Medical information intelligent auditing method and device based on large model
By using a large-model-based intelligent medical information review method, which compares medical images and text information using a parsing module and a review module, the problem that deep convolutional models cannot review information on their own is solved, and more accurate medical information review is achieved.
Patent Information
- Application Number
- CN202510920872.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-04
- Publication Date
- 2025-11-21
AI Technical Summary
Existing deep convolutional models cannot independently verify the accuracy of medical information and need to be improved.
A medical information intelligent review method based on a large model is adopted. By acquiring medical image data and corresponding text information, the parsing module and review module in the medical review large model are compared to generate review results, including consistency assessment, error prompts and modification suggestions.
It improves the accuracy of medical information review, makes up for the deficiency that deep convolutional models cannot review themselves, and provides more accurate diagnostic and identification results.
Smart Images
Figure CN120998523A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of medical image processing technology, and more specifically, to a method and apparatus for intelligent review of medical information based on a large model. Background Technology
[0002] Current medical images are used for diagnosis, recognition, and planning, but the accuracy of this medical information needs to be improved. However, the deep convolutional models that output the above information cannot verify it themselves.
[0003] Therefore, there is an urgent need for a method to review medical information in order to overcome the above-mentioned deficiencies. Summary of the Invention
[0004] To address the aforementioned issues, the first aspect of this application provides a medical information intelligent review method based on a large model, which includes:
[0005] Acquire medical image data and corresponding text information;
[0006] The medical image data and the text information are input into the medical review model to obtain the review result; the medical review model includes a purpose parsing module for parsing text information and a review module for comparing text information.
[0007] The second aspect of this application provides a manufacturing system for a medical information intelligent review method based on a large model, which includes:
[0008] The data acquisition module is used to acquire medical image data and corresponding text information;
[0009] The medical review module is used to input the medical image data and the text information into the medical review model to obtain the review result; the medical review model includes a purpose parsing module for parsing text information and a review module for comparing text information.
[0010] A third aspect of this application provides an electronic device, including: a memory and a processor; the memory being configurable to store a program, and the processor being coupled to the memory for executing the program in the memory for:
[0011] Acquire medical image data and corresponding text information;
[0012] The medical image data and the text information are input into the medical review model to obtain the review result; the medical review model includes a purpose parsing module for parsing text information and a review module for comparing text information.
[0013] The fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the aforementioned intelligent medical information review method based on a large model.
[0014] In this application, the medical review model reviews medical image data and corresponding text information, thereby making up for the deficiency of deep convolutional models that only output data and cannot review it. Attached Figure Description
[0015] Figure 1 This is a flowchart of a medical information intelligent review method based on a large model according to an embodiment of this application;
[0016] Figure 2 This is an architecture diagram showing the incentive adjustment of the intelligent medical information review method based on a large model according to the embodiments of this application;
[0017] Figure 3 This is an architecture diagram of a medical information intelligent review device based on a large model according to an embodiment of this application;
[0018] Figure 4 This is an architectural diagram of an electronic device according to an embodiment of this application. Detailed Implementation
[0019] To make the above-mentioned objects, features, and advantages of this application more apparent and understandable, specific embodiments of this application will be described in detail below with reference to the accompanying drawings. Although exemplary embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of this application and to fully convey the scope of this application to those skilled in the art.
[0020] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application shall have the ordinary meaning as understood by one of ordinary skill in the art to which this application pertains.
[0021] This application provides the above-described intelligent medical information review method based on a large model. The specific scheme of this method is as follows: Figure 1 Figure 2 As shown, this method can be executed by a large-scale model-based intelligent medical information review device, which can be integrated into electronic devices such as computers, servers, computer clusters, and data centers. Combined with... Figure 1 As shown, the intelligent medical information review method based on a large model includes:
[0022] S101, acquire medical image data and corresponding text information;
[0023] In this application, medical image data includes imaging data such as X-ray films, CT scans, and MRI; text information includes diagnostic reports, medical records, and annotation information.
[0024] Preferably, the medical images are standardized (e.g., resolution adjustment, grayscale normalization), and the text information is cleaned (e.g., noise removal, standardization of terminology).
[0025] Preferably, data alignment is performed: ensuring that each medical image matches its corresponding text information, and associating them using timestamps or unique identifiers (such as patient ID, examination number).
