An information query method and device, a computer device, and a storage medium

Abstract

The application relates to an information query method and device, computer equipment and a storage medium. The method comprises the following steps: a first entity generates a query request package according to case information of a target patient; the first entity determines a plurality of candidate entities from a pre-stored searchable entity set and sends the query request package through a constructed target entity prediction model based on the query request package; the candidate entity generates a response summary package according to a local case library and feeds back to the first entity in response to receiving the query request package; the first entity performs similarity verification on the plurality of response summary packages in sequence in response to receiving the plurality of response summary packages, screens out one or more referenceable cases, and takes the corresponding candidate entity as a second entity; and the first entity establishes a data encryption transmission channel with the second entity to obtain full case information of the referenceable case stored by the second entity. The method can improve the accuracy of case matching and the data transmission efficiency, and at the same time, guarantee the patient privacy and data security.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information query, in particular to an information query method and device, computer equipment and a storage medium. BACKGROUND

[0002] With the development of medical level and the promotion of regional medical cooperation, patients increasingly frequently go to different medical institutions for treatment and referral. In actual clinical work, hospitals often need to rely on historical cases, medical images and other data from other hospitals due to their own experience and insufficient case resources in the hospital.

[0003] However, the existing case information query technology still has many technical and privacy problems in actual application, for example: first, the matching efficiency is low, and the current "broadcast search" method is mainly used to request other hospitals to provide relevant case data in the whole country, which is time-consuming, has low response rate and lacks precise positioning ability; second, the data redundancy is serious, especially the medical images such as CT and MRI have large volume, and direct transmission causes high bandwidth occupation and obvious system delay; third, the privacy leakage risk is high, and some platforms mainly use original data in plaintext transmission, lack necessary encryption and identity verification mechanism, and patient privacy is easy to be exposed. Therefore, it is necessary to propose a new information query method to solve the above problems. SUMMARY

[0004] Therefore, it is necessary to provide an information query method, device, computer equipment and storage medium which can improve the accuracy of case matching and data transmission efficiency, and protect patient privacy and data security.

[0005] In one aspect, an information query method is provided, the method comprising:

[0006] The first entity extracts disease feature data from the case information of the target patient, combines the image feature data, and generates a query request package;

[0007] The first entity determines a plurality of candidate entities from a pre-stored searchable entity set based on the query request package through a constructed target entity prediction model, and sends the query request package;

[0008] The candidate entity, in response to receiving the query request package, generates a response summary package from a local case library and feeds back to the first entity;

[0009] The first entity, in response to receiving a plurality of response summary packages, sequentially performs similarity verification on the plurality of response summary packages, filters out one or more referenceable cases, and takes the corresponding candidate entity as a second entity;

[0010] The first entity generates a key according to an interaction parameter between the query request package and a response summary package fed back by the second entity, establishes a data encryption transmission channel with the second entity, and obtains full case information of the reference case stored by the second entity.

[0011] In one of the embodiments, the first entity extracts disease feature data from case information of a target patient, combines image feature data, and generates a query request package, which includes:

[0012] The case information of the target patient is parsed to obtain the disease feature data, which at least includes one of the following: International Classification of Diseases code, chief complaint, initial diagnosis record, and physiological parameter.

[0013] The disease query summary feature is generated according to the disease feature data.

[0014] The image examination data of the target patient is obtained, and image pyramid feature extraction under a first scale is performed on the image examination data to obtain image feature data under the first scale.

[0015] The image query summary feature is generated according to the image feature data under the first scale, the case summary feature is generated by combining the disease query summary feature, and the query request package is constructed.

[0016] In one of the embodiments, the first entity determines a plurality of candidate entities from a pre-stored searchable entity set by a constructed target entity prediction model based on the query request package, and sends the query request package, which includes:

[0017] The case summary feature in the query request package is extracted, and a structured disease summary vector and an image summary feature vector are respectively parsed.

[0018] The structured disease summary vector and the image summary feature vector are fused to construct a case joint input vector.

[0019] The historical case sharing distribution information of a plurality of searchable entities in the searchable entity set is obtained to construct corresponding entity attribute vectors.

[0020] For a plurality of searchable entities, the case joint input vector and the corresponding entity attribute vector are respectively fused to generate case entity matching items and input to the target entity prediction model, and the correlation probability scores between a plurality of searchable entities and case information of the target patient are respectively calculated.

[0021] Based on the correlation probability scores, a plurality of searchable entities are sorted, and a plurality of searchable entities with the highest scores are selected as the candidate entities.

[0022] In one embodiment, the plurality of searchable entities are respectively combined with the case joint input vector and the corresponding entity attribute vector to generate a case entity matching item and input to the target entity prediction model, and the correlation probability score between the plurality of searchable entities and the case information of the target patient is calculated, including:

[0023] The case summary embedding network, the graph neural network propagation layer, and the output score layer are sequentially arranged to form the target entity prediction model;

[0024] Based on the case joint input vector, a case summary embedding vector is generated through the case summary embedding network, and the case summary embedding network at least includes a multi-layer perception network for encoding structured disease features and an image feature encoding network for encoding image summary features;

[0025] The case summary embedding vector and the corresponding entity attribute vector are fused to obtain the case entity matching item;

[0026] The plurality of searchable entities are taken as graph nodes, the corresponding case entity matching items are taken as the initial features of the graph nodes, and the historical referral relationships and cooperation relationships between the searchable entities are taken as edges to construct an entity association graph;

[0027] Based on the entity association graph, neighbor node feature aggregation is performed on the graph nodes through the graph neural network propagation layer to generate a graph embedding vector of case entity relationships;

[0028] Based on the graph embedding vector of case entity relationships, the correlation probability score between the plurality of searchable entities and the case is calculated and output through the output score layer.

