Teaching system, method and electronic equipment for sharing digital slices

Abstract

The present invention provides a teaching system, method, and electronic device for sharing digital slices, including: a lecturer end and a student end; the lecturer end is used to receive the original PPT courseware and digital slices uploaded by the lecturer, obtain the slice features of the digital slices through AI feature extraction, and associate the digital slices with the original PPT courseware to obtain the target PPT courseware; when the lecturer is giving a live lecture, the lecturer end is used to respond to the lecturer's first click operation, share the target PPT courseware with the student end for display, obtain the positioning information of the slice features of the digital slices through AI semantic recognition, and synchronize the positioning information to the student end through AI control mode and / or lecturer manual control mode; the student end is used to view the target PPT courseware synchronously with the lecturer end and operate the digital slices. The present invention can share digital slices between the lecturer end and the student end, improving the efficiency of teaching interaction.

Description

Technical Field

[0001] The present invention relates to the technical field of pathology teaching, and in particular to a teaching system, method and electronic equipment for sharing digital slices. Background Art

[0002] With the development of digital pathology technology, pathology teaching has gradually transformed from traditional microscopic observation to digital teaching mode. Traditional pathology teaching relies on physical glass slides. Teachers need to organize teaching by having multiple people view the microscope together or distribute slides one by one. There are problems such as difficulty in sharing slides, high equipment loss rate, and low teaching efficiency. Although digital slide scanning systems have emerged in existing technologies to digitize slides, there are still many technical bottlenecks in actual teaching applications: (1) Live teaching lacks a real-time interactive slide observation mechanism; (2) Video courses cannot achieve intelligent matching of slide operation and teaching rhythm; (3) There is a lack of standardized control interface for collaborative operation between teachers and students; (4) The rapid positioning and sharing of slide features during the teaching process are inefficient. These problems have seriously restricted the popularization of digital pathology teaching. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a teaching system, method and electronic device for sharing digital slices, which can share digital slices between the lecturer and the student, thereby improving the efficiency of teaching interaction.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present invention provides a teaching system for sharing digital slices, comprising: a lecturer end and a student end; the lecturer end is used to receive the original PPT courseware and digital slices uploaded by the lecturer, obtain the slice features of the digital slices through AI feature extraction, and associate the digital slices with the original PPT courseware to obtain the target PPT courseware; when the lecturer is giving a live lecture, the lecturer end is used to respond to the lecturer's first click operation, share the target PPT courseware to the student end for display, and obtain the positioning information of the slice features of the digital slices through AI semantic recognition, and synchronize the positioning information to the student end through AI control mode and / or lecturer manual control mode; the student end is used to watch the target PPT courseware synchronously with the lecturer end and operate the digital slices.

[0006] Optionally, the lecturer obtains the slice features of the digital slice through AI feature extraction, including: extracting the structured information of the digital slice based on the NLP model; wherein the structured information includes at least: specimen location, cancer type; extracting the targeted features of the digital slice based on the convolutional neural network and the visual Transformer model; based on the structured information and targeted features of the digital slice, determining the structured annotation information of the slice features of the digital slice; wherein the structured annotation information includes: feature name and feature outer contour range.

[0007] Optionally, the lecturer associates the digital slices and the PPT courseware, including: in response to the lecturer's drag operation on at least one digital slice, dragging at least one digital slice to the target page area of ​​the PPT courseware, and generating a visual connection line between the digital slice and the target page area; in response to the lecturer's click operation on the visual connection line, setting the association parameters of the digital slice and the target page area; wherein the association parameters include at least: a default display level and a trigger condition; based on the associated digital slices and PPT courseware, establishing a three-dimensional association index table, and verifying the association relationship between the digital slices and the PPT courseware; wherein the three-dimensional association index table includes: PPT courseware ID, page number, digital slice number ID and spatial coordinate anchor point, and the spatial coordinate anchor point is used to represent the position of the digital slice in the associated target page area.

[0008] Optionally, the lecturer obtains the positioning information of the slice features of the digital slice through AI semantic recognition, including: when the lecturer is giving a live lecture, the lecturer's voice stream is collected in real time, and the voice stream is converted into text using an end-to-end speech recognition model; a bidirectional long short-term memory network and a conditional random field combination model is used to identify pathological feature keywords from the text; based on a pre-built feature word-spatial coordinate index library, the pathological feature keywords are converted into coordinate information on the digital slice.

[0009] Optionally, the AI ​​control mode includes: triggering automatic viewport positioning based on AI semantic recognition results, and obtaining coordinate positioning timestamps; sending positioning parameters to the student end through an independent command channel; wherein the positioning parameters include: coordinate information and coordinate positioning timestamps; the instructor manual control mode includes: obtaining the instructor's operation information for digital slices, and generating an operation instruction sequence based on the operation information; synchronizing the operation instruction sequence to the student end, so that the student end synchronizes the instructor's operations.

[0010] Optionally, the lecturer side is also used to: receive video teaching courseware uploaded by the lecturer, and analyze the video content and semantics of the video teaching courseware through AI to obtain the video picture features and audio semantic features of the video teaching courseware; obtain the digital slices uploaded by the lecturer, and associate the digital slices with the video teaching courseware, as well as associate the slice features, digital slice sharing points and synchronous viewing points of the digital slices based on the video picture features and audio semantic features of the video teaching courseware.

[0011] Optionally, the lecturer side is also used to: match the slice feature description information corresponding to the slice feature of the digital slice based on a pre-built pathology teaching knowledge graph, and input the audio semantic features and slice feature description information of the video teaching courseware into the graph neural network to obtain the semantic matching degree between the video teaching courseware and the digital slice; determine the visual similarity between the video teaching courseware and the digital slice based on the video picture features of the video teaching courseware; construct a teaching behavior model, and determine the teaching stage weight based on the teaching behavior model, the video picture features and the audio semantic features of the video teaching courseware; wherein the teaching behavior model is used to describe the teaching scene; perform time calibration on the video teaching courseware and the digital slice to obtain spatiotemporal consistency; perform weighted calculation based on semantic matching, visual similarity, teaching stage weight and spatiotemporal consistency to obtain the association score between the digital slice and the video teaching courseware.

[0012] Optionally, when the student is watching the video teaching courseware, the video teaching courseware is paused in response to the student's pause operation; when the video teaching courseware is paused, the student responds to the student's operation on the digital slice to zoom the digital slice, view the slice features of the digital slice, switch the digital slice, rotate the digital slice, and adjust the contrast of the digital slice.

