Digital pathological section scanner control system based on Internet of Things

Abstract

The invention provides a digital pathological section scanner control system based on the Internet of Things. The digital pathological section scanner control system comprises a user management module, a product classification module, an equipment management module, a pathological image data service module, a remote real-time monitoring and operation module and an intelligent fault diagnosis and early warning module. The system has the following beneficial effects: the product classification module realizes efficient classification management of pathological section scanner equipment; an intelligent section type identification sub-module in the pathological image data service module can automatically identify HE staining types and tissue features, and dynamically adjust scanning parameters according to staining quality, so that the stability of scanning quality is improved; a 3D digital slice reconstruction sub-module in the pathological image data service module can construct a three-dimensional organization structure view, so that the accuracy and reliability of pathological diagnosis are greatly improved; the intelligent fault diagnosis and early warning module adopts a machine learning technology to carry out fault prediction and real-time diagnosis, and equipment faults are effectively prevented.

Description

Technical Field

[0001] The present invention relates to the technical field of digital pathological slice scanner control, and particularly to a control system for a digital pathological slice scanner based on the Internet of Things. Background Art

[0002] With the continuous development of medical technology, pathological slice scanners play an increasingly important role in disease diagnosis. However, there are many limitations in the traditional control methods of pathological slice scanners, such as inconvenient operation, inability to achieve intelligent recognition of slice types, inability to remotely monitor, untimely fault diagnosis, insufficient data security, etc. At the same time, most of the current mainstream pathological slice scanners can only provide two-dimensional plane images. When observing, doctors can only obtain partial information of the slices, making it difficult to construct a stereoscopic structure understanding of the tissue. For complex tissue structures and lesion areas, two-dimensional images cannot accurately present their three-dimensional spatial relationships, which makes it difficult for doctors to judge the location, scope, and morphology of lesions, especially in the diagnosis of some diseases that require precise judgment of the depth and level of lesions (such as neurological diseases, etc.), limiting the accuracy and comprehensiveness of diagnosis. Therefore, an innovative control system is needed to solve these problems.

[0003] To overcome these shortcomings, it is necessary to develop and adopt solutions that can achieve intelligent recognition of slice types, 3D model reconstruction, device communication, data sharing, and process automation. By building an integrated medical device platform, the efficiency of pathological diagnosis can be improved, the rational allocation of medical resources can be promoted, and ultimately the overall quality of medical services can be enhanced. Summary of the Invention

[0004] In view of the above-mentioned disadvantages of the prior art, the purpose of the present invention is to provide a control system for a digital pathological slice scanner based on the Internet of Things, which is used to solve the problems of many limitations in the traditional control methods of pathological slice scanners, such as inconvenient operation, inability to achieve intelligent recognition of slice types, inability to perform 3D model reconstruction, inability to remotely monitor, untimely fault diagnosis, insufficient data security, etc.

[0005] To achieve the above purpose and other related purposes, the present invention provides the following technical solutions:

[0006] A digital pathological section scanner control system based on the Internet of Things, comprising a user management module, a product classification module, an equipment management module, a pathological image data service module, a remote real-time monitoring and operation module, and an intelligent fault diagnosis and early warning module; the user management module is used for centralized management of user accounts, and is also responsible for the creation, maintenance and permission allocation of user accounts, as well as monitoring user activities, and also provides user behavior analysis and feedback; the product classification module is used for classifying different types of scanner devices according to preset strategies, then numbering the scanner devices according to the classification results, and creating product files for each scanner device; the equipment management module is used for centralized management and maintenance of all networked scanner devices, and the equipment life cycle management includes equipment list display, equipment addition and deletion, viewing of equipment historical data, and real-time operation of equipment; the pathological image data service module is used for identifying the types of pathological sections, and is also used for 3D model reconstruction of two-dimensional section images, and is also used for centralized management of image data generated by scanning all scanner devices, and can realize the viewing, retrieval, calling and plotting of digital pathological images by users at the network end.

[0007] The remote real-time monitoring and operation module is used for remotely monitoring the working status of scanner devices through the Internet of Things, and is also used for remotely controlling scanner devices to perform scanning operations, and receiving feedback information from scanner devices in real time; the intelligent fault diagnosis and early warning module is used for analyzing the operation data of scanner devices using machine learning algorithms, predicting potential faults according to the analysis results, and sending fault early warnings to relevant staff; wherein, the pathological image data service module includes a slice type intelligent identification sub-module and a 3D digital slice reconstruction sub-module, the slice type intelligent identification sub-module is used for identifying the types of slice images pre-scanned by scanner devices, and obtaining the best scanning parameters according to the identification results, and the 3D digital slice reconstruction sub-module is used for constructing two-dimensional slice images scanned by scanner devices into 3D stereoscopic tissue structure models.

[0008] In an embodiment of the present invention, the classifying different types of scanner devices according to preset strategies, then numbering the scanner devices according to the classification results, and creating product files for each scanner device includes: classifying each scanner device according to product model, product specification and product status; after classification, assigning a unique identification number to each scanner device, and then creating a detailed file for each numbered scanner device; entering the information of each scanner device into the equipment management module.

[0009] In an embodiment of the present invention, the recognition scanner device pre-scans the type of the sliced image and obtains the optimal scanning parameters according to the recognition result. Among them, the sliced types include staining types and tissue types, including: obtaining the key feature vectors extracted by the scanner device from the sliced image, recognizing the type of the sliced image according to the key feature vectors, and performing a staining quality assessment on the sliced image of the recognized type; confirming whether the sliced image is a standard staining according to the assessment result; if so, calling the optimal parameters in the historical database as the optimal scanning parameters, where the optimal parameters include the light source brightness range automatically adjusted according to the staining depth, the exposure time interval optimized based on the tissue type, and the focusing interval standard determined according to the sample characteristics; if not, analyzing the staining abnormality type, calculating the deviation degree from the standard value, and dynamically adjusting the parameters according to the calculated deviation degree to obtain the optimal scanning parameters.

[0010] In an embodiment of the present invention, the calculation of the deviation degree from the standard value includes: calculating the deviation degree from the standard value according to the following formula, including: intensity deviation = (measured value - standard value) / standard value; uniformity deviation = (measured uniformity - standard uniformity) / standard uniformity; Among them, the intensity deviation includes H intensity deviation and E intensity deviation, and the uniformity deviation includes H uniformity deviation and E uniformity deviation. The meaning of H is the basic dye in the HE staining technique, and the meaning of E is the acidic dye in the HE staining technique;

[0011] The dynamic adjustment of the parameters according to the calculated deviation degree to obtain the optimal scanning parameters includes: dynamically adjusting the parameters according to the following formula, including: light source intensity adjustment coefficient = -20*(H intensity deviation + E intensity deviation) / 2; exposure time adjustment coefficient = -10*(0.7*intensity factor + 0.3*uniformity factor); focusing interval adjustment coefficient = -2*(H uniformity deviation + E uniformity deviation) / 2; where the adjustment of the light source intensity is mainly based on the staining intensity deviation; the adjustment of the exposure time mainly comprehensively considers the staining intensity and uniformity; the adjustment of the focusing interval is mainly based on the staining uniformity.