[0029] In one embodiment, in response to receiving the query request package, the candidate entity generates a response summary package according to the local case library and feeds back to the first entity, including:

[0030] The candidate entity analyzes the query request package to obtain the disease query summary feature and the image query summary feature;

[0031] In the local case library, historical case records with a field matching rate greater than or equal to a preset structured matching threshold with the disease query summary feature are retrieved as a candidate reference case set;

[0032] Based on the image query summary feature, a query image vector is constructed;

[0033] Obtaining medical image data corresponding to a plurality of cases in the candidate reference case set, performing image pyramid feature extraction under the first scale to generate a first response image vector;

[0034] Performing first scale image similarity calculation on the query image vector and the first response image vector, and in response to the image feature similarity between the first response image vector and the query image vector being greater than or equal to a preset image matching threshold, regarding the corresponding case as a response case;

[0035] Performing image pyramid feature extraction under the second scale on the entity image data of the response case to obtain a second response image vector and generate a response image summary feature;

[0036] Generating a structure field summary according to the structured data of the response case in the local case library;

[0037] Generating a unique identification code according to the case identifier of the response case and the entity identifier of the candidate entity, combining the response image summary feature, the structure field summary, and the image feature similarity to generate the response summary package.

[0038] In one embodiment, in response to receiving a plurality of response summary packages, the first entity sequentially performs similarity verification on a plurality of response summary packages, filters out one or more referenceable cases, and regards the corresponding candidate entity as a second entity, including:

[0039] The first entity analyzes a plurality of received response summary packages to obtain a plurality of identification codes and corresponding response image summary features, structure field summaries, and image feature similarities;

[0040] Based on the image feature similarity, the plurality of identification codes are sorted to generate a verification sequence;

[0041] Performing image pyramid feature extraction under the second scale on the image examination data of the target patient to generate an image verification vector;

[0042] Based on the verification sequence, a plurality of response image summary features and the image verification vector are sequentially subjected to multi-layer matching analysis to calculate a second scale image verification similarity;

[0043] Aligning the structure field summary with the disease feature data by field, and calculating a structure field matching degree according to a preset field weight rule;

[0044] Based on a preset joint scoring function, the second scale image verification similarity and the structure field matching degree are combined to calculate a comprehensive similarity corresponding to the identification code.

[0045] In response to the comprehensive similarity being greater than or equal to a preset available threshold, it is determined that the corresponding response case is the referenceable case, the corresponding response image summary feature and the structure field summary are recorded as reference case summary information, and the corresponding candidate entity is determined as the second entity.

[0046] In one of the embodiments, the first entity generates a key according to the interaction parameters between the query request package and the response summary package fed back by the second entity, establishes a data encryption transmission channel with the second entity, and obtains the full case information of the referenceable case stored by the second entity, including:

[0047] The first entity extracts the sending timestamp of the query request package and a negotiated random factor, and combines the identification code and the receiving timestamp of the response summary package to construct an interaction parameter set;

[0048] The interaction parameter set is input into a preset elliptic curve key agreement algorithm to generate a one-time communication key;

[0049] A temporary encrypted communication channel is established based on the one-time communication key, and the identity verification and handshaking confirmation operations are completed in the channel initialization process.

[0050] On the other hand, an information query device is provided, and the device comprises:

[0051] A query generation module is configured to extract disease feature data from case information of a target patient by a first entity, and generate a query request package in combination with image feature data;

[0052] A candidate entity determination module is configured to determine a plurality of candidate entities from a pre-stored searchable entity set by a constructed target entity prediction model based on the query request package, and send the query request package;

[0053] A response receiving module is configured to generate a response summary package according to a local case library and feed back to the first entity in response to the candidate entity receiving the query request package;

[0054] A similarity verification module is configured to perform similarity verification on a plurality of response summary packages in sequence to filter out one or more referenceable cases and take the corresponding candidate entity as a second entity in response to the first entity receiving a plurality of response summary packages;

[0055] A key negotiation communication module is configured to generate a key based on interaction parameters between the query request package and the response summary package fed back by the second entity, establish a data encryption transmission channel with the second entity, and obtain full case information of the referenceable case stored by the second entity.

[0056] In still another aspect, a computer device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the following steps when executing the computer program:

[0057] The first entity extracts disease feature data from case information of a target patient, combines image feature data, and generates a query request package;

[0058] The first entity determines a plurality of candidate entities from a pre-stored searchable entity set based on the query request package through a constructed target entity prediction model and sends the query request package;

[0059] The candidate entity, in response to receiving the query request package, generates a response summary package based on a local case library and feeds back to the first entity;

[0060] The first entity, in response to receiving a plurality of the response summary packages, sequentially performs similarity verification on the plurality of the response summary packages, filters out one or more referenceable cases, and takes the corresponding candidate entity as a second entity;

[0061] The first entity generates a key based on interaction parameters between the query request package and the response summary package fed back by the second entity, establishes a data encryption transmission channel with the second entity, and obtains full case information of the referenceable case stored by the second entity.

[0062] In still another aspect, a computer readable storage medium is provided, which stores a computer program, and the computer program, when executed by a processor, implements the following steps:

[0063] The first entity extracts disease feature data from case information of a target patient, combines image feature data, and generates a query request package;

[0064] The first entity determines a plurality of candidate entities from a pre-stored searchable entity set based on the query request package through a constructed target entity prediction model and sends the query request package;

[0065] The candidate entity, in response to receiving the query request package, generates a response summary package based on a local case library and feeds back to the first entity;

[0066] The first entity performs similarity verification on the plurality of response summary packages in sequence, screens one or more referenceable cases, and takes the corresponding candidate entity as a second entity in response to receiving the plurality of response summary packages.