[0013] In a second aspect, the present invention provides a teaching method for sharing digital slices, which is applied to any one of the teaching systems for sharing digital slices provided in the first aspect, the method comprising: receiving the original PPT courseware and digital slices uploaded by the lecturer through the lecturer end, obtaining the slice features of the digital slices through AI feature extraction, and associating the digital slices with the original PPT courseware to obtain the target PPT courseware; when the lecturer conducts live teaching, the lecturer end responds to the lecturer's first click operation, shares the target PPT courseware to the student end for display, obtains the positioning information of the slice features of the digital slice through AI semantic recognition, and synchronizes the positioning information to the student end through the AI ​​control mode and / or the lecturer manual control mode; the student end and the lecturer end synchronously watch the target PPT courseware and operate the digital slices.

[0014] In a third aspect, the present invention provides an electronic device comprising a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the steps of the method provided in the second aspect above.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method provided in the second aspect are executed.

[0016] The present invention brings the following beneficial effects:

[0017] The above-mentioned teaching system, method and electronic device for sharing digital slices provided by the embodiments of the present invention include: a lecturer end and a student end; the lecturer end is used to receive the original PPT courseware and digital slices uploaded by the lecturer, obtain the slice features of the digital slices through AI feature extraction, and associate the digital slices with the original PPT courseware to obtain the target PPT courseware; when the lecturer conducts live teaching, the lecturer end is used to respond to the lecturer's first click operation, share the target PPT courseware to the student end for display, and obtain the positioning information of the slice features of the digital slices through AI semantic recognition, and synchronize the positioning information to the student end through AI control mode and / or lecturer manual control mode; the student end is used to watch the target PPT courseware synchronously with the lecturer end and operate the digital slices. The above system can locate the slice features of digital slices through AI feature extraction, thereby improving the efficiency of feature positioning; in live teaching, it can realize the real-time conversion of teaching instructions through AI semantic recognition, thereby improving the efficiency of teaching interaction; students can watch PPT courseware and associated digital slices synchronously, and can watch the digital slices, while supporting manual switching, zooming, rotation, contrast adjustment and other operations, thereby improving the teaching effect.

[0018] Other features and advantages of the present invention will be described in the subsequent description, some of which may be directly reflected in the description or clarified through the implementation of the present invention. The objectives and other advantages of the present invention are realized and achieved through the structures specifically described in the description, claims, and drawings.

[0019] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the preferred embodiments are specifically listed below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 A schematic diagram of the structure of a teaching system for sharing digital slices provided by an embodiment of the present invention;

[0022] Figure 2 A schematic diagram of a live class teaching process for a teaching system sharing digital slices provided by an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of a video teaching process of a teaching system for sharing digital slices provided by an embodiment of the present invention;

[0024] Figure 4 A flowchart of a teaching method for sharing digital slices provided by an embodiment of the present invention;

[0025] Figure 5 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0027] Currently, pathology teaching has the following limitations:

[0028] 1. Limitations of physical slicing teaching:

[0029] (1) Difficulty in sharing: Most traditional microscopes only support 2-8 people for simultaneous observation, which cannot meet the needs of large-scale teaching.

[0030] (2) Slices are fragile: The annual loss rate of glass slices is 15%-20%, and the maintenance cost is high (and slices that can be used for teaching are extremely valuable, especially for rare cases. Once the slices are damaged, it may take months or even years to find them again).

[0031] (3) Operation threshold: Differences in students’ proficiency in microscope operation lead to inconsistent teaching progress.

[0032] 2. Disadvantages of existing digital teaching methods:

[0033] (1) One-way transmission mode: Currently, most teaching methods only use one-way display of slice images, lacking two-way interaction between teachers and students.

[0034] (2) Low positioning efficiency: Traditional digital slices require manual search for characteristic areas layer by layer, and the average positioning time is more than 45 seconds per time.

[0035] (3) Lack of intelligent association: There is no dynamic spatiotemporal association between teaching courseware (PPT / video) and slice data. Lecturers need to find the corresponding slices during the lecture, which is time-consuming and laborious.

[0036] 3. Insufficient adaptation to teaching scenarios:

[0037] (1) During live teaching, teachers need to manually operate digital slices to locate the feature points, which affects the fluency of teaching.

[0038] (2) During video teaching, the courseware cannot carry digital slices and locate the explanation content, resulting in a disconnect with the explanation content, affecting the quality of learning.

[0039] (3) Dual-screen display lacks an intelligent layout strategy, and the occlusion rate of interface elements is as high as 30%-40%.

[0040] Based on this, the embodiments of the present invention provide a teaching system, method, and electronic device for sharing digital slices, which can share digital slices between the lecturer and the student, thereby improving the efficiency of teaching interaction.

[0041] To facilitate understanding of this embodiment, a teaching system for sharing digital slices disclosed in an embodiment of the present invention is first introduced in detail. Figure 1 The structural diagram of a teaching system for sharing digital slices is shown, indicating that the teaching system mainly includes: a lecturer end and a student end.

[0042] Among them, the lecturer side is used to receive the original PPT courseware and digital slices uploaded by the lecturer, obtain the slice features of the digital slices through AI feature extraction, and associate the digital slices with the original PPT courseware to obtain the target PPT courseware.

[0043] In one embodiment, the lecturer can log in to the system before the live class to create the PPT courseware required for the class. After the lecturer logs in to the system, he / she uploads the new PPT courseware and digital slices. The lecturer can click on AI feature extraction in the lower right corner of the digital slice upload interface, and then the feature extraction interface pops up. After entering the specimen site and cancer type (such as lung, squamous cell carcinoma), the slice features of the digital slice are extracted. The AI ​​feature extraction function can also be automatically triggered when the lecturer uploads the slice. After entering the specimen site and cancer type, the system automatically analyzes the digital slice, and marks the feature area on the slice with a heat map, and displays the feature name. At the same time, the system also supports manual adjustment of the annotation name and position.

[0044] Furthermore, the system displays a library of uploaded digital slides on the right side of the courseware editing interface (sorted by slide ID, such as 000256358); and thumbnails of the PPT courseware pages (e.g., "Lung Cancer Pathology Teaching Chapter 2.ppt," a 15-page document) on the left. Instructors can drag and drop digital slides to link with the original PPT courseware. Once linked, the target PPT courseware is created, and the instructor can publish the course.

