Pathological section auxiliary reading method and system based on multi-path PIS

Abstract

The invention discloses a multi-path PIS-based pathological section auxiliary reading method and system, and relates to the technical field of information and communication for processing medical data. The method comprises the following steps: S1, obtaining a WIS image; s2, based on the number of online PIS terminals, whether frame reduction processing needs to be carried out on the WIS image is judged, if yes, the step S3 is executed, and if not, the step S5 is executed; s3, carrying out frame reduction processing on the WIS image; s4, judging whether the WIS image subjected to frame reduction processing needs to be restored or not, if not, executing the step S5, and if yes, executing the step S5 after restoration processing; and S5, identifying and marking the WIS image by adopting an AI pathological auxiliary diagnosis technology. According to the AI pathology auxiliary diagnosis method, frame reduction processing is carried out on WIS images transmitted by multiple paths of PIS to guarantee the transmission speed, meanwhile, pathological section images after frame reduction are judged, and images needing restoration processing and images not needing restoration processing are processed through different threads respectively to guarantee that the processing efficiency of AI pathology auxiliary diagnosis is good, and the real-time performance is good.

Description

Technical Field

[0001] The present invention relates to the field of information and communication technology for processing medical data, and in particular to a pathological slice auxiliary reading method and system based on a multi-channel PIS. Background Art

[0002] Pathology Information System (PIS) is a part of Hospital Information System (HIS) and is mainly used by the hospital pathology department. PIS can cover the whole process management from patient information entry, pathology application, application registration, fee registration, gross image registration and acquisition, to tracking the slice production process (including inputting gross description and slice information), obtaining pathology images, writing pathology reports, filling out pathology records, sending pathology reports, locking records, etc. Most of the existing PIS systems are open source systems, which makes it easy for different hospitals to adjust their functions according to their own needs.

[0003] Pathology-assisted diagnosis refers to scanning the pathology slice image (WSI), cutting the pathology image into multiple smaller image blocks, and marking and classifying specific areas or structures in the image. The marking and classification process is called annotation. The annotated WSI is convenient for subsequent analysis, research, diagnosis or machine learning model training.

[0004] In the prior art, research on pathological sections mainly focuses on image processing or algorithms, with the main purpose of reducing the error between AI recognition and manual recognition and using AI to help reduce doctors' workload.

[0005] The pathology department of the hospital needs to support the work of different departments. In fact, there are multiple PIS terminals to ensure the needs of various departments for pathological sections. When AI technology is used to assist in the identification of pathological section images in the actual implementation in the hospital, an AI server is introduced to process a large amount of data, which occupies a large bandwidth. Pathological sections usually contain high-resolution image data, so a large amount of bandwidth is required to transmit this data. If the network bandwidth is insufficient, the transmission speed may be slow or even fail. Therefore, under the condition of multiple PIS terminals, the demand for bandwidth is greater. Therefore, in the prior art, when the AI ​​server is deployed for image recognition and marking, the processing efficiency of the AI ​​server is not high enough, and it is impossible to provide efficient feedback to each PIS terminal, and the real-time performance is poor. Summary of the invention

[0006] The purpose of the present invention is to provide a pathology slice assisted reading method and system based on multi-channel PIS, which performs frame reduction processing on the pathology slices transmitted by the multi-channel PIS to ensure the transmission speed, and at the same time judges the pathology slice images after frame reduction, and processes the images that need to be restored and those that do not need to be restored by different threads, so as to ensure the processing efficiency of AI pathology assisted diagnosis and better real-time performance.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is as follows: In one aspect of the present invention, a method for assisting pathological slice reading based on a multi-channel PIS is disclosed, and its implementation includes the following steps: Step S1: Obtain WIS image; Step S2: Based on the number of online PIS terminals, determine whether the WIS image needs to be down-framed, if so, execute step S3, otherwise execute step S5; Step S3: performing frame reduction processing on the WIS image; Step S4: determine whether the WIS image after frame reduction processing needs to be restored, if not, execute step S5, if yes, execute step S5 after restoration processing; Step S5: Using AI pathology-assisted diagnosis technology to identify and mark the WIS image; Step S6: Transmit the WIS image with the identification mark to the display terminal.