[0067] The first entity generates a key according to an interaction parameter between the query request package and the response summary package fed back by the second entity, establishes a data encryption transmission channel with the second entity, and obtains full case information of the referenceable case stored by the second entity.

[0068] The information query method, device, computer device, and storage medium described above, by constructing a structured query request package based on the disease characteristic data and image characteristic data of the target patient by the first entity, replacing the direct transmission of the original case, not only effectively avoids the clear text diffusion of sensitive medical data between institutions, but also realizes the compressed expression of the query information, thereby reducing the complexity of the matching calculation and the system overhead under the premise of protecting the privacy of the patient, and significantly improving the matching efficiency; at the same time, the first entity screens a plurality of candidate entities for accurate request through the constructed target entity prediction model, avoiding the resource waste and low response rate caused by the "broadcasting type finding" in the prior art, and fundamentally solving the problem of low matching efficiency; further, by adopting the response summary package returned by the candidate entity, the full case information is obtained only after confirming the similarity to establish an encryption channel, avoiding the redundancy burden caused by the frequent transmission of large-volume images or full case information in the early stage, and effectively solving the problem of serious data redundancy; in addition, by generating a one-time communication key according to the interaction parameter to establish a temporary encryption channel, point-to-point encryption is realized, the identity verification and data encryption in the patient information transmission process are ensured, the information security level is significantly improved, and the privacy leakage risk in the prior art is avoided. BRIEF DESCRIPTION OF DRAWINGS

[0069] In order to more clearly illustrate the embodiments of the present application, the drawings required in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0070] Figure 1 It is an application environment diagram of the information query method in an embodiment;

[0071] Figure 2 It is a flowchart of the information query method in an embodiment;

[0072] Figure 3 It is a flowchart of the information query step in an embodiment;

[0073] Figure 4 A structural block diagram of an information query device in an embodiment;

[0074] Figure 5 An internal structural diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0075] For the purpose, technical solutions and advantages of the present application to be clearer, the present application is further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0076] The information query method provided by the present application can be applied to an application environment as shown in Figure 1 The user of the first entity enters the case information of the target patient through the front-end system deployed on the terminal 102, and the terminal 102 sends the case information to the server 104 for processing. The server 104 locally constructs a query request package, and then performs intelligent screening and scoring on a preset searchable entity set based on the constructed target entity prediction model, selects a plurality of candidate entities, and pushes the query request package to the systems accessed by the candidate entities. After the candidate entities receive the query request, they execute response summary generation logic through a local case library, and feed back a response summary package to the server 104 for unified processing and similarity verification, and select a second entity. Then, the server 104 generates a one-time communication key based on the interaction parameters in the query request package and the response summary package, and establishes an encrypted transmission channel with the second entity to obtain complete reference case information for clinical reference. The entity can be a single device or a cluster device, such as a hospital server. The terminal 102 can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers and portable wearable devices, which are used to initiate case query requests and receive feedback results at the entity user end. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers, which is used to support the running of the target entity prediction model, the screening of the candidate entities, the collection of the response summary and the similarity calculation, etc.

[0077] In one embodiment, as shown in Figure 2 An information query method is provided, which is described below by taking the server 104 in Figure 1 as an example, including the following steps:

[0078] Step 201, the first entity extracts disease feature data according to the case information of the target patient, and generates a query request package in combination with image feature data;

[0079] At step 202, the first entity determines a plurality of candidate entities from the pre-stored searchable entity set based on the query request package and through the constructed target entity prediction model, and sends the query request package;

[0080] At step 203, the candidate entity generates a response summary package according to the local case library in response to receiving the query request package, and feeds back to the first entity;

[0081] At step 204, the first entity performs similarity verification on the plurality of response summary packages in sequence in response to receiving the plurality of response summary packages, screens out one or more referenceable cases, and takes the corresponding candidate entity as a second entity;

[0082] At step 205, the first entity generates a key according to the interaction parameters between the query request package and the response summary package fed back by the second entity, establishes a data encrypted transmission channel with the second entity, and obtains the full case information of the referenceable case stored by the second entity.

[0083] In the above information query method, the first entity constructs a structured query request package based on the disease characteristic data and image characteristic data of the target patient, instead of directly transmitting the original case, which not only effectively avoids the clear diffusion of sensitive medical data between institutions, but also realizes the compressed expression of query information, thereby reducing the complexity of matching calculation and system overhead under the premise of protecting patient privacy, and significantly improving the matching efficiency. At the same time, the first entity screens out several candidate entities for accurate request through the constructed target entity prediction model, avoiding the resource waste and low response rate caused by the "broadcasting type search" in the prior art, and fundamentally solving the problem of low matching efficiency. Further, by using the candidate entity to return the response summary package, the full case information is obtained through the encrypted channel only after confirming the similarity, avoiding the redundancy burden caused by the frequent transmission of large-volume images or full case information in the early stage, and effectively solving the problem of serious data redundancy. In addition, by generating a one-time communication key according to the interaction parameters to establish a temporary encrypted channel, point-to-point encryption is realized, the identity verification and data encryption in the patient information transmission process are ensured, the information security level is significantly improved, and the privacy leakage risk in the prior art is avoided.