[0045] When the lecturer is giving a live lecture, the lecturer side is used to respond to the lecturer's first click operation, share the target PPT courseware to the student side for display, and obtain the positioning information of the slice features of the digital slice through AI semantic recognition, and synchronize the positioning information to the student side through AI control mode and / or lecturer manual control mode.

[0046] In one implementation, during a live lecture, the instructor can activate a pre-created PowerPoint presentation and switch pages during the course, sharing digital slides with students. In practice, sharing digital slides includes synchronized viewing and delegated viewing. Synchronized viewing allows students to view digital slides simultaneously with the instructor, including synchronization of any actions the instructor performs on the slides with the students. Delegated viewing allows students to manually manipulate the digital slides, such as zooming, switching slides, rotating, and adjusting contrast.

[0047] During the simultaneous viewing process, the instructor can achieve precise synchronization between the instructor's operations and the student's digital slide display through intelligent semantic recognition and real-time data distribution technology. Specifically, it can adopt a dual mode of AI automatic positioning and instructor manual control to ensure the real-time observation and operational consistency of pathology slides during teaching.

[0048] The student side is used to watch the target PPT courseware synchronously with the lecturer side and operate the digital slices.

[0049] In one implementation, after entering a live course, students can simultaneously view the target PPT courseware. Specifically, this can be done through a dual-screen model: the PPT courseware is viewed on a small screen and the digital slides are viewed on a large screen. Furthermore, students can manipulate the digital slides while the instructor delegates viewing authority.

[0050] The above-mentioned teaching system for sharing digital slices provided by the embodiment of the present invention can locate the slice features of digital slices through AI feature extraction, thereby improving the efficiency of feature positioning; in live teaching, it can realize real-time conversion of teaching instructions through AI semantic recognition, thereby improving the efficiency of teaching interaction; the students can watch PPT courseware and associated digital slices synchronously, and can watch the digital slices, while supporting manual switching, zooming, rotation, contrast adjustment and other operations, thereby improving the teaching effect.

[0051] In one embodiment, AI feature extraction is divided into three stages: information preprocessing, targeted feature extraction, and structured annotation output and display. Based on this, the instructor obtains the slice features of the digital slice through AI feature extraction, including the following steps 1 to 3:

[0052] Step 1: Extract structured information of digital slices based on the NLP model; the structured information includes at least: specimen location and cancer type.

[0053] In practice, we first extract structured information using an NLP model, extracting key fields (i.e., structured information) from digital slides. This includes: specimen location (breast, lung, prostate, etc.); cancer type (adenocarcinoma, squamous cell carcinoma, small cell carcinoma, etc.). We then calibrate the extracted results by matching them against a cancer type database. For example: ("breast": ["invasive ductal carcinoma", "lobular carcinoma", "DCIS"], ​​"lung": ["adenocarcinoma", "squamous cell carcinoma", "small cell carcinoma"]).

[0054] Step 2: Extract the targeted features of digital slices based on convolutional neural network and visual Transformer model.

[0055] In the specific implementation, a convolutional neural network (CNN) and a visual transformer (ViT) are used to extract targeted features in combination with comparison. The main steps include:

[0056] (1) Global feature positioning.

[0057] Specifically, through low-power microscopy (5x / 10x) analysis, the characteristics of tissue architecture destruction (such as disappearance of breast lobular structure), tumor boundary morphology (invasive / expanding growth) and distribution pattern characteristics of necrotic areas are obtained.

[0058] (2) Local key area detection.

[0059] Specifically, ROI extraction was performed using a high-power microscope (40x), including: using the Attention mechanism to locate abnormal cell clusters; using instance segmentation to identify individual tumor cells (nuclear-to-cytoplasmic ratio > 0.8); and quantifying the mitotic index (counting per 10 HPF).

[0060] (3) Detection of specific markers.

[0061] Specifically, a signature library is loaded for specific cancer types and specific markers are detected. For example, for lung adenocarcinoma, acinar formation and TTF-1 immunohistochemical simulation are examined; for hepatocellular carcinoma, abnormal bile duct plate structure and AFP expression patterns are examined.

[0062] Step 3: Based on the structured information and targeted features of the digital slice, determine the structured annotation information of the slice features of the digital slice; wherein the structured annotation information includes: feature name and feature outer contour range.

[0063] During the specific implementation, the system will dynamically load the structured annotation information output by AI and display it on the interface simultaneously. Among them, the structured annotation output by AI includes the following:

[0064] (1) Name of the characteristic feature (e.g., squamous cell carcinoma characteristic - keratinized pearls);

[0065] (2) Feature outer contour range; such as polygon

[0066] [{"x":7,"y":237},{"x":35,"y":1},{"x":34,"y":2},{"x":34,"y":16},{"x":1,"y":60},{"x":22,"y":73},{"x":66,"y":123} ,{"x":95,"y":154},{"x":120,"y":169},{"x":144,"y":169},{"x":169,"y":169},{"x":185,"y":171},{"x":199,"y":171}]);

[0067] Heatmap details (such as blue (#1E9FFF)

[0068] [{"x":7,"y":237},{"x":35,"y":1},{"x":34,"y":2},{"x":34,"y":16},{"x":1,"y":60},{"x":22,"y":73},{"x":66,"y":123} ,{"x":95,"y":154},{"x":120,"y":169},{"x":144,"y":169},{"x":169,"y":169},{"x":185,"y":171},{"x":199,"y":171}])).

[0069] In one embodiment, when the lecturer associates the digital slices with the original PPT courseware, the drag-and-drop association mechanism used includes:

[0070] First, in response to the lecturer's dragging operation on at least one digital slice, the at least one digital slice is dragged to a target page area of ​​the PPT courseware, and a visual connection line between the digital slice and the target page area is generated.

[0071] During specific implementation, the lecturer drags the thumbnail of the digital slice to the target page area of ​​the PPT courseware (such as page 5). The system also supports the simultaneous dragging and associating of multiple slices, that is, the lecturer can drag and associate multiple digital slices at a time. A single page of PPT can be associated with 3-5 main and supplementary slices, including HE and immunohistochemistry slices, and set the priority mark of the digital slice, including: main slice (displayed by default); supplementary slice (need to click the expansion button to display).

[0072] After dragging and dropping is completed, the system automatically generates visual connection lines and highlights the associated areas of the digital slices and PPT courseware.

[0073] Then, in response to the lecturer's click operation on the visual connection line, the association parameters of the digital slice and the target page area are set; wherein the association parameters at least include: a default display level and a trigger condition.