[0026] S102, input the medical image data and the text information into the medical review model to obtain the review result; the medical review model includes a purpose parsing module for parsing text information and a review module for comparing text information.
[0027] The comprehensive medical review model comprises two core modules:
[0028] Objective parsing module: Parses text information and extracts key content (such as lesion description and diagnostic conclusion); Review module: Compares the parsed text information with medical image data, checks for consistency, and generates review results.
[0029] The output format of the audit results may include: audit conclusion (such as "consistent" or "inconsistent"), error message (such as "text description does not match the image: lesion location is incorrect"), and modification suggestions (such as "it is recommended to modify the lesion size to 3cm×2cm").
[0030] In this application, the medical review model reviews medical image data and corresponding text information, thereby making up for the deficiency of deep convolutional models that only output data and cannot review it.
[0031] In one implementation, inputting the medical image data and the text information into a large-scale medical review model to obtain the review result includes:
[0032] The input text information is parsed to obtain the target description;
[0033] The same descriptive processing was performed on the medical image data to obtain the analysis results;
[0034] The analysis results and textual information are compared to obtain the review results.
[0035] In this application, the goal of the comparison is to check whether the analysis results (from medical images) are consistent with the text information (from diagnostic reports). The review results include consistency assessment, error messages, and modification suggestions.
[0036] In this application, key fields (such as anatomical location, lesion type, and size) in the analysis results and text information are compared one by one.
[0037] Example:
[0038] Text description: {"Anatomical location":"Right lower lobe","Lesion type":"Nodule","Size":"3cm×2cm"}.
[0039] Image analysis: {"Anatomical location":"Right lower lobe","Lesion type":"Nodule","Size":"3.5cm × 100cm"}
[0040] 2.5cm"}.
[0041] Comparison results: Sizes are inconsistent (error range can be set to ±0.5cm).
[0042] In this application, for special fields (such as anatomical locations), fuzzy matching algorithms (such as Levenshtein distance and Jaccard similarity) are used to handle synonyms or differences in expression.
[0043] Preferably, a medical rule base (such as the law of conservation of energy and unit consistency) is introduced for logical verification.
[0044] Example: Check if the units of measurement for lesion size are consistent (e.g., "cm" vs "mm"), and check if the location of the lesion conforms to anatomical common sense.
[0045] In one implementation, parsing the input text information to obtain the target description includes:
[0046] Obtain historical descriptive data for medical images;
[0047] Construct the target topology based on historical descriptive data;
[0048] The input text information is parsed to obtain parsed data;
[0049] The parsed data is mapped to the topological structure to obtain the target description.
[0050] In this application, historical descriptive data includes textual descriptions of similar medical images in the past (such as diagnostic reports and annotation information).
[0051] In this application, we ensure that the historical description data covers a variety of cases (such as normal, diseased, and post-operative) and a variety of expressions (such as technical terms and colloquial language).
[0052] In this application, the target topology is a logical framework used to represent the key content of textual information and its relationships. Specifically, it can be a tree structure, a graph structure, or a rule base.
[0053] Example of a tree structure: The root node is "anatomical location", and the child nodes are "lesion type", "size", "shape", etc. In this way, the hierarchy is clear and easy to parse and map.
[0054] Preferably, a medical terminology database and anatomical knowledge are incorporated to enhance the professionalism of the topological structure.
[0055] In this application, the goal of text parsing is to extract key information (such as anatomical location, lesion type, size, etc.) from the input text information. The parsed data is a structured representation of the original text information, used for subsequent mapping to a topological structure.
[0056] In this application, Natural Language Processing (NLP) technologies (such as BERT and BioBERT) are used for entity recognition (NER) and relation extraction. The key extracted information includes: anatomical location (such as "left upper lobe of the lung"), lesion type (such as "nodule" or "effusion"), and lesion characteristics (such as size, shape, and density).
[0057] Preferably, common patterns in historical descriptive data are incorporated to improve parsing accuracy.
[0058] Example: If "lower right lung" is often described as "lower right lobe" in historical data, then map to that term first.
[0059] In this application, the goal of mapping is to embed parsed data into the topology to generate a standardized purpose description.
[0060] In this application, parsed data is mapped to corresponding nodes or relationships according to the rules in the topology. For descriptions that are not completely consistent (such as "lower right lung" vs "lower right lung lobe"), fuzzy matching algorithms (such as Levenshtein distance and Jaccard similarity) are used to find the closest node.