[0012] In an embodiment of the present invention, the construction of a 3D three-dimensional tissue structure model from the two-dimensional slice images scanned by a scanner device includes: obtaining multiple layers of two-dimensional slice images scanned by the scanner device according to optimal scanning parameters; preprocessing the multiple layers of two-dimensional slice images, where the preprocessing includes denoising and contrast enhancement processing; determining the positions and shapes of the isosurfaces in the preprocessed two-dimensional slice images, and connecting the isosurfaces into triangular patches to gradually construct a 3D three-dimensional tissue structure model. Among them, in the process of gradually constructing the 3D three-dimensional tissue structure model, different tissue regions on the 3D three-dimensional tissue structure model are distinguished and labeled according to information such as the tissue characteristics and staining characteristics of the two-dimensional slice images obtained by the slice type intelligent recognition sub-module; the reconstructed 3D three-dimensional tissue structure model is rendered using the ray casting method to simulate the propagation and reflection of light in the tissue structure, and the final 3D three-dimensional tissue structure model is obtained according to the rendering result.

[0013] In an embodiment of the present invention, the determination of the positions and shapes of the isosurfaces in the preprocessed two-dimensional slice images and the connection of the isosurfaces into triangular patches to gradually construct a 3D three-dimensional tissue structure model includes: setting a gray threshold, and after the gray values of the pixel points in adjacent layer slice images satisfy the corresponding relationship with the gray threshold, calculating the intersection coordinates of the isosurface and the voxel edge using a linear interpolation formula; connecting the intersection coordinates into triangular patches according to a pre-set voxel configuration table, where the pre-set voxel configuration table describes the topological structure of the intersection of the isosurface and the voxel under different gray value distributions.

[0014] In an embodiment of the present invention, the distinction and labeling of different tissue regions on the 3D three-dimensional tissue structure model according to information such as the tissue characteristics and staining characteristics of the two-dimensional slice images obtained by the slice type intelligent recognition sub-module includes: on the basis of obtaining the tissue characteristics and staining characteristics of the two-dimensional slice images using the slice type intelligent recognition sub-module, extracting tissue texture features and morphological features, and inputting the tissue texture features and morphological features into a trained tissue classification model to obtain a tissue classification result through the trained tissue classification model; labeling different tissue regions on the 3D three-dimensional tissue structure model with different colors or symbols according to the tissue classification result, where the labeling information is stored in a data structure associated with the three-dimensional model.

[0015] In an embodiment of the present invention, rendering the reconstructed 3D stereoscopic tissue structure model by using the ray casting method, simulating the propagation and reflection of light in the tissue structure, and obtaining the final 3D stereoscopic tissue structure model according to the rendering result, including: determining the virtual view point coordinates and the ray direction vector, generating a ray through the ray parameter equation according to the virtual view point coordinates and the ray direction vector, and emitting the ray from the virtual view point to the three-dimensional model space; traversing all triangular patches in the 3D stereoscopic tissue structure model, calculating the intersection points of the ray and all triangular patches, and obtaining the effective intersection point closest to the virtual view point; calculating the color value of each effective intersection point according to the tissue region where the effective intersection point is located and a preset lighting model, and back-projecting the color value along the ray to the screen pixel to complete the model rendering, presenting a 3D stereoscopic tissue structure view with a three-dimensional sense and a sense of reality, wherein the preset lighting model is: I is the final color value, I a is the ambient light intensity, k a is the ambient light reflection coefficient, I Li is the intensity of the i-th light source, k d is the diffuse reflection coefficient, N is the intersection point normal vector, L i is the unit vector pointing to the i-th light source, k s is the specular reflection coefficient, R i is the reflected ray direction vector, V is the unit vector pointing to the view point, and s is the specular reflection exponent.

[0016] In an embodiment of the present invention, analyzing the operation data of the scanner device by using a machine learning algorithm, predicting potential faults according to the analysis result, and sending a fault warning to relevant staff, including: obtaining the historical fault data and normal operation data of the scanner device, and training a machine learning model by using the data, obtaining a trained machine learning model according to the training result; obtaining various parameter data during the operation of the device sent by the sensors installed on the scanner device, and inputting the various parameter data into the trained machine learning model; confirming whether the various parameter data during the operation of the scanner device is abnormal through the trained machine learning model, if it is abnormal, triggering an alarm, and predicting the future fault risk by combining the historical data and the current state, and automatically triggering an early warning mechanism.

[0017] In an embodiment of the present invention, it further includes a device OTA upgrade module, which is used to upgrade the firmware of the scanner device through OTA, and update the firmware or software of the scanner device based on the wireless network. Among them, the process of upgrading the firmware of the scanner device through OTA includes: the cloud platform adds a firmware upgrade requirement, and the scanner device reports the current firmware version when it wakes up or sends a heartbeat; the cloud system determines whether an upgrade is needed based on the reported version information, and after confirming the need, issues an instruction containing the URL of the upgrade firmware package; after receiving the upgrade URL, the scanner device downloads the firmware package and executes the upgrade; after the scanner device successfully upgrades, it reports the new version information, and after receiving this information, the cloud system updates the status of the scanner device.

[0018] As described above, a digital pathological section scanner control system based on the Internet of Things of the present invention has the following beneficial effects: The product classification module in the present invention realizes the efficient classification management of pathological section scanner devices; the pathological image data service module can centrally manage image data and enhance the data retrieval and analysis capabilities; the slice type intelligent recognition sub-module in the pathological image data service module can automatically identify HE staining types and tissue characteristics, and dynamically adjust scanning parameters according to staining quality, significantly reducing the operation difficulty of technicians and improving the stability of scanning quality; the 3D digital slice reconstruction sub-module in the pathological image data service module can construct a three-dimensional view of the organizational structure, greatly improving the accuracy and reliability of pathological diagnosis. Compared with traditional two-dimensional slice observation, the 3D view can more intuitively present the three-dimensional spatial relationship of the organizational structure, helping doctors more accurately judge the location, scope and morphology of lesions, and avoiding misdiagnosis or missed diagnosis caused by the limitations of two-dimensional images; the device management module and the remote real-time monitoring and operation module can centrally manage and remotely control pathological section scanner devices, greatly improving the maintenance efficiency and use flexibility of the scanner devices; the intelligent fault diagnosis and early warning module uses machine learning technology for fault prediction and real-time diagnosis, effectively preventing equipment failures and ensuring the stability and efficiency of medical services; the OTA device upgrade module supports wireless firmware updates, ensuring the continuous improvement and performance optimization of pathological section scanner devices; the present invention has a high degree of integration, intelligence and scalability, can meet the changing medical needs and technological progress, and is thus convenient for promotion and use. Brief Description of the Drawings

[0019] Figure 1 It shows the overall structural schematic diagram of the digital pathological section scanner control system based on the Internet of Things disclosed in the embodiment of the present invention;