[0074] During implementation, instructors can right-click on the connection line to set associated parameters, including: default display level (40% transparency overlay / independent window pop-up), trigger conditions (automatically load the page after 3 seconds or activate by clicking the icon).

[0075] Finally, based on the associated digital slices and PPT courseware, a three-dimensional association index table is established, and the association relationship between the digital slices and PPT courseware is verified; among them, the three-dimensional association index table includes: PPT courseware ID, page number, digital slice number ID and spatial coordinate anchor point, and the spatial coordinate anchor point is used to represent the position of the digital slice in the associated target page area.

[0076] In specific implementation, after the digital slices are associated with the PPT courseware, a three-dimensional association index table will be established in the system's data association layer, including: PPT courseware ID, page number, digital slice number ID and spatial coordinate anchor point. Such as:

[0077] PPT courseware ID: P2024001;

[0078] Page number: 5;

[0079] Digital slide ID: 000256358;

[0080] Space coordinate anchor point: x:320,y:180.

[0081] The spatial coordinate anchor point definition rules include:

[0082] (1) Convert the PPT page to a 1024×768 pixel coordinate system;

[0083] (2) Record the location of the slice icon (e.g. coordinates 320,180);

[0084] (3) Store the initial display magnification of the slice (such as the level corresponding to a 20× objective lens).

[0085] Furthermore, the system can also perform association verification and detect abnormal associations in real time, such as: repeated association warnings for a single slice number on the same PPT page; invalid drag and drop interception of slice files that have not been uploaded or have been physically deleted, etc.

[0086] In this embodiment of the present invention, the visual association between PPT pages and digital slides can be recorded using an SVG vector layer, and a relative coordinate storage scheme can be used to adapt to display on devices with different resolutions. Furthermore, a bidirectional search mechanism can be used to search for slides or PPT teaching nodes, including: forward search: retrieving associated slides and initial view parameters based on the PPT page number; and reverse search: searching all associated PPT teaching nodes based on the slide number.

[0087] In one implementation, to ensure real-time and operational consistency of pathology slide observation during teaching, the system's real-time communication layer can establish a bidirectional communication channel based on the WebSocket protocol, supporting low-latency command transmission. The data distribution layer can utilize a hierarchical content delivery network (CDN), with GPU-accelerated slice rendering servers deployed on primary nodes and high-frequency slice data cached on edge nodes. The state management module can record the slice ID, viewport parameters (zoom ratio, coordinates, rotation angle, etc.), and operation sequence version number of the current shared session.

[0088] When the instructor clicks the "Share Digital Slice" button, the system sends a sharing request to the server via a RESTful interface, carrying the course ID, slice UUID, and initial viewport parameters. After the server verifies permissions, it broadcasts the sharing start command to all online students and continuously listens for synchronization requests from newly connected students.

[0089] In its implementation, the system utilizes a real-time data synchronization mechanism and a multi-level caching strategy to enhance system performance. This real-time data synchronization mechanism includes incremental transmission, which transmits only the difference in viewport changes (such as coordinate offsets and zoom adjustments), and uses a binary protocol to compress data packets. The multi-level caching strategy includes: Students locally cache the base-level data of loaded slices; edge nodes pre-cache frequently accessed areas of the instructor's current slice; and a dynamic preloading algorithm predicts the next viewport location based on the instructor's operational trajectory.

[0090] In one embodiment, the lecturer obtains the positioning information of the slice features of the digital slice through AI semantic recognition, including: first, when the lecturer is giving a live lecture, the lecturer's voice stream is collected in real time, and the voice stream is converted into text using an end-to-end speech recognition model; then, a bidirectional long short-term memory network and conditional random field combination model is used to identify pathological feature keywords from the text; finally, based on a pre-built feature word-spatial coordinate index library, the pathological feature keywords are converted into coordinate information on the digital slice.

[0091] In specific implementation, AI semantic recognition includes: (1) Speech recognition: Real-time acquisition of the lecturer's speech stream and conversion to text through an end-to-end speech recognition model. (2) Medical entity extraction: Using a combined model of a bidirectional long short-term memory network (BiLSTM) and a conditional random field (CRF) model, pathological feature keywords (such as "lymphocyte aggregation area") are identified from the text. (3) Coordinate mapping: Using a pre-built feature word-spatial coordinate index library, semantic features are converted into precise coordinates on the slice, triggering automatic viewport positioning.

[0092] Furthermore, the present invention utilizes both AI control and instructor manual control to ensure real-time and consistent pathology slide observation during instruction. The AI ​​control mode includes triggering automatic viewport positioning based on AI semantic recognition results, obtaining coordinate positioning timestamps, and sending positioning parameters to the student via an independent command channel. The positioning parameters include coordinate information and coordinate positioning timestamps.

[0093] Specifically, the voice command recognition result triggers the viewport to automatically jump, the system records the coordinate positioning timestamp, and sends the positioning parameters through an independent command channel. This method takes precedence over the normal operation data packet transmission.

[0094] The instructor's manual control mode includes: obtaining the instructor's operation information on the digital slices, and generating an operation instruction sequence based on the operation information; synchronizing the operation instruction sequence to the student end so that the student end synchronizes the instructor's operation.

[0095] Specifically, the instructor's manual operations (dragging, scaling, rotating, etc.) are captured, a standardized operation instruction sequence is generated, and an operation log versioning mechanism is adopted to ensure the eventual consistency of out-of-order data packets.

[0096] In one embodiment, the system may also continuously monitor synchronization requests from new students. The synchronization process for new students includes:

[0097] (1) Detect the current sharing status flag when the student terminal accesses;

[0098] (2) Obtain basic slice data and the latest viewport parameter snapshot from the server;

[0099] (3) Parallel loading of historical operation logs to achieve status catch-up;

[0100] (4) Enter the real-time instruction stream monitoring state and complete seamless access.

[0101] In addition, the system has optimized the student disconnection and reconnection process, which can retain a circular buffer of the operation logs of the last 5 minutes and request incremental updates based on the last received instruction ID when reconnecting.

[0102] The above-mentioned system provided by the embodiment of the present invention controls the end-to-end operation delay within 500ms when the lecturer and the student end watch the digital slices synchronously, supports viewport parameter synchronization of 4K-level digital slices, and the coordinate error is less than 5 pixels; a single instance can carry 500+ terminals concurrent synchronization operations; and complete status synchronization is completed within 1 second after a new student joins.

[0103] The above-mentioned teaching system for sharing digital slices provided by the embodiment of the present invention can also perform video teaching. The lecturer uploads the video teaching courseware on the lecturer side, and the students can watch the video teaching courseware on the student side to learn.