[0061] Preferably, if the parsed data cannot fully match the topology, the topology is dynamically expanded.
[0062] In this application, by introducing historical descriptive data of medical images and constructing a target topology, the input text information can be mapped into a standardized framework, thereby obtaining a more accurate and consistent target description.
[0063] In one implementation, the step of performing the same descriptive processing on the medical image data to obtain the analysis results includes:
[0064] Medical image data is preprocessed to obtain preprocessed images;
[0065] Based on the objective topology, the objective description is decomposed to obtain the decomposed objective;
[0066] Based on each decomposition target, the preprocessed image is processed to obtain the decomposition analysis;
[0067] The analysis results are obtained by summarizing the above decomposition analysis.
[0068] In this application, the objective description is usually a complex whole, and the task of decomposing the objective is to break it down into multiple specific sub-tasks.
[0069] In this application, the objective description is broken down into multiple subtasks based on the nodes and relationships in the topology.
[0070] Preferably, if the purpose description contains vague or ambiguous content, it should be supplemented based on historical data or a rule base.
[0071] In this application, for each decomposition target, a corresponding decomposition analysis is generated using the appropriate image processing or analysis method.
[0072] In this application, the analytical method is as follows:
[0073] Localize anatomical locations: Use anatomical knowledge or pre-trained models (such as U-Net, nnU-Net) for region segmentation.
[0074] Detecting lesion types: Using deep learning classification models (such as ResNet and EfficientNet) to identify lesion types.
[0075] Measuring lesion features: lesion size is calculated using bounding box detection or segmentation mask.
[0076] In this application, all decomposition analyses are combined in a logical order to form a complete result.
[0077] Preferably, during the aggregation process, consistency between the decomposition analyses is checked, and the analysis results are standardized using a domain terminology library (such as SNOMED CT).
[0078] In one implementation, it further includes:
[0079] The aforementioned objective description is elevated to a higher level, resulting in multiple similar objectives;
[0080] Similar processing was performed on medical image data to obtain supplementary results;
[0081] Based on the supplementary results, a supplementary description of the audit results is provided.
[0082] In this application, the purpose of hypergeneracy is to abstract specific purpose descriptions into broader categories, thereby discovering potential related information.
[0083] In this application, a similar purpose of higher-level analysis is used to perform additional analysis on medical image data to generate supplementary results that can reveal information not covered by the original purpose description.
[0084] In this application, multimodal models (such as CLIP, MedViLM) are used to jointly analyze images and for similar purposes.
[0085] In this application, the supplementary results are integrated into the review results to generate a more comprehensive and detailed description. The supplementary description may include newly discovered lesion areas or other relevant information.
[0086] Preferably, the supplementary results are added to the review results in a logical order, and the supplementary results are sorted according to the severity of the lesion or its clinical significance.
[0087] In one implementation, the description of the objective is generalized to obtain multiple similar objectives, including:
[0088] Based on the target topology, determine the higher-level node describing the target;
[0089] Based on the parent node, determine the parallel nodes in the target topology;
[0090] Based on parallel nodes, identify similar purposes described in the purpose description.
[0091] In this application, the destination topology is a hierarchical logical framework, where the higher-level node is a category at a higher level than the current destination description and is usually used for abstract description.
[0092] In this application, a parallel node refers to other nodes at the same level as the current node. These nodes can help expand the scope of the purpose description and discover more potential similar purposes.
[0093] In this application, other nodes at the same level as the node (located under the same parent node) are searched in the topology.
[0094] In this application, a similar purpose is to combine the diverse descriptions generated by the superior node and the parallel node, which can reveal other potential information related to the original purpose description.
[0095] In this application, multiple similar destinations are generated by utilizing the hierarchical relationships (superior and inferior nodes) and parallel relationships (sibling nodes) in the destination topology.
[0096] In this application, higher-level nodes and parallel nodes are combined to generate similar purposes, and more specific descriptions are generated by combining lesion types and anatomical locations.
[0097] Example:
[0098] Superior node: "Right lung lesion", Parallel nodes: "Left lung lesion", "Both lung lesions".
[0099] Similar purposes: "Left lung lesions", "Bilateral lung lesions", "Nodular lesions in the right lung".
[0100] Preferably, similar purposes are sorted according to clinical significance or probability of occurrence, and a threshold is set to control the number of similar purposes to avoid excessive redundancy in the results.