[0020] Figure 2It shows a schematic diagram of the overall structure including sub - modules of the digital pathological slice scanner control system based on the Internet of Things disclosed in the embodiments of the present invention;

[0021] Figure 3 It shows a schematic diagram of the process of classifying scanner devices by the product classification module in the digital pathological slice scanner control system based on the Internet of Things disclosed in the embodiments of the present invention;

[0022] Figure 4 It shows a schematic diagram of the process of optimizing parameters of the slice image obtained by pre - scanning by the intelligent slice type recognition sub - module in the digital pathological slice scanner control system based on the Internet of Things disclosed in the embodiments of the present invention;

[0023] Figure 5 It shows a schematic diagram of the process of reconstructing a 3D model from a two - dimensional slice image by the 3D digital slice reconstruction sub - module in the digital pathological slice scanner control system based on the Internet of Things disclosed in the embodiments of the present invention;

[0024] Figure 6 It shows a schematic diagram of the process of predicting potential faults by the intelligent fault diagnosis and warning module in the digital pathological slice scanner control system based on the Internet of Things disclosed in the embodiments of the present invention;

[0025] Figure 7 It shows a schematic diagram of the process of firmware upgrading of the scanner device by the device OTA upgrade module in the digital pathological slice scanner control system based on the Internet of Things disclosed in the embodiments of the present invention. Detailed implementation manners

[0026] The following uses specific specific examples to illustrate the implementation manners of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. It should be noted that, without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0027] The present invention relates to a digital pathological slice scanner control system based on the Internet of Things. Please refer to Figure 1 and Figure 2 , which includes a user management module, a product classification module, an equipment management module, a pathological image data service module, a remote real - time monitoring and operation module, an intelligent fault diagnosis and warning module, and a device OTA upgrade module. Among them, OTA is Over - The - Air, which means transmission through wireless signals.

[0028] The user management module is used to centrally manage user accounts, responsible for the creation and maintenance of user accounts, permission allocation and management, and monitoring of user activities to ensure the security and compliance of the system. At the same time, it provides user behavior analysis and feedback to optimize system performance and user experience. The user management module includes a user permission sub-module, which is responsible for defining and managing the access permissions of different user roles, such as doctors, technicians, experts, administrators, etc., to ensure that only authorized medical personnel can access and operate the scanner, protecting data security and patient privacy. Through the user permission configuration sub-module, users of different roles can quickly access the functions and services they need, ensuring that resources are reasonably allocated to the users who truly need them, and improving the operation efficiency of the entire system.

[0029] Specifically, the user management module is used to centrally manage user accounts, for ordinary users to register and log in. To start the registration process: the user opens the registration interface, and there are two main registration options provided on the page: email registration and mobile phone number registration. By default, email password registration is shown. After the user enters the login interface, there are three login methods: account password, mobile phone number verification code, and WeChat scan code. By default, account password login is shown. The administrator is responsible for functions such as user list display, editing, role assignment, and deletion. Role management is an important part of permission management, allowing the administrator to view and manage the overview of all roles in the system, which is used to organize and control the permission allocation and functional division of users in the system.

[0030] The product classification module is used to classify and manage different types of scanner devices, providing classification criteria and guidelines to help users select appropriate scanning devices according to medical needs. Among them, the product classification module divides different types of scanner devices according to a preset strategy, and then numbers the products according to the division results, which helps to manage products in a refined manner and improve the accuracy of information collection and data processing.

[0031] Specifically, the preset strategy includes the following steps: First, classify the scanner devices according to product models and product specifications, and at the same time consider product status, such as newness, maintenance cycle, and usage frequency, to ensure the comprehensiveness and practicality of the classification. After the classification is completed, assign a unique identification number to each product. For example, "KScan-1234-A" represents a scanner with a model of KScan, a number of 1234, and a specification of A. At the same time, establish a detailed file for each numbered product, including product description, technical parameters, usage instructions, maintenance records, and product illustrations, etc., to form a complete product information library. For details, please refer to Figure 3 ;

[0032] More specifically, an operation example: Suppose a hospital pathology department needs to classify and manage newly purchased pathological slice scanners. First, according to the preset strategy, classify the pathological slice scanners by model (such as 1.5T or 3.0T), specification (such as open or closed), and status (such as brand new, second-hand); then assign a unique number to each pathological slice scanner, such as "KScan-5678-B", where "KScan" represents the product type, "5678" is the equipment serial number, and "B" represents the 3.0T open specification; then create a product file for the pathological slice scanner, including detailed product descriptions, technical parameters, operation manuals, maintenance logs, and clear product pictures; finally, enter this information into the hospital's equipment management module to facilitate the daily management and operation of the pathology department staff; through such a product classification module, the hospital can achieve refined management of pathological slice scanners, improve equipment usage efficiency, ensure the timeliness of equipment maintenance and updates, and at the same time facilitate the operation and maintenance of the equipment by medical staff.

[0033] The equipment management module is used to centrally manage and maintain all networked scanner devices, mainly serving equipment life cycle management, covering operations such as equipment list display, equipment addition, viewing equipment historical data, real-time operation of equipment, and deleting equipment; user permissions are assigned by the super administrator, and ordinary users can only see and manage the devices bound to them, and can share the equipment usage rights according to needs;

[0034] Specifically, the equipment management module includes an equipment list display sub-module, a real-time operation of equipment sub-module, and an equipment sharing management sub-module; the equipment list display sub-module provides a user interface to display a list of all registered scanner devices, including information such as device status, last maintenance date, etc., and at the same time allows users to view the historical usage and maintenance records of the scanner devices to facilitate the analysis of device usage trends and maintenance requirements; the real-time operation of equipment sub-module realizes remote control of the equipment through Internet of Things technology, including operations such as starting scanning and adjusting settings;

[0035] The device sharing management sub-module allows users to share their bound scanner devices with other users. Through this function, users can simplify the binding process of scanner devices for the shared users and set different access permissions as needed, thereby enhancing the overall user experience. Operation example: User A hopes to share a pathological section scanner device with User B. User A first selects the device from the device list and enters the sharing settings. In the sharing settings, User A selects "valid for one week" as the link validity period and assigns the "read-only" permission to User B through the preset role, ensuring that User B can only view the device status and information and cannot perform any operations. Then User A specifies the recipient by entering User B's mobile phone number or email address and selects to send a notification. After receiving the notification, User B clicks on the link in the notification and enters the password to complete the authentication. After successful authentication, User B enters the sharing information page, which displays the basic sharing information, including the sharer, shared content, and expiration time. User B clicks to receive the device. If not yet registered or logged in, the system guides them to complete the registration or login process. After registration or login, User B clicks to receive the device again. The system updates the sharing status of the device and assigns the corresponding access permissions to User B according to the selected permissions. After successful reception, the device appears in User B's device list and indicates the device source and current permissions, and User B can use the shared device within the authorized permissions.