[0008] In the present invention, a decision function is used to determine whether the WIS image needs to be downgraded, and the implementation process is as follows: Step S2.1: Calculate the total expected transmission demand and expected processing requirements , calculation method: ; ;

[0009] in, is the number of online PIS terminals, is the average transmission rate required for a single raw WIS image (Mbps); It is The workload of each terminal, ; Step S2.2: Determine by using the first decision function; the first decision function is:

[0010] Among them, B represents the currently available network bandwidth; C represents the maximum number of images that the AI ​​server can process per second; TB represents the network bandwidth threshold ratio; TC: AI server processing capacity threshold ratio.

[0011] In the present invention, a downsampling algorithm is used to perform frame reduction processing on the WIS image, and the specific implementation process is as follows: A multi-resolution image set is generated based on the WIS image. The image set includes several WIS images with different levels of resolution. Except for the highest level image, each level of image is a reduced version of the previous level image.

[0012] In the present invention, the downsampling algorithm is a bicubic interpolation algorithm; the bicubic interpolation algorithm is called to downsample the WIS image to generate a WIS image of the next level, and then the bicubic interpolation algorithm is used to downsample the WIS image of the next level again, until the resolution of the generated image reaches a threshold, and the downsampling is stopped; The detailed implementation process of stopping downsampling is as follows: Step S3.1: downsampling the WIS image; Step S3.2: Calculate the resolution of the current WIS image after downsampling to determine whether it is below the threshold; if yes, execute step S3.3; otherwise, return to step S3. Step S3.3: Return the previous level image and use it as the lowest level image of the image collection.

[0013] In the present invention, in step S4, it is determined whether the WIS image subjected to frame reduction processing needs to be restored. If so, the implementation method for performing the restoration processing is as follows: Step S4.1: starting from the lowest level image in the image collection, obtaining the judgment index of the image; the judgment index includes any one or more of the following: contrast, brightness, color fidelity, noise intensity, sharpness; Step S4.2: Determine whether the image is qualified based on the judgment index. If it is qualified, then there is no need to restore and then execute step S5. Otherwise, execute step S4.3. Step S4.3: Obtain the judgment index of the upper level image acquisition graph; Step S4.4: Determine whether the image is qualified based on the judgment index. If it is qualified, there is no need to restore and execute step S5; otherwise, return to step S4.3.

[0014] In the present invention, the process of determining whether an image is qualified based on the determination index is as follows: The decision function is used to determine whether the image is qualified. The determination process is as follows: It is determined in turn whether several judgment indicators of the image reach the minimum threshold. If all indicators reach the minimum threshold, the image is judged to be qualified. If any indicator does not reach the threshold, it is an unqualified image.

[0015] In the present invention, in step S5, multiple thread pools are established to execute the task of identifying and marking the WIS image.

[0016] In the present invention, a PID controller is used to allocate the number of threads to multiple thread pools, and its specific implementation includes the following process: S5.1: Setting parameters; the parameters include: target CPU utilization, initial number of threads for each task type, and PLD control parameters; S5.2: Monitor the resource status of each thread pool, including the current CPU utilization and current memory usage; S5.3: Calculate the error based on the current CPU utilization and the target CPU utilization; S5.4: Calculate the adjustment amount using the PID formula based on the current error and historical error; S5.5: Calculate a new number of threads based on the calculated adjustment amount, and control the number of threads to be between a set high and low threshold; S5.6: Adjust the number of threads in each thread pool according to the calculation results.

[0017] Another aspect of the present invention discloses a pathological slice auxiliary reading system based on a multi-channel PIS, including a PIS module, a decision module, an image processing module, a first storage module, a display module, an AI server, a resource allocation module, and a second storage module; The PIS module performs the task of obtaining the scanned WIS images of pathological sections; The data collection module is used to collect the number of PIS terminals through heartbeat signals, and collect the judgment indicators of the required level in the image collection composed of different levels of resolution of WIS images; A decision module stores a decision function, determines whether the WIS image is to be downgraded based on the number of PIS terminals collected, and determines whether the WIS image after the downgrade needs to be restored based on the determination index of the WIS image; The image processing module uses a downsampling algorithm to downsample the WIS image to generate a pyramid-level image collection with different resolution levels; each WIS image collection includes: down-frame image, restored image, and original image; A first storage module, used to store image collections; A display module, on which the stored images called from the first storage module are displayed, and the WIS images with identification marks are received and displayed, so as to facilitate the doctor to read the images; An AI server storing an AI image recognition model to perform recognition and marking of WIS images based on AI pathology-assisted diagnosis technology; Resource allocation module, establishes thread pools to handle different image processing tasks, and uses PID controller to allocate the number of threads in each thread pool; The second storage module is used to store the WIS images that have been identified and marked.