[0084] In one embodiment, the first entity extracts disease characteristic data from the case information of the target patient, combines image characteristic data, and generates a query request package, including:

[0085] The case information of the target patient is analyzed to obtain disease characteristic data, and the disease characteristic data at least includes one of the following: international disease classification code, chief complaint, initial diagnosis record, and physiological parameter;

[0086] A disease query summary feature is generated according to the disease characteristic data;

[0087] Obtain the image examination data of the target patient, and perform image pyramid feature extraction at a first scale on the image examination data to obtain image feature data at the first scale;

[0088] Generate image query summary features according to the image feature data at the first scale, combine the disease query summary features, generate case summary features, and construct a query request package.

[0089] Specifically, in the embodiment, the case information of the target patient is structurally analyzed to extract disease characteristics including international disease classification codes, chief complaints, initial diagnosis records, and physiological parameters, and combined with the patient image examination data, multi-scale image features are extracted through image pyramid, and on this basis, representative disease summary features and image summary features are generated, thereby constructing a standardized query request package. Not only does it significantly simplify the required data dimension and scale, but also improves the response speed of the subsequent query matching process, and avoids direct transmission of the original case data, effectively reducing the risk of privacy leakage, achieving an optimized balance between privacy protection and matching efficiency.

[0090] In one embodiment, the first entity determines a plurality of candidate entities from the pre-stored searchable entity set based on the query request package through the constructed target entity prediction model and sends the query request package, including:

[0091] Extract the case summary features in the query request package, and respectively analyze to obtain a structured disease summary vector and an image summary feature vector;

[0092] Fuse the structured disease summary vector and the image summary feature vector to construct a case joint input vector;

[0093] Obtain the historical case sharing distribution information of a plurality of searchable entities in the searchable entity set, and construct the corresponding entity attribute vector;

[0094] For a plurality of searchable entities, respectively fuse the case joint input vector and the corresponding entity attribute vector to generate case entity matching items and input them into the target entity prediction model, and respectively calculate the correlation probability scores between the plurality of searchable entities and the case information of the target patient;

[0095] Sort the plurality of searchable entities based on the correlation probability scores, and select a plurality of searchable entities with the highest scores as candidate entities.

[0096] Specifically, in this embodiment, a target entity prediction model is constructed. Based on the case joint input vector extracted from the query request packet and the historical shared attribute information of searchable entities, a structured matching mechanism is established. This enables the accurate screening of highly relevant candidate entities from a preset entity set. Compared with traditional broadcast matching or manual selection mechanisms, this significantly improves the accuracy and automation of target entity positioning, reduces the participation frequency of irrelevant entities, shortens the response cycle, and overcomes the technical bottlenecks of "low response rate and inaccurate positioning" in the prior art. This provides intelligent support for efficient case transfer in multi-center collaborative scenarios.

[0097] In one embodiment, for multiple searchable entities, the case joint input vector and the corresponding entity attribute vector are respectively fused to generate case entity matching items, which are then input into the target entity prediction model. The relevance probability scores between the multiple searchable entities and the case information of the target patient are calculated, including:

[0098] A target entity prediction model is formed by sequentially setting up a case summary embedding network, a graph neural network propagation layer, and an output scoring layer;

[0099] Based on the joint input vector of cases, a case summary embedding vector is generated through a case summary embedding network. The case summary embedding network includes at least a multilayer perceptron network for encoding structured disease features and an image feature encoding network for encoding image summary features.

[0100] The case summary embedding vector and the corresponding entity attribute vector are fused to obtain the case entity matching item;

[0101] Multiple searchable entities are used as graph nodes, the corresponding case entity matching items are set as the initial features of the graph nodes, and the historical referral and collaboration relationships between searchable entities are set as edges to construct an entity association graph.

[0102] Based on the entity association graph, the graph nodes are aggregated with neighbor node features through the graph neural network propagation layer to generate graph embedding vectors of case entity relationships;

[0103] Based on the graph embedding vector of case entity relationships, the correlation probability score between multiple searchable entities and cases is calculated and output through the output scoring layer.

[0104] Specifically, in the embodiment, a multi-modal summary embedding network and a graph neural network propagation mechanism are introduced into the target entity prediction model, the collaborative embedding capability of structured disease information and image summaries is utilized, the referral relationship and cooperation relationship actually existing between entities are combined, an entity graph network model is constructed, and the features between nodes are propagated and aggregated by the graph neural network. Thus, the relevance score between each entity and the target case can be output based on the structural features and graph structure information, the understanding ability of the model for the medical network relationship is enhanced, the accuracy of candidate entity prediction is improved, and better entity recommendation results are ensured under the condition of limited medical resources.

[0105] In one specific embodiment, a case entity matching item is generated and input into the target entity prediction model to calculate the relevance probability score between the case information of the target patient and a plurality of searchable entities, including:

[0106] The input of the target entity prediction model is set to be a case summary embedding vector, denoted as:

[0107] ;

[0108] wherein X represents the case summary embedding vector, represents a structured disease summary vector, represents an image summary feature vector, MLP(·) represents a multi-layer perceptron network function, which is used for nonlinear feature coding of the structured disease summary vector, realizing dimension compression and semantic enhancement, and outputting an embedding vector of structured information, CNN(·) represents an image feature coding network function, which is used for feature extraction and dimension reduction of the image summary feature vector, and only retains key image pattern information related to disease representation, and outputs an image embedding vector. The embedding vector of structured information and the image embedding vector are spliced to obtain the case summary embedding vector X.