[0104] Based on this, the lecturer side is also used to: receive the video teaching courseware uploaded by the lecturer, and analyze the video content and semantics of the video teaching courseware through AI to obtain the video picture features and audio semantic features of the video teaching courseware; obtain the digital slices uploaded by the lecturer, and associate the digital slices with the video teaching courseware, as well as associate the slice features, digital slice sharing points and synchronous viewing points of the digital slices based on the video picture features and audio semantic features of the video teaching courseware.

[0105] In one implementation, instructors can upload video courseware on the instructor's end, and AI analyzes the video content and semantics. Specifically, the system accepts mainstream video formats (MP4, AVI, MOV, etc.) and automatically transcodes them into the unified H.264 encoding format to optimize streaming performance. Using Dynamic Adaptive Streaming over HTTP (DASH) technology, videos are cut into 2-5 second segments for parallel processing and rapid location. Basic video attributes (resolution, duration, frame rate) are automatically extracted to generate a video fingerprint for deduplication verification.

[0106] In specific implementation, the multimodal AI analysis process includes:

[0107] (1) Visual content analysis.

[0108] First, we detect shot cuts using a dynamic thresholding method, extracting 1-3 representative frames (keyframes) from each scene. We then use a ResNet-50 model to encode image features and generate a 128-dimensional feature vector. Next, we use a YOLOv7 model to identify teaching tools in the video (laser pointer trajectory, microscope operation gestures). Finally, we use the PaddleOCR engine to capture embedded text in the video (such as courseware titles and pathology diagnosis conclusions).

[0109] (2) Phonetic semantic analysis.

[0110] Specifically, the system deploys an end-to-end speech recognition model (such as Conformer) and BERT fine-tuned for the medical field. It supports mixed Chinese and English recognition. The end-to-end speech recognition model outputs a time-stamped text stream with word-level time alignment accuracy of ±0.3 seconds. First, speech recognition is performed based on the end-to-end speech recognition model to generate text; then, medical semantic analysis is performed on the text, including:

[0111] 1) Constructing a domain knowledge graph: Integrate the pathology terminology library (ICD-O, SNOMED CT standards) to obtain a domain knowledge graph.

[0112] 2) Semantic role labeling: The BERT-BiLSTM-CRF model is used to identify teaching key points and label semantic roles.

[0113] 3) Entity extraction: Locate entities such as pathological feature words (such as "adenocarcinoma differentiation zone") and anatomical sites.

[0114] 4) Action recognition: Parsing teaching instructions (such as: "Zoom in this area", "Compare to normal tissue").

[0115] (3) Spatiotemporal correlation modeling.

[0116] Specifically, first, a cross-modal attention mechanism is established to align speech text with visual features (for example, when identifying "cell abnormalities here", the corresponding video image is associated with it), generate a spatiotemporal heat map, and mark areas of high teaching value in the video (such as the slice area repeatedly mentioned by the lecturer).

[0117] Then, the timeline is marked, including: automatic marking of potential teaching nodes, voice emphasis points (sudden increase in volume, slowdown in speaking speed), visual focus events (areas where the laser pen stays for more than 3 seconds), courseware page turning timestamps, etc.

[0118] After the above AI analysis, the teaching semantic map of the video is output, including video image features and audio semantic features. Such as:

[0119] {

[0120] "timepoint": "00:02:15",

[0121] "Visual focus": {"coordinates": [x1,y1,x2,y2], "type": "Pathology slide display"},

[0122] "Semantic tags": ["squamous cell carcinoma", "high-power microscope observation"],

[0123] "Associated operations": ["Zoom in 1.5 times", "Switch to H&E stained sections"],

[0124] }).

[0125] The AI ​​analysis method provided by the embodiment of the present invention has the following advantages:

[0126] (1) Multi-granularity parsing capability: capable of simultaneously processing pixel-level visual features and word-level semantic features;

[0127] (2) Understanding teaching intentions: Identifying potential teaching intentions through the command-action mapping model (e.g., "pay attention to this area" → marking the key area);

[0128] (3) Real-time processing performance: 1080P video parsing speed reaches 1.2 times real-time (full analysis of a 30-minute video is completed within 18 minutes);

[0129] (4) Explainable output: Generate a visual analysis report to show the teaching logic link recognized by AI.

[0130] Furthermore, after the lecturer uploads the video teaching courseware, the following steps are also included:

[0131] (1) Upload digital slices and associate them with video time points, and use AI to extract the slice features of the digital slices. Specifically, the instructor can manually associate the uploaded digital slices with specific video time points (for example, select the slice number ID000256358, then select the associated video file "Video Teaching Materials for Chapter 2 of Lung Cancer Pathology Teaching", and then select the slice from 24 minutes and 15 seconds to 29 minutes and 30 seconds).

[0132] (2) Associating slice features through AI. Specifically, when extracting features from the video teaching courseware, the semantic feature {lung squamous cell carcinoma - different nuclei size} is extracted; the system will automatically search for matching and associated digital slice features, and generate an associated map to locate the corresponding position. The association is completed after the instructor confirms it.

[0133] (3) Manual or AI-automated setting of digital slice sharing points. Specifically, when extracting features from video teaching courseware, if the semantic feature {see slice} is extracted or the video image feature switches from {case information to digital slice}, the system will automatically search for matching and associated digital slice sharing points. The setting is completed after the instructor confirms.

[0134] (4) AI automatically sets the synchronization viewing point. Specifically, when extracting video teaching courseware features, if the video screen features {digital slices appear dragged, enlarged, etc.} are extracted; the system will automatically set the current slice viewing position as the synchronization viewing point. The setting is completed after the instructor confirms it.

[0135] In specific implementation, the aforementioned associated slice features, digital slice sharing points, and synchronized viewing points extract and recognize the audio semantic features and video image features of the video, and take corresponding actions based on the extracted content. Specifically, the process includes the following:

[0136] (1) Data input and feature alignment.

[0137] Specifically, the input source feature library includes: ① Video node features: timestamp (accurate to milliseconds), visual focus coordinates (ROI region in the video), semantic label sets (pathological feature words, operation instructions), and speech emotion intensity (quantified by the lecturer's emphasis through voiceprint analysis). ② Digital slide features: morphological feature vectors (cell density, staining intensity, tissue structure), spatial coordinate metadata (feature point location in the full slide coordinate system), and hierarchical multi-resolution features (regional salience at 5X / 10X / 20X / 40X magnification).