[0018] Furthermore, the PIS terminal is equipped with a PIS module, a decision module, a data acquisition module, an image processing module, a first storage module, and a display module; the host computer is equipped with an AI server and a resource allocation module. The present invention has the following beneficial effects: In the present invention, whether the image needs to be "frame-reduced" is determined according to the number of online PIS terminals, so that the amount of data transmission can be flexibly adjusted according to the bandwidth occupancy to meet the network congestion problem under the premise of multiple PIS terminals.

[0019] In the present invention, the pathology scan slice (WIS) image is processed using the image pyramid technology, so that a collection of images containing different levels of resolution is formed for the same pathology slice and stored locally. This processing method does not occupy the existing hospital bandwidth, and the images in the image collection that meet the minimum standard of identification and marking requirements can be transmitted to the AI ​​server for processing. On the basis of ensuring the accuracy of auxiliary diagnosis technology (quality requirements for images), the data transmission volume is reduced and network congestion is avoided. On the premise of ensuring multi-channel PIS terminal data transmission, a large amount of bandwidth can be left for the AI ​​server to perform identification and marking work to ensure efficient data processing of the AI ​​server. In addition, the original-lowest level images with different resolutions can be called out from the local storage medium at any time, which does not occupy bandwidth, ensuring transmission efficiency and data processing efficiency.

[0020] In the present invention, multiple thread pools are established to execute the task of identifying marks, so that resources can be allocated according to the data volume of the image based on PID control to ensure the real-time processing of WIS image annotation analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a flow chart of the auxiliary film reading method in the embodiment; Figure 2 This is a flow chart of downsampling a WIS image in an embodiment; Figure 3 is a flowchart of image restoration in an embodiment; Figure 4 Schematic diagram of the auxiliary film reading system in the embodiment. DETAILED DESCRIPTION

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0023] It should be understood that although the terms first, second, etc. may be used herein to describe various modules, these modules should not be limited by these terms. These terms are only used to distinguish one module from another module. For example, a first module can be referred to as a second module, and similarly, a second module can be referred to as a first module without departing from the scope of the exemplary embodiments of the present invention.

[0024] It should be understood that the term "and / or" that may appear in this document is only a description of the association relationship between associated objects, indicating that there may be three relationships. For example, A and / or B can mean: A exists alone, B exists alone, and both exist. There are three situations of A and B; the term " / and" that may appear in this article describes another kind of relationship between associated objects, indicating that two relationships may exist. For example, A / and B can represent two situations: A exists alone, and A and B exist alone; in addition, the character " / " that may appear in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0025] Embodiment: like Figure 1-Figure 3 As shown, the first aspect of this embodiment provides a pathological slice auxiliary reading method based on a multi-channel PIS, and its implementation includes the following steps: Step S1: Obtain WIS image; In this step, the WIS image is obtained by scanning the pathological section to the PIS terminal.

[0026] Step S2: Based on the number of online PIS terminals, determine whether the WIS image needs to be down-framed, if so, execute step S3, otherwise execute step S5; Specifically, the first decision function is used to determine whether the WIS image needs to be downgraded, and the implementation process is as follows: Step S2.1: Calculate the total expected transmission demand and expected processing requirements , calculation method: ; ; in, is the number of online PIS terminals, is the average transmission rate required for a single raw WIS image (Mbps); It is The workload of each terminal, .

[0027] Step S2.2: Determine by using the first decision function; the first decision function is:

[0028] Among them, B represents the currently available network bandwidth (unit: Mbps); C represents the maximum number of images that the AI ​​server can process per second (unit: images / second); T B Represents the network bandwidth threshold ratio; T C : AI server processing capacity threshold ratio.

[0029] Furthermore, this step collects the online number of PIS terminals through heartbeat signals.

[0030] Step S3: Perform frame reduction processing on the WIS image.