[0109] The case summary embedding vector and the corresponding entity attribute vector are fused to obtain a case entity matching item, denoted as: ; wherein represents the case entity matching item of the i-th searchable entity, H(i) represents the entity attribute vector of the i-th searchable entity, i.e., the frequency statistics of related cases provided by the entity in history, disease distribution, etc., and ReLU(·) represents a rectified linear unit activation function, which is used for nonlinear transformation of the fused feature vector, sets all values less than 0 in the feature vector to 0, and retains the part greater than or equal to 0 unchanged, in order to enhance the feature sparsity and model expression ability, and the output result is a non-negative feature vector with the same dimension as the input, which is used as the initial representation of the case entity matching item and is used for constructing the node features in the graph neural network subsequently;

[0110] Based on the entity association graph, the neighbor node feature aggregation is performed on the graph nodes by the graph neural network propagation layer to generate the graph embedding vector of the case entity relationship. The neighbor node feature aggregation process is based on the following formula: ; wherein, is the feature vector of the i-th graph node (i.e., the i-th searchable entity) in the k-th layer, which is initially , is the new feature of the i-th graph node in the k+1-th layer, which is the weighted aggregation result of the neighbor node information; is the feature vector of node j, which is a neighbor (i.e., a possible regional cooperative entity) of node i; N(i) is the neighbor set of node i; W is a weight matrix, which is a learnable linear transformation matrix used to transform the neighbor node features; is a weight coefficient used to identify the influence degree of node j on node i, which can be weighted by the co-occurrence frequency / geographical relationship or learned through an attention mechanism; is an activation function used to perform non-linear mapping on the feature vector aggregated by the neighbor nodes in the graph neural network propagation layer to enhance the model expression ability and fitting ability for complex patterns. It is a non-linear transformation function such as ReLU, LeakyReLU or ELU, and the output result is still a vector with the same dimension as the input, only after non-linear transformation processing, it provides stable and trainable feature input for the subsequent scoring layer;

[0111] The feature vector of the i-th image node under the final K-layer aggregation, i.e., the graph embedding vector, is denoted as ;

[0112] Based on the graph embedding vector of the case entity relationship, the correlation probability score between the multiple searchable entities and the case is calculated and output by the following formula: ;

[0113] wherein, represents the correlation probability score between the i-th searchable entity and the case, i.e., the probability that the i-th searchable entity has a similar case given the case entity matching item; sigmoid(·) represents an activation function that limits the output result to 0-1; represents the weight vector corresponding to the graph embedding vector, which is used for weighted summation of the graph embedding vector ; is a bias term used to adjust the output of the model, which is obtained together with the weight vector by training the model.

[0114] In one specific embodiment, before generating the case entity matching item and inputting it into the target entity prediction model, it further includes:

[0115] Obtain historical case request records and a set of response entities, perform real label annotation, and construct a training set;

[0116] Based on the preset loss function, the target entity prediction model is trained on the training set, and in response to the target entity prediction model converging to the preset confidence threshold on the preset loss function, it is determined that the target entity prediction model training is completed.

[0117] Specifically, in one embodiment, the preset loss function is:

[0118] ;

[0119] Wherein, represents the prediction deviation value, that is, the difference between the overall prediction result of the model and the true result in the target entity prediction of the current case; represents the total number of searchable entities, represents whether the i-th entity is a candidate entity, and the value is 0 or 1;

[0120] Based on the loss function, the target entity prediction model is trained on the training set, and in response to the target entity prediction model converging to the prediction deviation value change less than 0.1 on the preset loss function, it is determined that the target entity prediction model training is completed.

[0121] Specifically, in this embodiment, the target entity prediction model is trained based on the training set labeled by the true label, and the model is optimized using the preset loss function, which can effectively determine whether the model converges and completes the training, ensures that the model reaches a balance in prediction accuracy and convergence speed, and improves the prediction performance of the model, ensures its reliability and effectiveness in actual application, especially in the information query between entities, which can improve the accuracy of case matching.

[0122] In one embodiment, the candidate entity responds to receiving the query request package, and generates a response summary package according to the local case library and feeds back to the first entity, including:

[0123] The candidate entity parses the query request package to obtain the disease query summary feature and the image query summary feature;

[0124] Retrieving the historical case records in the local case library that have a field matching rate with the disease query summary feature greater than or equal to a preset structured matching threshold as a candidate reference case set;

[0125] Based on the image query summary feature, a query image vector is constructed;

[0126] Obtaining the medical image data corresponding to multiple cases in the candidate reference case set, performing image pyramid feature extraction at the first scale to generate a first response image vector;

[0127] perform first scale image similarity calculation on the query image vector and the first response image vector, and in response to the image feature similarity between the first response image vector and the query image vector being greater than or equal to a preset image matching threshold, the corresponding case is taken as a response case;

[0128] perform image pyramid feature extraction at a second scale on entity image data of the response case to obtain a second response image vector and generate a response image summary feature;

[0129] generate a structured field summary according to structured data of the response case in the local case library;

[0130] generate a unique identification code according to the case identifier of the response case and the entity identifier of the candidate entity, and combine the response image summary feature, the structured field summary, and the image feature similarity to generate a response summary package.

[0131] Specifically, in the embodiment, the response summary package is used to replace the original case data as the information carrier for the preliminary interaction, the structured matching field and the image pyramid feature in the local case library are extracted through the candidate entity, and preliminary image similarity determination is performed, and only the response cases with similarity meeting the threshold requirement are taken as the candidate results for feedback, which significantly reduces the data amount and invalid data interference of network transmission, improves the matching efficiency, effectively reduces the bandwidth load and system delay caused by full-amount case transmission, controls the information exposure range, and improves the use safety of patient data.