[0138] Feature alignment includes time-space reference alignment: establishing a unified time axis to synchronize video timestamps with slice operation logs at the millisecond level; building a spatial coordinate system mapping system, including: video screen coordinate system (X / Y axis pixel position), digital slice full-frame coordinate system (based on the physical coordinates of the slice scanner) feature space relative coordinate system (local feature coordinates with specific pathological structures as the origin).

[0139] (2) Cross-modal association.

[0140] 1) Semantic level association.

[0141] First, a knowledge graph for pathology teaching is constructed, which includes: the mapping relationship between disease types and typical slice features, and the association rules between teaching scenarios and common operation modes (such as "explaining cancerous areas" → high-magnification observation); then, a graph neural network (GNN) is used for semantic reasoning. The input of the GNN is the semantic label in the video (i.e., video semantic features) + slice feature description; the output is the association confidence score (range 0-1).

[0142] 2) Visual feature association.

[0143] Specifically, visual feature extraction and intent analysis include page transition detection. Differential analysis is used to capture changes in visual features when switching video pages, including:

[0144] Content type analysis: comparison of the image-text ratio on the before and after pages (case text page: text ratio > 60% → digital slide page: image ratio > 90%);

[0145] Structural feature extraction:

[0146] Text page features: detect title bars, bullet points, and text paragraph boundaries;

[0147] Slice page features: identification of full-screen image containers, digital slice thumbnail matrices (e.g., 4x4 slice preview layouts);

[0148] Transition marker recognition: Capture page switching effects (such as "fade in and fade out" animations) and build a probability model for switching intentions.

[0149] 3) Dynamic correlation between time and space.

[0150] Specifically, a teaching behavior model is constructed. A hidden Markov model (HMM) is constructed to describe a typical teaching scenario. The model includes:

[0151] State space: {slice overview → feature focus → comparative analysis → conclusion emphasis};

[0152] Observation sequence: operation rhythm + speech semantic changes in the video;

[0153] Dynamically adjust the association weights according to the current teaching stage (e.g., prioritize the association of diagnostic features in the "conclusion emphasis" stage).

[0154] Furthermore, dynamic time axis calibration is performed, including: detecting the deviation of operation rhythm between video and slices; applying the dynamic time warping (DTW) algorithm to optimize the time correspondence; and generating a time correlation matrix with confidence intervals.

[0155] (3) Association rule engine.

[0156] Specifically, a multi-factor decision model is used to calculate the association score through a weighted scoring function:

[0157] Association score = α*semantic matching + β*visual similarity + γ*spatiotemporal consistency + δ*teaching stage weight.

[0158] In specific implementation, the coefficients can be dynamically adjusted according to the following rules: (1) When an operation instruction (such as "zoom in here") is detected in the video, the spatial coordinate weight is significantly increased; (2) When a strong association rule in the knowledge graph appears, the semantic weight is increased to 70%.

[0159] When a single video node corresponds to multiple slice features, the most relevant features are selected based on the teaching stage (for example, "beginner mode" prioritizes typical features); high-frequency related items are selected based on historical data; and a confidence threshold alarm is triggered (manual confirmation is requested when it is <0.6).

[0160] (4) Output of associated data.

[0161] Specifically, the structured association relationship table includes:

[0162] Video timestamp: 00:05:23.450;

[0163] Slice UUID: 000256358;

[0164] Association type: semantically dominant;

[0165] Confidence level: 0.92;

[0166] Spatial mapping parameters: [x1,y1,x2,y2].

[0167] Furthermore, an interactive spatiotemporal relationship map can be generated, including:

[0168] Time dimension: heat map of correlation density on the teaching progress axis;

[0169] Spatial dimension: projection mapping of slice feature points and video ROI area;

[0170] Semantic dimension: pathological concept association network (including weights).

[0171] The above association method provided by the embodiment of the present invention has the following advantages:

[0172] (1) Three-dimensional association system: simultaneously satisfying the triple constraints of temporal synchronization, spatial correspondence, and semantic consistency;

[0173] (2) Dynamic teaching perception: intelligently adjust the association strategy according to the real-time teaching status;

[0174] (3) Multi-granularity verification: providing cross-scale verification from cell level (20X) to tissue level (5X);

[0175] (4) Self-optimization capability: Continuously improve the association accuracy through teaching feedback data.

[0176] Based on this, in an embodiment of the present invention, the lecturer end is also used to: first, match the slice feature description information corresponding to the slice feature of the digital slice based on the pre-constructed pathology teaching knowledge graph, and input the audio semantic features and slice feature description information of the video teaching courseware into the graph neural network to obtain the semantic matching degree between the video teaching courseware and the digital slice; then determine the visual similarity between the video teaching courseware and the digital slice based on the video picture features of the video teaching courseware; then construct a teaching behavior model, and determine the teaching stage weight based on the teaching behavior model, the video picture features and the audio semantic features of the video teaching courseware; specifically, the system can evaluate the importance of the content based on the attention weight, automatically generate a teaching focus timeline, and calculate the teaching node weight; thereafter, perform time calibration on the video teaching courseware and the digital slice to obtain spatiotemporal consistency; finally, perform weighted calculation based on semantic matching, visual similarity, teaching stage weight and spatiotemporal consistency to obtain the association score between the digital slice and the video teaching courseware.

[0177] In one embodiment, while watching a video courseware, a student can pause the video courseware to access the unauthorized viewing mode and perform operations on the digital slices. Specifically, when watching the video courseware, the student's pause operation is responded to by the student; while the video courseware is paused, the student can respond to the student's operations on the digital slices to zoom in and out of the digital slices, view the slice features of the digital slices, switch the digital slices, rotate the digital slices, and adjust the contrast of the digital slices.

[0178] For ease of understanding, the present invention also provides a teaching process for sharing digital slices. Figure 2 and Figure 3 As shown in the flowchart, it mainly includes the following steps:

[0179] For live classes, the instructor's pre-class preparation includes:

[0180] Step 1: Log in to the system.

[0181] Step 2: Add courseware.

[0182] Step 2.1: Upload the PPT courseware.

[0183] Step 2.2: Upload digital slides.

[0184] Step 2.3: Manually associate digital slices to courseware pages.

[0185] Step 2.4: AI digital slice feature extraction.

[0186] Step 2.5: Complete the courseware production.

[0187] Step 3: Publish the course.

[0188] The instructor course includes:

[0189] Step 4: Start the class / activate the courseware.