[0031] Specifically, in this step, the WIS image is a high-resolution digital pathology slice scan image, which is not in frames like a video. However, during transmission, a technique similar to "frame reduction" can be used to reduce the amount of data.

[0032] Specifically, the frame reduction method may be to process the WIS image using a compression algorithm, or to establish an image pyramid for the WIS image to reduce the frame.

[0033] In this embodiment, an image pyramid is used as a specific case for illustration: A multi-resolution image collection is created based on the WIS image. The image collection includes several WIS images of different levels of resolution. Except for the highest level image, each level of image is a reduced version of the previous level image. These image collections are stored on the local device.

[0034] In some embodiments, a downsampling algorithm is used to process WIS images, and the processed WIS images are used to construct an image collection.

[0035] Specifically, in this step, the downsampling algorithm is a bicubic interpolation algorithm, and the interface in the Python library is used to call the bicubic interpolation algorithm to downsample the WIS image to generate the next level of WIS image, and then the next level is downsampled again based on the WIS image using the bicubic interpolation algorithm until the resolution of the generated image reaches the threshold and the downsampling is stopped.

[0036] After multiple downsampling processes, a pyramid-like image collection is constructed based on WIS. The highest level of the image collection is the original image, and the image collection is saved to the local device. The reason for using the bicubic interpolation algorithm is that this technology can better preserve image details and is suitable for the processing of pathological slice images.

[0037] The detailed implementation process of stopping downsampling is as follows: Step S3.1: downsampling the WIS image; Step S3.2: Calculate the resolution of the current WIS image after downsampling to determine whether it is lower than the threshold; if yes, execute step S3.3, otherwise return to step S3; Step S3.3: Return the previous level image and store it as the lowest level image of the image collection.

[0038] Step S4: Determine whether the WIS image after frame reduction processing needs to be restored, if not, execute step S5, if yes, execute step S5 after restoration processing.

[0039] In this step, it is determined whether the WIS image processed by frame reduction needs to be restored. If so, the implementation method of the restoration process is as follows: Step S4.1: Starting from the lowest level image in the image collection, obtain the judgment index of the graphic; the judgment index includes but is not limited to any one or more of the following: contrast, brightness, color fidelity, noise intensity, sharpness.

[0040] Step S4.2: Determine whether the image is qualified based on the judgment index. If it is qualified, then there is no need to restore and then execute step S5. Otherwise, execute step S4.3. Step S4.3: Obtain the judgment index of the upper level image acquisition graph; Step S4.4: Determine whether the image is qualified based on the judgment index. If it is qualified, then there is no need to restore and execute step S5; otherwise, return to step S4.3; Furthermore, the process of determining whether an image is qualified based on the determination index is as follows: The second decision function is used to determine whether the image is qualified. The determination process of the second decision function is as follows: It is determined in turn whether several judgment indicators of the image reach the minimum threshold. If all indicators reach the minimum threshold, the image is judged to be qualified. If any indicator does not reach the threshold, it is an unqualified image.

[0041] Specifically, the minimum threshold is the lowest standard of image quality that can be processed by AI pathology-assisted diagnosis technology.

[0042] This embodiment provides a detailed process for obtaining image determination indicators, specifically: (1) The method for obtaining the sharpness of an image is as follows: after converting the image to grayscale, use the cv2.Laplacian function in the OpenCV library to apply the Laplacian filter to the grayscale image and calculate its variance. The variance value represents the sharpness of the image.

[0043] (2) The method for obtaining the contrast of an image is as follows: after converting the image to grayscale, the standard deviation of all pixel values ​​of the grayscale image is calculated. The standard deviation is the performance data of the contrast. The formula for calculating the standard deviation is:

[0044] in, is the standard deviation, is the total number of pixels, is the grayscale value of each pixel, is the average grayscale value of all pixels.

[0045] (3) The method for obtaining the brightness of an image is as follows: after grayscale conversion of the image, the average brightness is calculated using the following formula:

[0046] in, is the total number of pixels, is the grayscale value of each pixel.

[0047] The method for obtaining the color fidelity of an image is: A1: Load the image of the level to be evaluated as the image to be evaluated, and load the highest level image as the reference image; A2: Then resize the two images to the same size; in this step, the reference image is cropped to match the image to be evaluated; A3: Convert the two images from RGB color space to CIELAB color space; A4: Calculate the distance between two colors As a color fidelity index, the calculation method is a prior art and will not be described in detail in this embodiment.