[0132] In one embodiment, in response to receiving a plurality of response summary packages, the first entity performs similarity verification on the plurality of response summary packages in sequence, filters out one or more referenceable cases, and takes the corresponding candidate entity as a second entity, including:

[0133] The first entity analyzes the received plurality of response summary packages to obtain a plurality of identification codes and corresponding response image summary features, structured field summaries, and image feature similarities;

[0134] Sort the plurality of identification codes based on the image feature similarity to generate a verification sequence;

[0135] Perform image pyramid feature extraction at a second scale on the image examination data of the target patient to generate an image verification vector;

[0136] Based on the verification sequence, perform multi-layer matching analysis on the plurality of response image summary features and the image verification vector in sequence to calculate a second scale image verification similarity;

[0137] Align the structured field summary with the disease feature data, and calculate a structured field matching degree according to a preset field weight rule;

[0138] Based on a preset joint scoring function, the joint second scale image verification similarity and the structure field matching degree are calculated to obtain a comprehensive similarity corresponding to the recognition code;

[0139] In response to the comprehensive similarity being greater than or equal to a preset available threshold, it is determined that the corresponding response case is a referenceable case, the corresponding response image summary features and the structure field summary are recorded as reference case summary information, and the corresponding candidate entity is determined as a second entity.

[0140] Specifically, in the embodiment, after the first entity receives a plurality of response summary packages, based on a bidirectional comparison mechanism of image summary features and structure field summaries, a multi-scale image similarity analysis is performed through an image verification vector, and a final similarity score is output by combining field alignment and weight matching degree calculation of the structure field. Compared with the traditional single image comparison or field matching method, the embodiment can take into account image details and clinical semantic information to ensure high credibility of the selected referenceable cases.

[0141] In one embodiment, as shown in Figure 3 After the corresponding candidate entity is determined as a second entity, it further includes:

[0142] Step 301, the first entity generates a confirmation information package according to the determined second entity and the information summary of the referenceable case, and sends it to the second entity;

[0143] Step 302, in response to the second entity receiving the confirmation information package, a second query request package is generated according to the determined information summary of the referenceable case, and is sent to a second searchable entity in a second searchable entity set associated with the second entity to obtain a secondary response summary package;

[0144] Step 303, in response to the second entity receiving one or more secondary response summary packages, a similarity verification is performed to filter out one or more second referenceable cases and feed back to the first entity.

[0145] It is worth noting that after the preliminary query and matching of the first entity to the second entity are completed, the second entity system will automatically extract the core structure disease features and image summary features in the selected referenceable cases as corresponding information summaries, and regenerate a second query request package. The request package contains referenceable case features alone, or mixed feature expressions after fusion with the original request features of the target patient;

[0146] When the second query request package is generated, the second entity sends the second query request package to its associated second searchable entity set, which includes historical cooperative entities, subordinate entities, associated community entities, special permission entities, etc.

[0147] After receiving the second query request package, the second searchable entity performs the same local digest matching, response digest package generation and encrypted return steps as the first round, feeds back the second-level response digest package to the second entity, which then aggregates and securely forwards it to the first entity at one time.

[0148] Specifically, in the present embodiment, by generating a second query request package based on the internally matched referenceable cases by the second hospital, the risk of mis-matching due to incomplete information, atypical symptoms or entry differences, etc. of the original target case characteristics can be avoided, and the generated query request package has stronger clinical representativeness and structural standardization features, which is conducive to improving the ability of the next-level query hospital to judge the similarity of the case; Furthermore, through the control mechanism that the first hospital allows the second hospital to initiate a second-level query request only after sending a confirmation information package, the fine management of the cross-hospital data exchange authority is realized, unauthorized information diffusion is avoided, and the security compliance of the system is enhanced; At the same time, the introduced second searchable hospital set constitutes a secondary association node in the hospital network, covering historical cooperation hospitals, subordinate institutions, community hospitals and special permission institutions, etc., which is conducive to further expanding the breadth of the matchable data source and improving the response coverage capability for rare diseases, marginal cases or referral problems; In addition, the second hospital uniformly aggregates and feeds back all second-level response digest packages, which can pre-judge, re-screen and digest the results in the second query process, so as to ensure that the first hospital obtains refined feedback with high value and high relevance, and reduces the processing burden of the first hospital. Therefore, the multi-level query mechanism not only enhances the depth ability and intelligent evolution characteristics of the entire information query system, but also improves the comprehensive performance of the existing case information collaboration mechanism from the dimensions of privacy protection, efficiency control and precise matching, etc., has outstanding technical effects and practical value.

[0149] In one embodiment, the first entity generates a key according to the interaction parameters between the query request package and the response digest package feedback by the second entity, establishes a data encryption transmission channel with the second entity to obtain the full case information of the referenceable cases stored by the second entity, including:

[0150] The first entity extracts the sending timestamp of the query request package and the negotiated random factor, and combines the identification code and the receiving timestamp of the response digest package to construct an interaction parameter set;

[0151] The interaction parameter set is input into a preset elliptic curve key agreement algorithm to generate a one-time communication key;

[0152] A temporary encrypted communication channel is established based on the one-time communication key, and the identity verification and handshaking confirmation operations of both parties are completed in the channel initialization process.