[0190] Step 5: Switch the courseware page.

[0191] Step 6: Share the digital slice.

[0192] Among them, shared digital slices include: synchronized viewing and delegated viewing.

[0193] Simultaneous viewing includes:

[0194] Step 6.1: AI recognizes the instructor's semantics and locates feature points.

[0195] Step 6.2: The instructor manually switches feature points and changes the film.

[0196] Step 7: End of the course.

[0197] It should be noted that the live class system has the function of automatically recording and converting it into video teaching materials. It can automatically record video teaching materials during the live class, and recognize the instructor's operation of sharing digital slices during the live broadcast to automatically set sharing points and synchronous viewing points.

[0198] The student side pre-class includes:

[0199] Step 1: Register / Log in to the system.

[0200] Step 2: Find out the course times.

[0201] Step 3: Enter the live broadcast room.

[0202] The student-side course includes:

[0203] Step 4: Watch the courseware, including:

[0204] Dual-screen mode: small screen for courseware and large screen for digital slices.

[0205] Manipulate digital slices: zoom, switch slices, rotate, adjust contrast.

[0206] Step 5: End of study.

[0207] For video teaching materials, the instructor's pre-class preparation mainly includes:

[0208] Step 1: Log in to the system.

[0209] Step 2: Add video teaching courseware.

[0210] Step 2.1: Upload the video and AI will analyze the video content and semantics.

[0211] Step 2.2: Upload digital slices, manually associate video time points, and extract AI digital slice features.

[0212] Step 2.3: AI associates slice features.

[0213] Step 2.4: Set the digital slice sharing point manually or automatically by AI.

[0214] Step 2.5: AI automatically sets the synchronized viewing point.

[0215] Step 2.6: Complete the courseware production.

[0216] Step 3: Publish the course.

[0217] The student side pre-class includes:

[0218] Step 1: Register / Log in to the system.

[0219] Step 2: Find out the course times.

[0220] Step 3: Enter the live broadcast room.

[0221] The student-side course includes:

[0222] Step 4: Watch the courseware, including:

[0223] Dual-screen mode: small screen for courseware and large screen for digital slices.

[0224] Pausing the video allows students to watch the video and manipulate the digital slices, including zooming, clicking to view feature points, switching slices, rotating, and adjusting contrast.

[0225] Step 5: End of study.

[0226] The above-mentioned system provided by the embodiment of the present invention, through the deep integration of multimodal interaction architecture and artificial intelligence algorithms, has achieved significant breakthroughs in the following three dimensions, effectively solving the key bottleneck problems existing in the existing technology:

[0227] (1) Improved teaching interaction efficiency.

[0228] Feature positioning efficiency: AI-assisted positioning algorithm can shorten the time of slice feature positioning from 45 seconds per time of traditional manual operation to ≤3 seconds per time.

[0229] Real-time collaborative response speed: millisecond-level synchronization of teacher and student operation instructions is achieved through standardized API interfaces, and interaction delay is ≤120ms.

[0230] Resource matching efficiency: The courseware-slice dynamic association engine reduces the time required to prepare teaching resources from the traditional 2-3 hours per class to 18-25 minutes, increasing efficiency by 5.2 times.

[0231] (2) Hardware resource optimization.

[0232] Slice reuse rate: By replacing physical slices with digital slices for teaching, the annual loss rate of precious slices can be reduced from 15%-20% to 0.

[0233] Hardware compatibility: supports cross-platform device access (including PC and mobile).

[0234] (3) Enhanced adaptation to teaching scenarios.

[0235] Live teaching: The intelligent semantic analysis module realizes real-time conversion of teaching voice commands, with a response accuracy rate of 92.4%; the dual-screen dynamic layout algorithm reduces the interface element occlusion rate from 30%-40% to 6.8%.

[0236] Video teaching: The timestamp-slice feature association module achieves frame-level synchronization between courseware content and digital slices; the pause-trigger authorized viewing function enables students to actively operate digital slices and achieve all-round observation.

[0237] Regarding the teaching system for sharing digital slices provided in the above embodiment, the present invention also provides a teaching method for sharing digital slices, see Figure 4 The flowchart of a teaching method for sharing digital slices is shown, which illustrates that the method mainly includes the following steps S401 to S403:

[0238] Step S401: receiving the original PPT courseware and digital slices uploaded by the lecturer through the lecturer end, obtaining the slice features of the digital slices through AI feature extraction, and associating the digital slices with the original PPT courseware to obtain the target PPT courseware.

[0239] Step S402: When the lecturer is giving a live lecture, the lecturer's end responds to the lecturer's first click operation, shares the target PPT courseware to the student end for display, and obtains the positioning information of the slice features of the digital slice through AI semantic recognition, and synchronizes the positioning information to the student end through AI control mode and / or lecturer manual control mode.

[0240] Step S403: The student terminal and the lecturer terminal synchronously watch the target PPT courseware and operate the digital slices.

[0241] The above-mentioned teaching method of sharing digital slices provided by the embodiment of the present invention can locate the slice features of digital slices through AI feature extraction, thereby improving the efficiency of feature positioning; in live teaching, it can realize real-time conversion of teaching instructions through AI semantic recognition, thereby improving the efficiency of teaching interaction; the students can watch PPT courseware and associated digital slices synchronously, and can manually slice the digital slices, thereby improving the teaching effect.

[0242] It should be noted that the implementation principles and technical effects of the methods provided in the embodiments of the present invention are the same as those of the aforementioned system embodiments. For the sake of brevity, any details not mentioned in the method embodiments can be referred to the corresponding contents of the aforementioned system embodiments. The specific numerical values ​​provided in the implementation of the present invention are merely exemplary and are not intended to be limiting.

[0243] An embodiment of the present invention further provides an electronic device. Specifically, the electronic device includes a processor and a storage device. The storage device stores a computer program, and when the computer program is executed by the processor, it executes the method described in any one of the above embodiments.

[0244] Figure 5 This is a structural diagram of an electronic device provided in an embodiment of the present invention. The electronic device 100 includes: a processor 50, a memory 51, a bus 52 and a communication interface 53. The processor 50, the communication interface 53 and the memory 51 are connected via the bus 52; the processor 50 is used to execute an executable module stored in the memory 51, such as a computer program.

[0245] The memory 51 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage. The system network element communicates with at least one other network element via at least one communication interface 53 (which may be wired or wireless), and may utilize the Internet, a wide area network, a local area network, a metropolitan area network, or the like.