[0048] The method for obtaining the noise intensity of an image is: B1: Load the image of the level to be evaluated as the image to be evaluated, and load the highest level image as the reference image; B2: Call the function from the Python library to calculate the average SSIM value of the image. The specific process is: The two images are resized to a uniform size; the reference image is cropped to match the image to be evaluated; the average SSIM value of the image is calculated as follows: For each pixel in the two images, the SSIM value at that point is calculated through a sliding window of a fixed size, and then all these local window SSIM values ​​are averaged.

[0049]

[0050] in, and Represent the reference image and the image to be evaluated respectively; and In the window and The average brightness of and In the window and The variance of It is in the window and The covariance of and are two constants.

[0051] Step S5: Use AI pathology-assisted diagnosis technology to identify and mark the WIS image.

[0052] In this step, the AI ​​pathology-assisted diagnosis technology is an AI image recognition model. Specifically, the trained YOLO V8 or higher AI computer vision model can be used to perform real-time identification and marking of abnormal targets (malignant glands / myoepithelial cells, etc.) in multiple WIS images.

[0053] In this step, multiple thread pools are established to perform the task of identifying and marking WIS images. In some embodiments, a thread pool is created for down-frame images, and another thread pool is created for restored images and original images. In other embodiments, a thread pool is created for down-frame images, restored images, and original images, respectively.

[0054] The down-frame image refers to the lowest level image processed by the image pyramid, the original image refers to the highest level image not processed by the image pyramid, and the restored image is between the down-frame image and the original image.

[0055] Furthermore, in some embodiments, a PID controller is used to allocate the number of threads to multiple thread pools, and its specific implementation includes the following process: S5.1: Set parameters; the parameters include: target CPU utilization, initial number of threads for each task type (degraded image, restored image, original image), and PLD control parameters (proportional gain Kp, integral gain Ki, differential gain Kd).

[0056] S5.2: Monitor the resource status of each thread pool, including the current CPU utilization and current memory usage; S5.3: Calculate the error based on the current CPU utilization and the target CPU utilization; S5.4: Calculate the adjustment amount using the PID formula based on the current error and historical error; S5.5: Calculate a new number of threads based on the calculated adjustment amount, and control the number of threads to be between a set high and low threshold; S5.6: Adjust the number of threads in each thread pool according to the calculation results.

[0057] Step S6: Transmit the WIS image with the identification mark to the display terminal.

[0058] In this step, the doctor can read the WIS image with identification marks through the display terminal to assist in diagnosis.

[0059] like Figure 4 As shown, the second aspect of this embodiment provides a pathology slice auxiliary reading system based on a multi-channel PIS, including a PIS module, a decision module, an image processing module, a first storage module, a display module, an AI server, a resource allocation module, and a second storage module.

[0060] Specifically, the PIS terminal is equipped with a PIS module, a decision module, a data acquisition module, an image processing module, a first storage module, and a display module. The host computer is equipped with an AI server, a resource allocation module, and a second storage module.

[0061] Specifically, the PIS module is used to perform all functions of the existing pathology information system, including the task of obtaining scanned WIS images of pathology sections.

[0062] The data collection module is used to collect the number of PIS terminals through heartbeat signals and to collect the judgment index of the required level in the image collection composed of different levels of resolution of WIS images; A decision module stores a decision function, determines whether the WIS image is to be downgraded based on the number of PIS terminals collected, and determines whether the WIS image after the downgrade needs to be restored based on the determination index of the WIS image; The image processing module uses a downsampling algorithm to downsample the WIS image to generate a pyramid-level image collection with different resolution levels; each WIS image collection includes: down-frame image, restored image, and original image; A first storage module, used to store image collections; A display module, on which the stored images called from the first storage module are displayed, and the WIS images with identification marks are received and displayed, so as to facilitate the doctor to read the images; An AI server storing an AI image recognition model to perform recognition and marking of WIS images based on AI pathology-assisted diagnosis technology; The resource allocation module establishes a thread pool to handle different image processing tasks, and uses a PID controller to allocate the number of threads in each thread pool.

[0063] The second storage module is used to store the WIS images that have been identified and marked.