[0036] The pathological image data service module is used to identify the type of pathological sections, is also used to perform 3D model reconstruction on two-dimensional section images, and is also used to centrally manage all the image data generated by scanning. It can enable users to view digital pathological images on the network side, support efficient retrieval and invocation; display the complete digital pathological section mosaic image, and can achieve stepless zoom of the image from 0 to 40 times, and drag the image in any direction; implement a simple image plotting function, which can draw straight lines, circles, rectangles, irregular closed curves, arrows, write text marks, etc.; support the comparison and viewing of multiple digital results on the same page, can limit the maximum number of images displayed simultaneously, and provide a convenient image sharing function; the purpose of the pathological image data service module is to support the efficient storage, rapid retrieval, data processing and analysis of pathological image data, as well as the integration with other systems; this module needs to have high reliability and scalability to cope with the growing data volume and changing requirements;

[0037] Specifically, the pathological image data service module includes a slice type intelligent recognition sub-module, which comprehensively recognizes pathological slices through three-level intelligent analysis: The first step is to obtain the key feature vectors extracted from the slice images by the scanner device. The slice images are obtained by the user pre-scanning the slices with a pathological slice scanner. The key feature vectors include the color features, texture features, and morphological features of the slice images. The second step is to identify the staining type through a deep learning model, and analyze the uniformity and intensity of the staining in combination with the color histogram. A deep learning model is constructed using the ResNet50 deep residual network model architecture for image classification. By collecting more than 1000 different types of stained slices, labeling the staining types (such as HE, immunohistochemistry, etc.), and then dividing the training set, validation set, and test set, the ResNet50 deep residual network model architecture can be trained to obtain the deep learning model. The color histogram analysis process is to convert from RGB to HSV color space, calculate the hue and saturation histograms, and then extract the color distribution feature vectors. The third step is to use ResNet50 to identify the specific tissue type (also using ResNet50), analyze the tissue structure and cell distribution characteristics, and finally output a comprehensive evaluation result including the slice type, staining characteristics, and tissue characteristics, providing an accurate decision-making basis for subsequent scanning parameter optimization. The comprehensive evaluation result may show that the slice type is "HE staining", the staining characteristic is "uniform staining", and the tissue characteristic is "normal tissue structure".

[0038] More specifically, an operation example: A pathological laboratory technician is preparing to scan a liver slice stained with HE (hematoxylin-eosin). However, due to limited experience, the technician is not sure about the best scanning parameters. First, place the prepared HE (hematoxylin-eosin) stained glass slide in the high-throughput digital pathological slice scanner. The technician finds the relevant device on the front-end interface to enter the operation interface and clicks to start the pre-scanning. At this time, the transmission system inside the device sends the slices to the scanning area in sequence. After all the slices are continuously and automatically scanned, the key feature vectors are extracted and uploaded to the server. The slice type intelligent recognition sub-module in the server combines image processing (including color feature analysis, morphological feature extraction, etc.) and artificial intelligence technology to automatically identify the HE (hematoxylin-eosin) stained liver slice and evaluate the staining quality. When the evaluation result is "standard staining", the optimal parameters in the historical database are called. The optimal parameters include the light source brightness range automatically adjusted according to the staining depth, the exposure time interval optimized based on the tissue type, and the focusing interval standard determined according to the sample characteristics.

[0039] When the evaluation result is "non-standard staining", automatically analyze the type of staining abnormality (such as overstaining, understaining or uneven staining), calculate the degree of deviation from the standard value, and dynamically adjust the parameters accordingly; among them, calculating the degree of deviation from the standard value includes: deviation degree calculation formula: intensity deviation = (measured value - standard value) / standard value; uniformity deviation = (measured uniformity - standard uniformity) / standard uniformity; Among them, the intensity deviation includes H intensity deviation and E intensity deviation, and the uniformity deviation includes H uniformity deviation and E uniformity deviation. In this embodiment, the meaning of H is hematoxylin, and the meaning of E is eosin;

[0040] Dynamically adjusting the parameters accordingly includes: light source intensity adjustment: mainly based on the staining intensity deviation; exposure time adjustment: comprehensively considering the staining intensity and uniformity; focus interval adjustment: mainly based on the staining uniformity; adjustment coefficient calculation: light source intensity adjustment coefficient = -20 * (H intensity deviation + E intensity deviation) / 2; exposure time adjustment coefficient = -10 * (0.7 * intensity factor + 0.3 * uniformity factor); focus interval adjustment coefficient = -2 * (H uniformity deviation + E uniformity deviation) / 2. For example, when there is understaining, increase the light source brightness and extend the exposure time, and at the same time display the staining quality assessment report and processing suggestions to the technician. For details, please refer to Figure 4 ; After the technician confirms the optimal scanning parameters and starts scanning, the system will monitor the image quality in real time and automatically record the scanning process data. For non-standard staining cases, the monitoring frequency will be increased and stricter quality control thresholds will be set; finally, the system receives the slice scanning pictures transmitted by the pathological slice scanner, updates the parameter database, and generates an operation report containing the complete scanning process and parameter adjustment history;

[0041] In addition, the pathological images are automatically uploaded to the server by the scanner device. In this process, the image data will be encapsulated and key metadata will be attached, including device information and patient information, to ensure the integrity and traceability of the data. An automatic trigger upload mechanism is integrated at the device end, and an encrypted network communication protocol is used to ensure data security. Data compression is adopted to reduce the transmission load, and error detection and retransmission are implemented to ensure data integrity. The receiving logic at the server end not only processes image storage, but also updates the database to record the detailed information of the device and the patient; it should be noted that the operation example: when the pathological scan is completed, the system automatically encrypts the image together with the device information and patient information, and transmits it to the server through a secure network. The user interface displays the upload progress in real time; after the upload is successful, the server end automatically stores the image and updates the database record, and at the same time the device end receives the confirmation information; if a transmission error occurs, the system will automatically retry or notify the technician when necessary; doctors can then access these encrypted images at the server end for diagnostic analysis, ensuring the efficiency, security and convenience of the entire process;

[0042] Furthermore, the pathological image data service module further includes a 3D digital slice reconstruction sub-module, which is used to construct a 3D stereoscopic tissue structure model from the two-dimensional slice images scanned by the scanner device; it should be noted that this sub-module first performs data preprocessing on the scanned multi-layer two-dimensional slice images, including operations such as denoising and enhancing contrast to improve the image quality; then uses the Marching Cubes algorithm to perform three-dimensional reconstruction on the preprocessed two-dimensional image data: that is, first determines the position and shape of the isosurface by analyzing the pixel relationship between adjacent slice images, and then connects the isosurfaces into triangular patches to gradually construct a stereoscopic tissue structure model. During the reconstruction process, different tissue regions are distinguished and labeled according to information such as tissue characteristics and staining characteristics for doctors to observe more clearly; at the same time, the ray casting method is used to render the reconstructed three-dimensional model to simulate the propagation and reflection of light in the tissue structure, enhancing the three-dimensional sense and realism of the model. For specific details, please refer to Figure 5 ; The specific operation process is as follows:

[0043] 1. After the scanner device completes scanning according to the optimal scanning parameters, the system automatically detects whether the 3D digital slice reconstruction function is enabled (which can be determined according to user preset or automatically according to the slice type and diagnostic requirements). If it is enabled, the system transfers the two-dimensional slice image data scanned to the 3D digital slice reconstruction sub-module;