[0101] In one implementation, the target topology may contain multiple parent nodes.
[0102] In this application, a single purpose description has multiple parent nodes, which need to be extracted separately.
[0103] In one implementation, after acquiring medical image data and before inputting it into a large model, the medical image data is blurred to improve the local randomness of the large model processing.
[0104] In one implementation, the fuzzification adjustment process includes:
[0105] Medical images are divided into blocks to obtain independent blocks;
[0106] For each independent block, obtain the first and second neighboring blocks with different spacings;
[0107] A first feature block is generated based on the independent block and the first neighboring block;
[0108] A second feature block is generated based on the independent block and the second neighboring block.
[0109] The first and second feature blocks are compressed to obtain a compressed block.
[0110] Iterate through all the individual blocks and generate a blurred and adjusted medical image based on the resulting compressed blocks.
[0111] In this application, medical images are divided into blocks, that is, the medical images are divided into corresponding image blocks using a checkerboard pattern; wherein, the image block can be at the pixel level (that is, each pixel is an image block) or other levels, and the specific division depends on the actual processing situation.
[0112] In this application, a sliding window or a fixed step size is used to divide the image into blocks of the same size.
[0113] Preferably, in this application, each image block has 1,001,000 pixels, thereby enabling more feature calculations between local regions while ensuring generation accuracy and reducing computational load.
[0114] In this application, an image block is selected as an independent block. The image blocks above, below, to the left, and to the right of this independent block are the first neighboring blocks; the image blocks one grid away from the top, bottom, left, and right of this independent block are the second neighboring blocks. The spacing between the first and second neighboring blocks and the independent block is different.
[0115] In this application, neighborhood information is extracted for each independent block to capture local structure.
[0116] In this application, generating the first feature block is to generate a local feature representation using an independent block and its first neighboring block. Specifically, this can be done by processing the independent block and the first neighboring block with convolutional layers and attention layers to obtain the first feature block.
[0117] In this application, the specific structure and parameters of the convolutional layer and attention layer can be obtained from the training data or determined according to the actual situation.
[0118] It should be noted that in this application, there are four first neighboring blocks and multiple first feature blocks.
[0119] In this application, the independent block and the first neighboring block are processed by convolutional layers and attention layers to obtain the first feature block. The specific process is as follows: the independent block and four neighboring blocks are concatenated together to form a multi-channel input, and the convolutional layer is used to extract features from the concatenated block; an important feature is enhanced by using a self-attention mechanism or a channel attention mechanism, the attention weight is calculated, and the output of the convolutional layer is weighted to enhance the important feature; the output of the attention layer is split into multiple feature blocks, and each feature block corresponds to the processing result of the independent block and at least one neighboring block.
[0120] In this application, a second feature block is generated to generate a broader local feature representation using the independent block and its second neighboring block. The specific generation process is the same as that of the first feature block, except that the parameters of the convolutional layer and the attention layer are different.
[0121] In this application, the generated feature blocks are compressed into a more compact representation to reduce computational cost while retaining key information. Feature compression is performed using pooling operations (such as max pooling or average pooling) or fully connected layers.
[0122] In this way, multiple first and second feature blocks are compressed into a single compressed block, which corresponds to the size and position of the individual blocks and is used to replace them. All image blocks are replaced by the compressed block, resulting in a blurred and adjusted medical image.
[0123] In this application, each image block of the medical image is traversed to obtain the corresponding compressed block.
[0124] In this application, for image blocks / independent blocks near the edge, their first and second neighboring blocks are incomplete. In this case, the incomplete blocks are completed by copying the first and second neighboring blocks in their relative positions. For example, if the first neighboring block above an independent block does not exist, the first neighboring block below it is copied and used as the block above it.
[0125] In this application, by completing the image blocks, the processing accuracy of adjacent image blocks is greatly improved.
[0126] Preferably, combined with Figure 2 As shown, after the fuzzification adjustment, the stimulus adjustment is also included.
[0127] The incentive adjustment process includes:
[0128] The input image is adjusted using a residual module with two convolutions;
[0129] The spatial information of the adjusted input image is obtained by adopting the concept of multi-scale / multi-branch: after a 3×3 convolution, it is divided into four branches, and each branch performs convolution with different kernel sizes (3, 5, 7, 9);
[0130] All four branches obtain channel weights through the SE module, resulting in different weighted feature maps;
[0131] The weighted feature maps of each branch are concatenated to obtain the concatenated feature map;
[0132] The concatenated feature map is added to the adjusted input image to obtain the activation feature map;
[0133] The activation feature map is combined with the input image to obtain the activation-adjusted input feature map.