[0064] Specifically, the first storage module may be a local memory on the PIS terminal. The memory refers to a carrier for storing data, which may include but is not limited to a hard disk, a flash memory, a USB disk and / or a memory stick.

[0065] The second storage module is a local database or a cloud database, preferably a local database.

[0066] The third aspect of this embodiment provides another pathology section assisted reading device based on a multi-channel PIS. Taking the device as an electronic device as an example, the device includes: a memory, a processor and a transceiver that are communicatively connected in sequence, wherein the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer program to execute the pathology section assisted reading method based on a multi-channel PIS as described in the first aspect of the embodiment.

[0067] For example, the memory may include, but is not limited to, random access memory (RAM), read only memory (ROM), flash memory, first input first output (FIFO) and / or first in last out (FILO), etc. Specifically, the processor may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). At the same time, the processor may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state.

[0068] In some embodiments, the processor may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. For example, the processor may not be limited to a microprocessor of the STM32F105 series, a reduced instruction set computer (RISC) microprocessor, an X86 architecture processor, or a processor with an integrated embedded neural network processing unit (NPU); the transceiver may be, but is not limited to, a wireless fidelity (WIFI) wireless transceiver, a Bluetooth wireless transceiver, a general packet radio service technology (General Packet Radio Service, GPRS) wireless transceiver, a ZigBee protocol (a low-power local area network protocol based on the IEEE802.15.4 standard, ZigBee) wireless transceiver, a 3G transceiver, a 4G transceiver and / or a 5G transceiver, etc. In addition, the device may also include, but is not limited to, a power module, a display screen, and other necessary components.

[0069] The working process, working details and technical effects of the electronic device provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.

[0070] The fourth aspect of this embodiment provides a storage medium storing instructions for the pathological section assisted reading method based on multiple PIS as described in the first aspect of the embodiment, that is, the storage medium stores instructions, and when the instructions are executed on a computer, the pathological section assisted reading method based on multiple PIS as described in the first aspect of the embodiment is executed.

[0071] The storage medium refers to the carrier for storing data, which may include but is not limited to floppy disks, CDs, hard disks, flash drives, etc. storage, USB flash drive and / or memory stick, etc., the computer may be a general-purpose computer, a special-purpose computer, a computer computer network, or other programmable device.

[0072] The working process, working details and technical effects of the storage medium provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.

[0073] A fifth aspect of the present embodiment provides a computer program product comprising instructions, which, when executed on a computer, causes the computer to execute the pathological slice auxiliary reading method based on a multi-channel PIS as described in the first aspect of the embodiment, wherein In the present invention, the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0074] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for assisting pathological slice reading based on multi-channel PIS, characterized in that: The implementation process is as follows: Step S1: Obtain WIS image; Step S2: Based on the number of online PIS terminals, determine whether the WIS image needs to be down-framed, if so, execute step S3, otherwise execute step S5; Step S3: performing frame reduction processing on the WIS image; Step S4: determine whether the WIS image after frame reduction processing needs to be restored, if not, execute step S5, if yes, execute step S5 after restoration processing; Step S5: Using AI pathology-assisted diagnosis technology to identify and mark the WIS image; Step S6: Transmit the WIS image with the identification mark to the display terminal.

2. A method for assisting pathological slice reading based on multi-channel PIS according to claim 1, characterized in that: The decision function is used to determine whether the WIS image needs to be downgraded. The implementation process is as follows: Step S2.1: Calculate the total expected transmission demand and expected processing requirements , calculation method: ; ; in, is the number of online PIS terminals, is the average transmission rate required for a single raw WIS image; It is The workload of each terminal, ; Step S2.2: Determine by using the first decision function; the first decision function is: Among them, B represents the currently available network bandwidth; C represents the maximum number of images that the AI ​​server can process per second; T B Represents the network bandwidth threshold ratio; T C : AI server processing capacity threshold ratio.

3. The method for assisting pathological slice reading based on multi-channel PIS according to claim 1, characterized in that: The downsampling algorithm is used to reduce the frame rate of WIS images. The specific implementation process is as follows: A multi-resolution image set is generated based on the WIS image. The image set includes several WIS images with different levels of resolution. Except for the highest level image, each level of image is a reduced version of the previous level image.