[0044] 2. Data preprocessing: a. Denoising: The Gaussian filtering algorithm is used to perform denoising processing on the scanned multi-layer two-dimensional slice images. Gaussian filtering uses the formula (where (x, y) are the pixel coordinates and σ is the standard deviation), through convolution operation with the image, the pixels around each pixel are weighted and summed according to the Gaussian kernel function to obtain the denoised pixel value. Select a suitable value according to the characteristics of the image noise to effectively suppress noise; it should be noted that when performing denoising processing, determine the standard deviation σ value of Gaussian filtering according to the characteristics of the image noise, and calculate each pixel of each layer of two-dimensional slice image using the above Gaussian filtering formula to obtain the denoised image. For example, if the image noise is small, σ = 0.5 can be selected; if the noise is large, the σ value can be appropriately increased, such as σ = 1.5; b. Contrast enhancement: The histogram equalization method is used to enhance the image contrast. First, calculate the gray histogram H(k) of the original image (k is the gray level, and H(k) represents the number of pixels of gray level k), and then calculate the cumulative histogram According to the total number of pixels N of the image, through the formula (Where L is the total number of gray levels, usually 256) Calculate the gray value T(k) after equalization, and finally perform gray mapping on the pixels of the original image according to T(k) to obtain the image with enhanced contrast; it should be noted that when enhancing the contrast, according to the steps of histogram equalization, first count the gray histogram, calculate the cumulative histogram, and then obtain the gray mapping value after equalization, perform gray transformation on the image, enhance the image contrast, and make the tissue structure more clearly distinguishable;

[0045] 3. Use the Marching Cubes algorithm for 3D reconstruction: a. Determine the position and shape of the isosurface: Set the gray threshold T (adaptively adjusted according to tissue type and staining characteristics). For adjacent two-layer slice images, traverse the pixel points; if the gray value I(x, y, z) of a certain pixel point P(x, y, z) and the gray value I′(x, y, z + 1) of the corresponding pixel point P′(x, y, z + 1) in its adjacent layer satisfy (I(x, y, z) - T) × (I′(x, y, z + 1) - T) < 0, then through the linear interpolation formula Calculate the intersection point P interp coordinates; it should be noted that when determining the position and shape of the isosurface, according to the tissue type and staining characteristics, determine the appropriate gray threshold T through a large amount of experimental data. For example, for slices with obvious contrast between normal tissues and diseased tissues, T can be set to an intermediate value so that the isosurface can accurately distinguish different tissue regions, and then according to the gray value relationship of adjacent slice pixel points, use the linear interpolation formula to calculate the intersection point coordinates and determine the isosurface position; b. Construct triangular patches: According to the predefined voxel configuration table (describing the topological structure of the intersection of the isosurface and voxels under different gray value distributions), connect the determined intersection points into triangular patches, and gradually construct a three-dimensional tissue structure model. For example, for cube voxels, connect the intersection points according to the vertex gray value distribution in the table to form triangular patches to approximate the isosurface;

[0046] 4. During the reconstruction process, regional differentiation and annotation are performed based on information such as tissue characteristics and staining characteristics: a. Feature extraction and analysis: Based on the existing staining and tissue type identification, further extract tissue texture features (calculate the gray-level co-occurrence matrix GLCM to analyze texture roughness, directionality, etc.) and morphological features (measure geometric parameters such as area, perimeter, and circularity), and establish a tissue classification model based on these features (such as using support vector machines SVM or deep learning classifiers) to classify different tissue regions; It should be noted that during feature extraction and analysis, the gray-level co-occurrence matrix is ​​calculated to obtain texture features , measure the geometric parameters of the tissue area to obtain the morphological features, and input these features into the trained tissue classification model (such as the SVM model, whose kernel function can select linear kernel, polynomial kernel or Gaussian kernel according to the data characteristics) to obtain the tissue classification results; b. Labeling method: according to the classification results, use different colors or symbols to label different tissue areas on the three-dimensional model. For example, fill the tumor tissue area with red and mark the boundary with a thick red line; fill the normal tissue with green and mark the boundary with a thin green line. The labeling information is stored in the data structure associated with the three-dimensional model for display when the doctor views the model;

[0047] 5. Finally, the reconstructed 3D model is rendered using the ray projection method: a. Ray generation and projection: Rays are emitted from the virtual viewpoint (screen observation angle position) to the 3D model space. The direction of the ray is determined by the viewpoint and screen pixel coordinates. The ray parameter equation is R(t)=E+tD (R(t) is the coordinate of the point on the ray, E is the viewpoint coordinate, D is the ray direction vector, and t is the ray propagation distance parameter). Corresponding rays are generated for each pixel on the screen. It should be noted that when generating and projecting rays, the virtual viewpoint coordinates are determined according to the doctor's observation angle on the screen, the ray direction vector is calculated, and the ray is generated according to the ray parameter equation. For example, when the doctor looks directly at the model, the ray direction vector Pointing perpendicular to the screen and into the model; when the viewing angle is rotated, the direction of the light changes accordingly; b. Calculation of intersection between light and model: Calculate the intersection between light and triangular patches in the three-dimensional model, traverse all patches of the model, solve the simultaneous equations of the light equation and the plane equation where the patch is located, determine the intersection and calculate the intersection coordinates and parameters, and select the intersection point closest to the viewpoint (minimum) as the valid intersection point; It should be noted that when calculating the intersection between light and model, traverse all triangular patches of the model, calculate the intersection between light and patches by accurately solving the equation group, and find the valid intersection point closest to the viewpoint; c. Color calculation and rendering: Calculate the color value according to the tissue area where the intersection point is located and the preset illumination model. The illumination model is: (I is the final color value, I a is the ambient light intensity, k a is the ambient light reflection coefficient, I Li is the intensity of the i-th light source, k d is the diffuse reflection coefficient, N is the normal vector of the intersection point, Li is the unit vector pointing to the i-th light source, k s is the specular reflection coefficient, R i is the reflected light direction vector, V is the unit vector pointing to the viewpoint, s is the specular reflection exponent), calculate the color of each intersection point and back-project it onto the screen pixels to complete the rendering and present a stereoscopic and realistic view; it should be noted that during color calculation and rendering, according to the tissue area where the intersection point is located and the preset lighting model parameters (such as the ambient light intensity I a = 0.3, the diffuse reflection coefficient k d = 0.7, the specular reflection coefficient k s = 0.2, the specular reflection exponent s = 10, etc., these parameters can be adjusted according to the actual observation effect), calculate the color value of the intersection point, back-project the color value along the light ray onto the screen pixels, complete the model rendering, and present a 3D stereoscopic tissue structure view with a sense of three-dimensionality and realism;

[0048] During the 3D reconstruction process, in order to achieve efficient interaction on the web page, the server first converts the rendered 3D model data into a format suitable for the web page (such as the JSON format supported by the Web Graphics Library WebGL). Then, according to the preset template or dynamically generate web page code to create a web page interface specifically for displaying the 3D pathological section model. The layout of this web page interface is reasonable and the functions are complete. Among them, the area for displaying the 3D model occupies the core position. In this area, doctors can perform convenient interaction operations with various input devices such as a mouse, keyboard, or touch screen. For example, use the left mouse button to drag the model to achieve smooth rotation and accurately control the viewing angle; scroll the wheel to easily achieve model scaling and carefully explore the details of the tissue structure; use the right mouse button to drag to flexibly complete model translation and comprehensively browse different parts of the section. These interaction operation instructions will be immediately sent to the server side through the stable and efficient communication connection between the web page and the server. After receiving the instructions, the server immediately performs corresponding transformation processing operations on the 3D model, including but not limited to accurately adjusting parameters such as the rotation angle, scaling ratio, and translation vector of the model, and quickly transmits the updated model data back to the web page. After receiving the new data, the browser immediately re-renders the model to ensure that the accurate effect after the doctor's operation is presented in real time, providing the doctor with the most real and intuitive observation experience.