[0134] This allows for the stimulation of channel and spatial information in the blurred image, improving the accuracy of subsequent recognition and diagnosis.
[0135] This application provides a medical information intelligent review device based on a large model, used to execute the medical information intelligent review method based on a large model described above. The following is a detailed description of the medical information intelligent review device based on a large model.
[0136] like Figure 3 As shown, the intelligent medical information verification device based on a large model includes:
[0137] Data acquisition module 101 is used to acquire medical image data and corresponding text information;
[0138] The medical review module 102 is used to input the medical image data and the text information into the medical review big model to obtain the review result; the medical review big model includes a purpose parsing module for parsing text information and a review module for comparing text information.
[0139] In one implementation, the medical review module 102 is further configured to:
[0140] The input text information is parsed to obtain the purpose description; the same purpose description processing is applied to medical image data to obtain the analysis results; the analysis results and text information are compared to obtain the review results.
[0141] In one implementation, the medical review module 102 is further configured to:
[0142] The stated purpose description is generalized to obtain multiple similar purposes; similar purpose processing is performed on medical image data to obtain supplementary results; based on the supplementary results, the review results are further described.
[0143] In one implementation, the medical review module 102 is further configured to:
[0144] Acquire historical descriptive data of medical images; construct the target topology based on the historical descriptive data; parse the input text information to obtain parsed data; map the parsed data to the topology to obtain the target description.
[0145] In one implementation, the medical review module 102 is further configured to:
[0146] Medical image data is preprocessed to obtain a preprocessed image; based on the target topology, the target description is decomposed to obtain the decomposed target; based on each decomposed target, the preprocessed image is processed to obtain the decomposition analysis; the decomposition analysis is summarized to obtain the analysis result.
[0147] In one implementation, the medical review module 102 is further configured to:
[0148] Based on the target topology, determine the parent node of the target description; based on the parent node, determine the parallel nodes in the target topology; based on the parallel nodes, determine similar targets to the target description.
[0149] In one implementation, the target topology may contain multiple parent nodes.
[0150] The medical information intelligent review device based on a large model provided in the above embodiments of this application corresponds to the medical information intelligent review method based on a large model provided in the embodiments of this application. Therefore, the specific content in this system corresponds to the medical information intelligent review method based on a large model. The specific content can be referred to the records in the medical information intelligent review method based on a large model, which will not be repeated in this application.
[0151] The medical information intelligent review device based on a large model provided in the above embodiments of this application and the medical information intelligent review method based on a large model provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the applications stored therein.
[0152] The above describes the internal functions and structure of the intelligent medical information review device based on a large model, such as... Figure 4 As shown, in practice, this medical information intelligent review device based on a large model can be implemented as an electronic device, including: a memory 301 and a processor 303.
[0153] Memory 301 can be configured to store a program.
[0154] Additionally, memory 301 can also be configured to store various other data to support operation on the electronic device. Examples of this data include instructions for any application or method used to operate on the electronic device, contact data, phonebook data, messages, pictures, videos, etc.
[0155] Memory 301 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. Processor 303, coupled to memory 301, is used to execute programs in memory 301 for:
[0156] Acquire medical image data and corresponding text information;
[0157] The medical image data and the text information are input into the medical review model to obtain the review result; the medical review model includes a purpose parsing module for parsing text information and a review module for comparing text information.
[0158] In one implementation, the processor 303 is further configured to:
[0159] The input text information is parsed to obtain the purpose description; the same purpose description processing is applied to medical image data to obtain the analysis results; the analysis results and text information are compared to obtain the review results.
[0160] In one implementation, the processor 303 is further configured to:
[0161] The stated purpose description is generalized to obtain multiple similar purposes; similar purpose processing is performed on medical image data to obtain supplementary results; based on the supplementary results, the review results are further described.
[0162] In one implementation, the processor 303 is further configured to:
[0163] Acquire historical descriptive data of medical images; construct the target topology based on the historical descriptive data; parse the input text information to obtain parsed data; map the parsed data to the topology to obtain the target description.