[0153] In one specific embodiment, the set of interaction parameters is input into a preset elliptic curve key agreement algorithm to generate a one-time communication key, including:

[0154] The first entity concatenates the query request sending timestamp, the negotiation random factor, the identification code of the response digest package, and the receiving timestamp to form a key derivation seed, and inputs the key derivation seed into a preset key derivation function to obtain a first private key, and calculates a corresponding first public key based on the first private key and an elliptic curve base point, wherein the key derivation function is an HKDF (HMAC-based Key Derivation Function) constructed based on an HMAC, which has the characteristics of high security and supports multi-purpose key expansion, and is suitable for high-security level negotiation scenarios of dynamic session keys;

[0155] The second entity also generates a second private key and a second public key based on the parameters in the set of interaction parameters and the same key derivation function as the first entity;

[0156] The first entity requests the second public key of the second entity, and inputs the first private key generated by itself and the second public key into a preset elliptic curve key agreement algorithm to generate a one-time communication key;

[0157] The second entity requests the first public key of the first entity, and inputs the second private key generated by itself and the first public key into a preset elliptic curve key agreement algorithm to generate a one-time communication key consistent with the first entity;

[0158] In a further embodiment, a temporary encrypted communication channel is established based on the one-time communication key, and the identity verification and handshake confirmation operations of both parties are completed in the channel initialization process, including:

[0159] Both parties input the shared communication key into a preset key derivation function to further derive multiple sub-keys for data encryption, message authentication, and identity authentication, including an encryption key, an authentication key, and a handshake key;

[0160] Based on the derived encryption key, the first entity and the second entity establish a temporary encrypted communication channel using a symmetric encryption algorithm (such as AES-GCM or SM4 algorithm conforming to the national standard) for subsequent encrypted transmission of sensitive data;

[0161] In the channel initialization phase, the first entity and the second entity perform identity verification through a handshake interaction process, which includes:

[0162] Both parties exchange digital certificates or preset identity credentials;

[0163] Both parties use the derived handshake key to sign or perform HMAC operation on the communication digest, thereby completing the legality confirmation of the identity of the communication party;

[0164] If the identity verification is passed, both parties complete the key agreement and authentication process by sending a handshake completion flag data packet (Finished message), and formally enter the encryption communication stage;

[0165] The temporary encrypted communication channel is automatically closed after the session ends or reaches the preset timeout, and the communication key and related intermediate materials used during the session are destroyed, to realize the privacy, integrity and forward security of the communication data.

[0166] Specifically, in the data acquisition stage, the interactive parameters such as the timestamp, random factor and identification code in the query request and response summary are extracted, and a one-time communication key is generated by combining the elliptic curve key agreement algorithm, to establish a secure temporary encrypted channel, complete identity authentication and data transmission, effectively prevent the original case information from being intercepted, tampered or forged in network transmission, solve the privacy leakage problem caused by the lack of encryption mechanism in the prior art, and ensure the security and compliance of medical data in cross-institutional sharing.

[0167] It should be understood that, although Figures 2-3 The steps in the flowchart are shown in sequence according to the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, Figures 2-3 At least part of the steps in the flowchart can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily sequential, but can be alternately executed with other steps or sub-steps or stages of other steps.

[0168] In one embodiment, as shown in Figure 4 An information query device is provided, comprising: a query generation module, a candidate entity determination module, a response receiving module, a similarity verification module and a key agreement communication module, wherein:

[0169] The query generation module is configured to extract disease feature data from the case information of the target patient according to the first entity, and generate a query request packet in combination with the image feature data;

[0170] The candidate entity determination module is configured to determine a plurality of candidate entities from a pre-stored searchable entity set based on the query request packet by constructing a target entity prediction model, and send the query request packet;

[0171] The response receiving module is configured to, in response to the candidate entity receiving the query request packet, generate a response summary packet according to the local case library and feed back to the first entity;

[0172] The similarity verification module is configured to, in response to the first entity receiving the plurality of response summary packets, sequentially perform similarity verification on the plurality of response summary packets, filter out one or more referenceable cases, and take the corresponding candidate entity as the second entity;

[0173] The key negotiation communication module is configured to generate a key according to the interaction parameters between the query request packet and the response summary packet fed back by the second entity, establish a data encryption transmission channel with the second entity, and obtain the full case information of the referenceable case stored by the second entity.

[0174] The specific limitations of the information query device can refer to the limitations of the information query method described above, and will not be repeated here. Each module in the above information query device can be realized by software, hardware and their combination in whole or in part. The above modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in the form of software, so as to call and execute the operations corresponding to each module by the processor.

[0175] In one embodiment, a computer device is provided, which can be a server, and its internal structure diagram can be as shown in Figure 5 The computer device includes a processor, a memory, a network interface and a database connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store information query data. The network interface of the computer device is used to communicate with the terminal outside through the network connection. The computer program is executed by the processor to implement an information query method.

[0176] Those skilled in the art can understand that Figure 5 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0177] In one embodiment, a computer device is provided, which includes a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented:

[0178] The first entity extracts the disease characteristic data according to the case information of the target patient, combines the image characteristic data, and generates a query request package;

[0179] The first entity determines a plurality of candidate entities from the pre-stored searchable entity set based on the query request package through the constructed target entity prediction model, and sends the query request package;

[0180] The candidate entity, in response to receiving the query request package, generates a response summary package according to the local case library and feeds back to the first entity;

[0181] The first entity, in response to receiving a plurality of response summary packages, sequentially performs similarity verification on the plurality of response summary packages, filters out one or more referenceable cases, and takes the corresponding candidate entity as a second entity;

[0182] The first entity generates a key according to the interaction parameters between the query request package and the response summary package fed back by the second entity, establishes a data encryption transmission channel with the second entity, and obtains the full amount of case information of the referenceable case stored by the second entity.