[0049] Meanwhile, when doctors use the annotation tool to annotate on the model, the annotation information is transmitted through the network to the server for storage and management; the server stores the annotation information associated with the corresponding 3D model data and displays the annotation content in real time on the web page to ensure the synchronous update of the annotation and the model; during the process of doctors observing and analyzing the 3D model, if the server updates the model data (such as optimizing the model according to new diagnostic algorithms or user feedback), the server will actively push the updated data to the web page. After receiving the data, the web page automatically updates the display of the 3D model to ensure that doctors always see the latest and accurate model information; in addition, all user interaction operation data on the web page (such as annotation information, observation perspective records, diagnostic opinions, etc.) will be fed back to the server in real time for storage and management, providing an important basis for subsequent research and improvement of the system;

[0050] Furthermore, the pathological image data service module also includes an image browsing sub-module, a multi-image comparison sub-module, and an image sharing sub-module; Image browsing sub-module: Use OpenLayers (Open Source Web Mapping Library, that is, the open-source web mapping library) to implement image browsing and annotation functions. When the user opens a pathological image, the system first loads the basic view of the image, and then dynamically loads higher-resolution image blocks according to the user's selection to achieve smooth zooming. The user can use the mouse wheel or toolbar buttons to zoom in or out of the image, or drag to view different parts of the image. The zoom level range is 0-40 times; for the annotation function, the user can draw various shapes (such as rectangles, ellipses, freehand drawing, etc.) on the image to mark the area of interest and add text annotations. These annotation information will be saved and associated with the image for subsequent viewing and analysis by the user;

[0051] Implementation method of the multi-image comparison sub-module: In the user interface, multiple image display areas are provided. These areas can be arranged side by side or in a grid pattern to accommodate multi-image display. The number of images for comparison can be customized by the user, with a maximum of no more than 4 images; Image sharing sub-module: The image sharing function provided by the system allows users to share pathological section images in three ways: select multiple pictures in the image list for sharing, click the share button on the single-slice image page to share the current field of view image, and package and share the annotated pathological section and its XML annotation file. The sharing settings include the required sharing name and link validity period, as well as optional description and sharing mechanisms, such as generating a secure sharing link and QR code for the recipient to access the content through password authentication; Users can also choose to share with specific registered users in the system and decide whether to enable the synchronization mode to achieve real-time update of the content; The image sharing list page displays the sharing name, description, and time, and supports filtering, sorting, and pagination functions; Users can view and delete their own shares, while visitors can access and view the image list through the sharing link or scanning the QR code and entering the password, and even save the images to their own accounts.

[0052] The remote real-time monitoring and operation module is used to provide medical staff with the ability to remotely monitor the working status of the scanner device through the Internet of Things, receive feedback information from the scanner device in real time, and support remotely controlling the scanner device to perform scanning operations, including parameter settings, scan start and stop, etc. By remotely operating, it reduces the direct contact between medical staff and the device, improving work efficiency and safety;

[0053] Specifically, the data transmission protocol uses MQTT (Message Queuing Telemetry Transport), a lightweight message publishing / subscribing protocol. The device side installs the MQTT client library and connects to the MQTT broker server through Ethernet. The MQTT broker server acts as the central node and is responsible for message forwarding; The user initiates a control operation on the front-end interface, and the front-end page constructs the corresponding MQTT message and publishes it to the specified topic. The device side subscribes to the corresponding topic and executes the corresponding actions after receiving the control command. In this process, the security and reliability of communication need to be considered, so it is necessary to use TLS / SSL to encrypt the MQTT connection and implement an authentication and authorization mechanism; A reconnection mechanism also needs to be implemented to handle network interruption situations.

[0054] The intelligent fault diagnosis and warning module is used to analyze the operation data of the scanner device using machine learning algorithms, predict potential faults, achieve real-time fault diagnosis, quickly locate problems and provide solutions, and send fault warning notifications to relevant personnel to take preventive measures in advance to prevent device failures;

[0055] Specifically, the specific steps for analyzing the operation data of the scanner device using machine learning algorithms are as follows: Data collection: Install sensors such as temperature sensors and vibration sensors on the key components of the scanner device to collect various parameters during device operation, and transmit the sensor data to the server platform in real time through Internet of Things (IoT) technology; Data preprocessing: Extract meaningful features from the raw data, such as mean values and standard deviations, remove outliers and missing values to ensure data quality, identify and select features or parameters related to device failures, and use historical failure data and normal operation data to train machine learning models, such as classification algorithms, clustering algorithms, or deep learning networks. Use historical data to train the model so that it can identify the differences between normal operation states and potential failures, and evaluate the accuracy of the model through methods such as cross-validation; Utilize stream processing technologies (such as Apache Kafka, Spark Streaming) to process sensor data in real time. When the trained machine learning model detects an abnormal pattern, it will trigger an alarm, and combine historical data and the current state to predict future failure risks, automatically trigger the early warning mechanism, and send failure warning information to maintenance personnel or relevant personnel via email, text message, application notification, etc. For details, please refer to Figure 6 ; Provide an intuitive dashboard for users on the user interface to display device status, failure risks, and solutions, facilitating user monitoring and operation.