[0164] In one implementation, the processor 303 is further configured to:
[0165] Medical image data is preprocessed to obtain a preprocessed image; based on the target topology, the target description is decomposed to obtain the decomposed target; based on each decomposed target, the preprocessed image is processed to obtain a decomposition analysis; the decomposition analysis is summarized to obtain the analysis result.
[0166] In one implementation, the processor 303 is further configured to:
[0167] Based on the target topology, determine the parent node of the target description; based on the parent node, determine the parallel nodes in the target topology; based on the parallel nodes, determine similar targets to the target description.
[0168] In one implementation, the target topology may contain multiple parent nodes.
[0169] In this application, Figure 4 The diagram only shows some components and does not mean that the electronic device includes only these components. Figure 4 The components shown.
[0170] The electronic device provided in this embodiment is based on the same inventive concept as the intelligent medical information review method based on a large model provided in this application embodiment, and has the same beneficial effects as the methods adopted, run or implemented by the application stored therein.
[0171] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0172] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0173] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0174] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory. Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.
[0175] This application also provides a computer-readable storage medium corresponding to the intelligent medical information review method based on a large model provided in the foregoing embodiments, wherein a computer program (i.e., a program product) is stored thereon. When the computer program is run by a processor, it executes the interactive image analysis assistance method for 3D aerial imaging provided in any of the foregoing embodiments.
[0176] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.
[0177] The computer-readable storage medium provided in the above embodiments of this application and the interactive image analysis assistance method for 3D aerial imaging provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the applications stored therein.
[0178] It should be noted that numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. In some instances, well-known structures and techniques have not been shown in detail so as not to obscure the understanding of this specification.
[0179] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0180] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.
Claims
1. A method for intelligent review of medical information based on a large model, characterized in that, include: Acquire medical image data and corresponding text information; The medical image data and text information are input into the medical review model to obtain the review results. The medical review model includes a purpose parsing module for parsing text information and a review module for comparing text information.
2. The method for intelligent review of medical information based on a large model according to claim 1, characterized in that, The process of inputting the medical image data and the text information into the medical review model to obtain the review result includes: The input text information is parsed to obtain the target description; The same descriptive processing was performed on the medical image data to obtain the analysis results; The analysis results and textual information are compared to obtain the review results.
3. The method for intelligent review of medical information based on a large model according to claim 2, characterized in that, Also includes: The aforementioned objective description is elevated to a higher level, resulting in multiple similar objectives; Similar processing was performed on medical image data to obtain supplementary results; Based on the supplementary results, a supplementary description of the audit results is provided.
4. The method for intelligent review of medical information based on a large model according to claim 3, characterized in that, The process of parsing the input text information to obtain the target description includes: Obtain historical descriptive data for medical images; Construct the target topology based on historical descriptive data; The input text information is parsed to obtain parsed data; The parsed data is mapped to the topological structure to obtain the target description.
5. The method for intelligent review of medical information based on a large model according to claim 4, characterized in that, The same descriptive processing is performed on the medical image data to obtain the analysis results, including: Medical image data is preprocessed to obtain preprocessed images; Based on the objective topology, the objective description is decomposed to obtain the decomposed objective; Based on each decomposition target, the preprocessed image is processed to obtain the decomposition analysis; The analysis results are obtained by summarizing the above decomposition analysis.
6. The method for intelligent review of medical information based on a large model according to claim 4, characterized in that, The description of the objective is elevated to a higher level, resulting in several similar objectives, including: Based on the target topology, determine the higher-level node describing the target; Based on the parent node, determine the parallel nodes in the target topology; Based on parallel nodes, identify similar purposes described in the purpose description.
7. The method for intelligent review of medical information based on a large model according to claim 4, characterized in that, In the target topology, the target describes multiple parent nodes.
8. A medical information intelligent verification device based on a large model, characterized in that, include: The data acquisition module is used to acquire medical image data and corresponding text information; The medical review module is used to input the medical image data and the text information into the medical review model to obtain the review result; the medical review model includes a purpose parsing module for parsing text information and a review module for comparing text information.
9. An electronic device, characterized in that, include: Memory and processor; The memory is used to store programs; The processor, coupled to the memory, is used to execute the program for: Acquire medical image data and corresponding text information; The medical image data and the text information are input into the medical review model to obtain the review result; the medical review model includes a purpose parsing module for parsing text information and a review module for comparing text information.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by a processor to implement the intelligent medical information review method based on a large model as described in any one of claims 17.