[0246] The bus 52 may be an ISA bus, a PCI bus, or an EISA bus. The bus may be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, Figure 5 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0247] Among them, the memory 51 is used to store programs, and the processor 50 executes the program after receiving the execution instruction. The method executed by the device for flow process definition disclosed in any embodiment of the above-mentioned embodiment of the present invention can be applied to the processor 50 or implemented by the processor 50.

[0248] The processor 50 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method may be completed by hardware integrated logic circuits or software instructions in the processor 50. The processor 50 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It may implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present invention may be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or the like. The storage medium is located in the memory 51 , and the processor 50 reads the information in the memory 51 and completes the steps of the above method in combination with its hardware.

[0249] The computer program product of the readable storage medium provided in the embodiment of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the method described in the previous method embodiment. The specific implementation can be referred to the previous method embodiment and will not be repeated here.

[0250] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0251] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the above-described embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-described embodiments within the technical scope disclosed by the present invention, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A teaching system for sharing digital slices, characterized by: include: Instructor side and student side; The lecturer end is used to receive the original PPT courseware and digital slices uploaded by the lecturer, obtain the slice features of the digital slices through AI feature extraction, and associate the digital slices with the original PPT courseware to obtain the target PPT courseware; When the lecturer is giving a live lecture, the lecturer terminal is used to respond to the lecturer's first click operation, share the target PPT courseware to the student terminal for display, and obtain the positioning information of the slice features of the digital slice through AI semantic recognition, and synchronize the positioning information to the student terminal through AI control mode and / or lecturer manual control mode; The student terminal is used to watch the target PPT courseware synchronously with the lecturer terminal and operate the digital slices.

2. The system according to claim 1, wherein: The lecturer obtains the slice features of the digital slice through AI feature extraction, including: Extracting structured information of digital slices based on the NLP model; wherein the structured information includes at least: specimen location and cancer type; Extracting targeting features of the digital slice based on a convolutional neural network and a visual Transformer model; Based on the structured information and the targeting feature of the digital slice, structured annotation information of the slice feature of the digital slice is determined; wherein the structured annotation information includes: a feature name and a feature outer contour range.

3. The system according to claim 1, wherein: The lecturer end associates the digital slice with the PPT courseware, including: In response to the lecturer's dragging operation on at least one digital slice, dragging at least one digital slice to a target page area of ​​the PPT courseware, and generating a visual connection line between the digital slice and the target page area; In response to the lecturer's click operation on the visual connection line, setting association parameters between the digital slice and the target page area; wherein the association parameters at least include: a default display level and a trigger condition; Based on the associated digital slices and PPT courseware, a three-dimensional association index table is established, and the association relationship between the digital slices and the PPT courseware is verified; wherein, the three-dimensional association index table includes: PPT courseware ID, page number, digital slice number ID and spatial coordinate anchor point, and the spatial coordinate anchor point is used to represent the position of the digital slice in the associated target page area.

4. The system according to claim 1, wherein: The lecturer obtains the positioning information of the slice features of the digital slice through AI semantic recognition, including: When the lecturer is giving a live lecture, the lecturer's voice stream is collected in real time and converted into text using an end-to-end speech recognition model; A bidirectional long short-term memory network and a conditional random field combination model are used to identify pathological feature keywords from the text; Based on a pre-built feature word-spatial coordinate index library, the pathological feature keywords are converted into coordinate information on the digital slice.

5. The system according to claim 4, characterized in that The AI ​​control modes include: Trigger automatic viewport positioning based on AI semantic recognition results and obtain coordinate positioning timestamps; Sending positioning parameters to the student terminal through an independent command channel; wherein the positioning parameters include: coordinate information and coordinate positioning timestamp; The instructor manual control mode includes: Acquiring the lecturer's operation information on the digital slice, and generating an operation instruction sequence based on the operation information; The operation instruction sequence is synchronized to the student terminal so that the student terminal synchronizes the operation of the lecturer.

6. The system according to claim 1, wherein: The lecturer terminal is also used to: Receive the video teaching courseware uploaded by the lecturer, and analyze the video content and semantics of the video teaching courseware through AI to obtain the video picture features and audio semantic features of the video teaching courseware; The digital slices uploaded by the lecturer are obtained, and the digital slices are associated with the video teaching courseware, and the slice features, digital slice sharing points and synchronous viewing points of the digital slices are associated based on the video picture features and audio semantic features of the video teaching courseware.

7. The system according to claim 6, characterized in that The lecturer terminal is also used to: Matching the slice feature description information corresponding to the slice feature of the digital slice based on a pre-built pathology teaching knowledge graph, and inputting the audio semantic features of the video teaching courseware and the slice feature description information into a graph neural network to obtain the semantic matching degree between the video teaching courseware and the digital slice; Determining the visual similarity between the video teaching courseware and the digital slice based on the video picture features of the video teaching courseware; Constructing a teaching behavior model, and determining teaching stage weights based on the teaching behavior model, the video image features, and the audio semantic features of the video teaching courseware; wherein the teaching behavior model is used to describe the teaching scenario; Performing time calibration on the video teaching courseware and the digital slices to obtain spatiotemporal consistency; A weighted calculation is performed based on the semantic matching degree, the visual similarity, the teaching stage weight and the spatiotemporal consistency to obtain an association score between the digital slice and the video teaching courseware.

8. The system according to claim 7, characterized in that When the student terminal is watching the video teaching courseware, in response to a pause operation by the student, the video teaching courseware is paused; When the video teaching courseware is paused, the student terminal responds to the student's operation on the digital slice to zoom the digital slice, view the slice features of the digital slice, switch the digital slice, rotate the digital slice, and adjust the contrast of the digital slice.

9. A teaching method for sharing digital slices, characterized in that: The teaching system for sharing digital slides according to any one of claims 1 to 8, wherein the method comprises: The lecturer receives the original PPT courseware and digital slices uploaded by the lecturer through the lecturer terminal, obtains the slice features of the digital slices through AI feature extraction, and associates the digital slices with the original PPT courseware to obtain the target PPT courseware; When the lecturer is giving a live lecture, the lecturer terminal responds to the lecturer's first click operation by sharing the target PPT courseware to the student terminal for display, and obtains the positioning information of the slice features of the digital slice through AI semantic recognition, and synchronizes the positioning information to the student terminal through AI control mode and / or lecturer manual control mode; The student terminal and the lecturer terminal synchronously watch the target PPT courseware and operate the digital slices.

10. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the steps of the method according to claim 9.

Images