[0056] The device OTA upgrade module is used to upgrade the firmware of the scanner device through OTA, and update the firmware or software of the scanner device based on the wireless network; Through OTA upgrade, the functions of the scanner device can be updated, vulnerabilities can be fixed, and performance can be optimized; The device OTA device firmware upgrade is a remote firmware update technology achieved through the wireless network, which allows the device to update functions, fix vulnerabilities, and optimize performance without direct interaction with the user;

[0057] Specifically, this process includes several key steps: First, the cloud platform adds firmware upgrade requirements, and the device reports its current firmware version when it wakes up or sends a heartbeat; The cloud system determines whether an upgrade is needed based on the reported version information, and after confirmation, issues an instruction containing the URL of the upgrade firmware package; After receiving the upgrade URL, the scanner device downloads the firmware package and executes the upgrade. The key status during the upgrade process, such as successful download, will be reported to the cloud in real time; Finally, after the device upgrade is successful, it will report the new version information. After receiving this information, the cloud updates the device status to "upgraded". For details, please refer to Figure 7 ;

[0058] In terms of design, the system pays particular attention to the reporting mechanism of firmware version information, ensuring automatic reporting during scanner device restart or heartbeat to avoid additional periodic reporting events. In addition, when adding a new firmware version to the cloud, it is necessary to ensure that the new version number is higher than the version numbers of all devices running this product to avoid version conflicts. The criterion for judging successful upgrade is that the version number reported by the device is consistent with the version number recorded in the cloud. The system also specially designs two mechanisms for firmware upgrade: link sharing and generating QR codes, enabling users to easily share the firmware upgrade package with others. The recipient can access the upgrade content by clicking the link or scanning the QR code and entering the password. At the same time, the system provides a synchronization mode option, allowing users to choose whether to update in real time when the content changes. The cloud function design further refines the operation process of OTA firmware upgrade. Administrators can add new firmware by selecting the product, naming the firmware, specifying the firmware version number, choosing the firmware type (full or differential firmware), and selecting the signature algorithm. The upload of the firmware file needs to meet specific size and format restrictions. Adding firmware description information helps record the functions and features of the firmware. The firmware list page displays detailed information about all available firmware, including firmware name, product, type, signature method, and creation time, and supports functions such as filtering, sorting, and pagination. Administrators can view, delete, verify, and perform batch upgrades on the firmware. Verifying the firmware function allows administrators to test the new firmware on a small scale to ensure its stability and compatibility and avoid potential impacts on a large number of devices.

[0059] During the verification process, administrators can select the version number to be upgraded and the devices to be verified, and set the device upgrade timeout. Once the verification is passed, the batch upgrade function will be activated, allowing administrators to push the new firmware to a large number of devices. The batch upgrade operation has strict requirements and prohibits operating on test devices or using unvalidated firmware. The system provides detailed upgrade parameter settings, including the version number to be upgraded and the upgrade scope, supports full upgrade or targeted upgrade, and allows users to view the upgrade status and failure reasons on the firmware details page. The entire OTA device firmware upgrade process is carefully designed to achieve efficient, secure, and user-friendly remote device management, ensure that the device always stays up-to-date, and minimize interference with user operations.

[0060] In summary, the product classification module in the present invention realizes the efficient classification management of the pathological slice scanner device; the pathological image data service module can centrally manage image data and enhance the data retrieval and analysis capabilities; the slice type intelligent recognition sub-module in the pathological image data service module can automatically identify the HE staining type and tissue characteristics, and dynamically adjust the scanning parameters according to the staining quality, significantly reducing the operation difficulty of technicians and improving the stability of scanning quality; the 3D digital slice reconstruction sub-module in the pathological image data service module can construct a three-dimensional view of the organizational structure, greatly improving the accuracy and reliability of pathological diagnosis. Compared with the traditional two-dimensional slice observation, the 3D view can more intuitively present the three-dimensional spatial relationship of the organizational structure, helping doctors to more accurately judge the location, scope and morphology of the lesion, and avoiding misdiagnosis or missed diagnosis caused by the limitations of two-dimensional images. In terms of surgical planning, doctors can better understand the anatomical relationship between the lesion and surrounding important tissues and organs through the 3D stereoscopic model, formulate more precise and safe surgical plans, and improve the success rate of surgery. In pathological research, researchers can use the 3D reconstructed images to perform three-dimensional spatial analysis of the organizational structure, deeply study the mechanism of disease occurrence and development, and provide strong support for the development of new diagnostic methods and treatment strategies. In addition, the 3D digital slice reconstruction function has important value in remote consultation. No matter where the doctor is, they can perform accurate diagnosis through the 3D view on the network side, promoting the sharing and optimal allocation of medical resources and improving the accessibility of medical services; the device management module and the remote real-time monitoring and operation module can centrally manage and remotely control the pathological slice scanner device, greatly improving the maintenance efficiency and use flexibility of the scanner device; the intelligent fault diagnosis and early warning module uses machine learning technology for fault prediction and real-time diagnosis, effectively preventing equipment failures and ensuring the stability and efficiency of medical services; the OTA device upgrade module supports wireless firmware updates, ensuring the continuous improvement and performance optimization of the pathological slice scanner device; the present invention has a high degree of integration, intelligence and scalability, can meet the changing medical needs and technological progress, and is thus convenient for promotion and use.

[0061] The above embodiments are only illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. All equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A digital pathology slide scanner control system based on the Internet of Things, characterized by: It includes user management module, product classification module, equipment management module, pathological image data service module, remote real-time monitoring and operation module and intelligent fault diagnosis and early warning module; The user management module is used to centrally manage user accounts and is also responsible for the creation, maintenance and authority allocation of user accounts, as well as monitoring user activities, while also providing user behavior analysis and feedback; The product classification module is used to classify different types of scanner devices according to a preset strategy, then number the scanner devices according to the classification results, and create a product profile for each scanner device; The device management module is used to centrally manage and maintain all networked scanner devices. The device lifecycle management includes device list display, device addition and deletion, device historical data viewing and real-time device operation; The pathological image data service module is used to identify the type of pathological slices, to reconstruct 3D models of two-dimensional slice images, and to centrally manage image data generated by scanning by all scanner devices, so that users can view, retrieve, call and plot digital pathological images on the network end; The remote real-time monitoring and operation module is used to remotely monitor the working status of the scanner device through the Internet of Things, and is also used to remotely control the scanner device to perform scanning operations and receive feedback information from the scanner device in real time; The intelligent fault diagnosis and early warning module is used to analyze the operating data of the scanner device using a machine learning algorithm, predict potential faults based on the analysis results, and send fault warnings to relevant staff; Among them, the pathological image data service module includes a slice type intelligent recognition submodule and a 3D digital slice reconstruction submodule. The slice type intelligent recognition submodule is used to identify the type of slice image obtained by pre-scanning of the scanner device and obtain the optimal scanning parameters based on the recognition result. The 3D digital slice reconstruction submodule is used to construct the two-dimensional slice image obtained by scanning the scanner device into a 3D three-dimensional tissue structure model.

2. According to the control system of the digital pathology slide scanner based on the Internet of Things according to claim 1, it is characterized in that: The method of classifying different types of scanner devices according to a preset strategy, then numbering the scanner devices according to the classification results, and creating product profiles for each scanner device includes: Classify each scanner device according to product model, product specification and product status; after classification, assign a unique identification number to each scanner device, and then create a detailed file for each numbered scanner device; enter the information of each scanner device into the device management module.

