, the fluorescence microscope image is cropped into multiple ordinary cell region cropped images; the ordinary cell region cropped image is determined by a square sampling frame, and the side length of the sampling frame is a preset multiple of the long side length of the ordinary cell detection frame in the fluorescence microscope image; The multiple cropped images of ordinary cell regions are input into a pre-trained dsDNA classification model to obtain the output of the dsDNA classification model; the output of the dsDNA classification model includes at least the original scores for positive and negative binary classification and the titer regression output; The dsDNA classification model output is subjected to dsDNA post-processing to obtain the dsDNA detection results of the fluorescence microscope image, which are used as the structured detection results; the dsDNA detection results include at least the positive / negative judgment result, the positive category probability, the titer level, and the titer value.

6. The method according to claim 5, characterized in that, The step of performing dsDNA post-processing on the output of the dsDNA classification model to obtain the dsDNA detection results of the fluorescence microscope images includes: The original scores of the binary classification were calculated using SoftMax to obtain the probabilities of the positive and negative categories. If the probability of the positive category is greater than the probability of the negative category, the titer regression output is rounded to a preset level to obtain the titer level, and the titer regression output is mapped to a preset set of titer values ​​to obtain the titer value.

7. An ANA and dsDNA fluorescence image processing device, characterized in that, The device includes: The acquisition module is used to acquire fluorescence microscope images to be processed; the fluorescence microscope images are images obtained by capturing the field of view of the sample under test through a fluorescence microscope; The selection module is used to determine the detection item identifier associated with the fluorescence microscope image in response to a detection item selection operation on the fluorescence microscope image; the detection item identifier is used to identify the detection mode used on the fluorescence microscope image; the detection mode includes one of ANA detection mode and dsDNA detection mode; The detection module is used to assign the fluorescence microscope images to corresponding image detection pipelines according to the detection item identifiers associated with the fluorescence microscope images. The image detection pipelines then perform detection processing on the fluorescence microscope images according to the image detection algorithm flow that matches the detection item identifiers, thereby obtaining the structured detection results corresponding to the fluorescence microscope images. The structured detection results include the detection results corresponding to each of the multiple detection items that match the detection item identifiers. The display module is used to display the structured detection results corresponding to the images captured by the fluorescence microscope via a display device.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

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

Images