4. A method for assisting pathological slice reading based on multi-channel PIS according to claim 3, characterized in that: The downsampling algorithm is a bicubic interpolation algorithm; the bicubic interpolation algorithm is called to downsample the WIS image to generate the next level of WIS image, and then the bicubic interpolation algorithm is used to downsample the next level based on the WIS image again until the resolution of the generated image reaches the threshold, and then the downsampling stops; The detailed implementation process of stopping downsampling is as follows: Step S3.1: downsampling the WIS image; Step S3.2: Calculate the resolution of the current WIS image after downsampling to determine whether it is lower than the threshold; if yes, execute step S3.3, otherwise return to step S3; Step S3.3: Return the previous level image and use it as the lowest level image of the image collection.

5. The method for assisting pathological slice reading based on multi-channel PIS according to claim 1, characterized in that: In step S4, it is determined whether the WIS image subjected to frame reduction processing needs to be restored. If so, the method for implementing the restoration processing is as follows: Step S4.1: starting from the lowest level image in the image collection, obtaining the judgment index of the image; the judgment index includes any one or more of the following: contrast, brightness, color fidelity, noise intensity, sharpness; Step S4.2: Determine whether the image is qualified based on the judgment index. If it is qualified, then there is no need to restore and then execute step S5. Otherwise, execute step S4.

3. Step S4.3: Obtain the judgment index of the upper level image acquisition graph; Step S4.4: Determine whether the image is qualified based on the judgment index. If it is qualified, there is no need to restore and execute step S5; otherwise, return to step S4.

3.

6. A method for assisting pathological slice reading based on multi-channel PIS according to claim 5, characterized in that: The process of determining whether an image is qualified based on the judgment index is as follows: The decision function is used to determine whether the image is qualified. The determination process is as follows: It is determined in turn whether several judgment indicators of the image reach the minimum threshold. If all indicators reach the minimum threshold, the image is judged to be qualified. If any indicator does not reach the threshold, it is an unqualified image.

7. A method for assisting pathological slice reading based on multi-channel PIS according to claim 1, characterized in that: In step S5, multiple thread pools are established to execute the task of identifying and marking the WIS image.

8. A method for assisting pathological slice reading based on multi-channel PIS according to claim 7, characterized in that: The PID controller is used to allocate the number of threads to multiple thread pools. The process includes: S5.1: Setting parameters; the parameters include: target CPU utilization, initial number of threads for each task type, and PLD control parameters; S5.2: Monitor the resource status of each thread pool, including the current CPU utilization and current memory usage; S5.3: Calculate an error based on the current CPU utilization and the target CPU utilization; S5.4: Calculate the adjustment amount using the PID formula based on the current error and historical error; S5.5: Calculate a new number of threads based on the calculated adjustment amount, and control the number of threads to be between a set high and low threshold; S5.6: Adjust the number of threads in each thread pool according to the calculation results.

9. A pathological slice auxiliary reading system based on multi-channel PIS, characterized in that: It includes a PIS module, a decision module, an image processing module, a first storage module, a display module, an AI server, a resource allocation module, and a second storage module; The PIS module performs the task of obtaining the scanned WIS images of pathological sections; The data collection module is used to collect the number of PIS terminals through heartbeat signals, and collect the judgment indicators of the required level in the image collection composed of different levels of resolution of WIS images; A decision module stores a decision function, determines whether the WIS image is to be downgraded based on the number of PIS terminals collected, and determines whether the WIS image after the downgrade needs to be restored based on the determination index of the WIS image; The image processing module uses a downsampling algorithm to downsample the WIS image to generate a pyramid-level image collection with different resolution levels; Each WIS image collection includes: downscaled image, restored image, and original image; A first storage module, used to store image collections; A display module, on which the stored images called from the first storage module are displayed, and the WIS images with identification marks are received and displayed, so as to facilitate the doctor to read the images; An AI server storing an AI image recognition model to perform recognition and marking of WIS images based on AI pathology-assisted diagnosis technology; Resource allocation module, establishes thread pools to handle different image processing tasks, and uses PID controller to allocate the number of threads in each thread pool; The second storage module is used to store the WIS images that have been identified and marked.

10. A pathological slice auxiliary reading system based on multi-channel PIS according to claim 9, characterized in that: The PIS terminal is deployed with a PIS module, a decision module, a data acquisition module, an image processing module, a first storage module, and a display module; the host computer is deployed with an AI server and a resource allocation module.

Images