3. According to the control system of the digital pathology slide scanner based on the Internet of Things in claim 1, it is characterized by: The identification of the type of slice image obtained by pre-scanning by the scanner device and obtaining the best scanning parameters according to the identification result, wherein the slice type includes the staining type and the tissue type, comprises: Acquire a key feature vector extracted from a slice image by a scanner device, identify the type of the slice according to the key feature vector, and perform staining quality assessment on the slice image of the identified type; confirming whether the slice image is standard stained according to the evaluation result; If so, the optimal parameters in the historical database are called as the optimal scanning parameters, wherein the optimal parameters include the light source brightness range automatically adjusted according to the depth of staining, the exposure time interval optimized based on the tissue type, and the focus interval standard determined according to the sample characteristics; if not, the type of abnormal staining is analyzed, the degree of deviation from the standard value is calculated, and the parameters are dynamically adjusted according to the calculated degree of deviation to obtain the optimal scanning parameters.

4. According to the control system of the digital pathology slide scanner based on the Internet of Things as claimed in claim 3, it is characterized in that: The degree of deviation between the calculation and the standard value includes: The degree of deviation from the standard value is calculated according to the following formulas, including: strength deviation = (measured value - standard value) / standard value; uniformity deviation = (measured uniformity - standard uniformity) / standard uniformity; Among them, the intensity deviation includes H intensity deviation and E intensity deviation, the uniformity deviation includes H uniformity deviation and E uniformity deviation, H means basic dyes in HE dyeing technology, and E means acid dyes in HE dyeing technology; The dynamically adjusting parameters according to the calculated deviation degree to obtain the optimal scanning parameters includes: Dynamically adjust parameters according to the following formulas, including: light source intensity adjustment coefficient = -20*(H intensity deviation + E intensity deviation) / 2; exposure time adjustment coefficient = -10*(0.7*intensity factor + 0.3*uniformity factor); focus interval adjustment coefficient = -2*(H uniformity deviation + E uniformity deviation) / 2; Among them, the light source intensity adjustment is mainly based on the dyeing intensity deviation; the exposure time adjustment mainly takes into account the dyeing intensity and uniformity; and the focus interval adjustment is mainly based on the dyeing uniformity.

5. According to the control system of the digital pathology slide scanner based on the Internet of Things in claim 1, it is characterized by: The method of constructing a 3D tissue structure model from a two-dimensional slice image scanned by a scanner device comprises: Acquire a multi-layer two-dimensional slice image scanned by a scanner device according to optimal scanning parameters; Preprocessing the multiple layers of the two-dimensional slice images, wherein the preprocessing includes denoising and contrast enhancement processing; Determine the position and shape of the isosurface in the preprocessed two-dimensional slice image, and connect the isosurface into triangular patches to gradually construct a 3D three-dimensional tissue structure model, wherein, in gradually constructing the 3D three-dimensional tissue structure model, different tissue regions on the 3D three-dimensional tissue structure model are distinguished and labeled according to information such as tissue features and staining features of the two-dimensional slice image obtained by the slice type intelligent recognition submodule; The reconstructed 3D tissue structure model is rendered using the ray casting method to simulate the propagation and reflection of light in the tissue structure, and the final 3D tissue structure model is obtained based on the rendering results.

6. According to the control system of the digital pathology slide scanner based on the Internet of Things according to claim 5, it is characterized in that: The method of determining the position and shape of the isosurface in the preprocessed two-dimensional slice image and connecting the isosurface into triangular patches to gradually construct a 3D tissue structure model includes: After setting the grayscale threshold and making the grayscale value of the pixel points in the adjacent slice images satisfy the corresponding relationship with the grayscale threshold, the intersection coordinates of the isosurface and the voxel edge are calculated using the linear interpolation formula; The intersection coordinates are connected into triangular facets according to a preset voxel configuration table, wherein the preset voxel configuration table describes the topological structure of intersections between isosurfaces and voxels under different grayscale value distributions.

7. According to the control system of the digital pathology slide scanner based on the Internet of Things in claim 5, it is characterized by: The different tissue regions on the 3D tissue structure model are distinguished and labeled based on the tissue features and staining features of the two-dimensional slice image obtained by the slice type intelligent identification submodule, including: On the basis of obtaining tissue features and staining features of the two-dimensional slice image using the slice type intelligent recognition submodule, extracting tissue texture features and morphological features, and inputting the tissue texture features and morphological features into a trained tissue classification model, and obtaining a tissue classification result through the trained tissue classification model; Different tissue regions are marked on the 3D tissue structure model using different colors or symbols according to the tissue classification result, wherein the marking information is stored in a data structure associated with the three-dimensional model.

8. According to the Internet of Things-based digital pathology slide scanner control system of claim 5, it is characterized by: The reconstructed 3D stereoscopic tissue structure model is rendered using a ray casting method to simulate the propagation and reflection of light in the tissue structure, and the final 3D stereoscopic tissue structure model is obtained according to the rendering result, including: Determine virtual viewpoint coordinates and light direction vectors, generate light through light parameter equations according to the virtual viewpoint coordinates and light direction vectors, and emit light from the virtual viewpoint to the three-dimensional model space; Traversing all triangular facets in the 3D stereoscopic tissue structure model, calculating the intersection points of the light ray with all triangular facets, and obtaining the valid intersection point closest to the virtual viewpoint; The color value of each valid intersection is calculated according to the tissue area where the valid intersection is located and the preset illumination model, and the color value is back-projected along the light to the screen pixel to complete the model rendering, presenting a 3D stereoscopic tissue structure view with a sense of three-dimensionality and realism, wherein the preset illumination model is: I is the final color value, I a is the ambient light intensity, k a is the ambient light reflection coefficient, I Li is the intensity of the i-th light source, k d is the diffuse reflection coefficient, N is the normal vector of the intersection point, L i is the unit vector pointing to the i-th light source, k s is the specular reflection coefficient, R i is the direction vector of the reflected light, V is the unit vector pointing to the viewpoint, and s is the mirror reflection index.

9. The control system of a digital pathology slide scanner based on the Internet of Things according to claim 1, characterized in that: The machine learning algorithm is used to analyze the operating data of the scanner device, predict potential failures based on the analysis results, and send failure warnings to relevant personnel, including: Acquire historical fault data and normal operating data of the scanner device, and use the data to train a machine learning model, and obtain a trained machine learning model based on the training results; acquire various parameter data of the device during operation sent by a sensor installed on the scanner device, and input the various parameter data into the trained machine learning model; confirm through the trained machine learning model whether the various parameter data of the scanner device during operation are abnormal, and if abnormal, trigger an alarm, and predict future fault risks in combination with historical data and current status, and automatically trigger an early warning mechanism.

10. The control system of a digital pathology slide scanner based on the Internet of Things according to claim 1, characterized in that: It also includes a device OTA upgrade module, which is used to upgrade the scanner device firmware through OTA, and update the scanner device firmware or software based on a wireless network; Among them, the scanner device firmware upgrade via OTA includes: the cloud platform adds a firmware upgrade requirement, and the scanner device reports the current firmware version when waking up or sending a heartbeat; the cloud system determines whether an upgrade is required based on the reported version information, and issues an instruction containing the upgrade firmware package URL after confirming the need; after the scanner device receives the upgrade URL, it downloads the firmware package and executes the upgrade; after the scanner device is successfully upgraded, it reports the new version information, and after receiving this information, the cloud system updates the status of the scanner device.

Images