[0183] In one embodiment, a computer readable storage medium is provided, and a computer program is stored on the computer readable storage medium. The computer program is executed by a processor to implement the following steps:

[0184] The first entity extracts the disease characteristic data according to the case information of the target patient, combines the image characteristic data, and generates a query request package;

[0185] The first entity determines a plurality of candidate entities from the pre-stored searchable entity set based on the query request package through the constructed target entity prediction model, and sends the query request package;

[0186] The candidate entity, in response to receiving the query request package, generates a response summary package according to the local case library and feeds back to the first entity;

[0187] The first entity, in response to receiving a plurality of response summary packages, sequentially performs similarity verification on the plurality of response summary packages, filters out one or more referenceable cases, and takes the corresponding candidate entity as a second entity;

[0188] The first entity generates a key according to the interaction parameters between the query request package and the response summary package fed back by the second entity, establishes a data encryption transmission channel with the second entity, and obtains the full amount of case information of the referenceable case stored by the second entity.

[0189] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing relevant hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, the processes of the above-mentioned embodiment methods can be included. Any reference to memory, storage, database or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0190] The technical features of the above embodiments can be combined in any manner. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist, they should be considered as the scope of the present application.

[0191] The above embodiments only express several implementation manners of the present application, and the description is specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, some modifications and improvements can be made, which are all within the scope of the present application. Therefore, the scope of the patent protection of the present application should be subject to the appended claims.

Claims

1. An information search method characterized by comprising: The method comprises the following steps: A first entity extracts disease feature data from the case information of a target patient, combines image feature data, and generates a query request package; The first entity determines a plurality of candidate entities from a pre-stored searchable entity set based on the query request package through a constructed target entity prediction model, and sends the query request package, including: extracting the case summary features in the query request package, and respectively analyzing to obtain a structured disease summary vector and an image summary feature vector, fusing the structured disease summary vector and the image summary feature vector to construct a case joint input vector, obtaining the historical case sharing distribution information of a plurality of searchable entities in the searchable entity set, constructing the corresponding entity attribute vector, for a plurality of searchable entities, respectively fusing the case joint input vector and the corresponding entity attribute vector to generate a case entity matching item and input it into the target entity prediction model, respectively calculating the correlation probability score between the plurality of searchable entities and the case information of the target patient, sorting the plurality of searchable entities based on the correlation probability score, and selecting the top several searchable entities as candidate entities; wherein, the calculation of the correlation probability score comprises: sequentially setting a case summary embedding network, a graph neural network propagation layer and an output score layer to form a target entity prediction model, generating a case summary embedding vector based on the case joint input vector through the case summary embedding network, the case summary embedding network at least including a multi-layer perception network for encoding structured disease features and an image feature encoding network for encoding image summary features, fusing the case summary embedding vector and the corresponding entity attribute vector to obtain a case entity matching item, taking the plurality of searchable entities as graph nodes, setting the corresponding case entity matching item as the initialization feature of the graph nodes, setting the historical referral relationship and cooperation relationship between the searchable entities as edges, constructing an entity association graph, based on the entity association graph, performing neighbor node feature aggregation on the graph nodes through the graph neural network propagation layer to generate a graph embedding vector of the case entity relationship, and based on the graph embedding vector of the case entity relationship, calculating and outputting the correlation probability score between the plurality of searchable entities and the case through the output score layer; The candidate entity generates a response summary package according to the local case library in response to receiving the query request package and feeds back to the first entity; The first entity performs similarity verification on the plurality of response summary packages in sequence in response to receiving the plurality of response summary packages, filters out one or more referenceable cases, and takes the corresponding candidate entity as a second entity; The first entity generates a confirmation information package according to the information summary of the determined second entity and referenceable case, and sends it to the second entity; in response to the second entity receiving the confirmation information package, the second entity generates a second query request package according to the information summary of the determined referenceable case, and sends it to the second searchable entity in the second searchable entity set associated with the second entity to obtain a secondary response summary package; in response to the second entity receiving one or more secondary response summary packages, the second entity performs similarity verification and feeds back one or more second referenceable cases to the first entity. The first entity generates a key according to an interaction parameter between the query request packet and the response summary packet fed back by the second entity, establishes a data encrypted transmission channel with the second entity, and obtains full case information of a reference case stored by the second entity.

2. The information query method of claim 1, wherein, The first entity extracts case characteristic data according to case information of a target patient, combines image characteristic data, and generates a query request packet, including: The case information of the target patient is parsed to obtain the case characteristic data, and the case characteristic data at least includes one of the following: an international disease classification code, a chief complaint, an initial diagnosis record, and a physiological parameter; A case query summary feature is generated according to the case characteristic data; Image examination data of the target patient is obtained, and image pyramid feature extraction under a first scale is performed on the image examination data to obtain image feature data under the first scale; An image query summary feature is generated according to the image feature data under the first scale, and the case summary feature is generated by combining the case query summary feature to construct the query request packet.

3. The information query method of claim 2, wherein, The candidate entity, in response to receiving the query request packet, generates a response summary packet according to a local case library and feeds back to the first entity, including: The candidate entity parses the query request packet to obtain the case query summary feature and the image query summary feature; In the local case library, historical case records with a field matching rate greater than or equal to a preset structured matching threshold with the case query summary feature are retrieved as a candidate reference case set; A query image vector is constructed based on the image query summary feature; Medical image data corresponding to a plurality of cases in the candidate reference case set is obtained, and image pyramid feature extraction under the first scale is performed to generate a first response image vector; First scale image similarity calculation is performed on the query image vector and the first response image vector, and in response to the image feature similarity between the first response image vector and the query image vector being greater than or equal to a preset image matching threshold, the corresponding case is taken as a response case; Second scale image pyramid feature extraction is performed on entity image data of the response case to obtain a second response image vector and generate a response image summary feature; A structured field summary is generated according to structured data of the response case in the local case library; A unique identification code is generated according to a case identifier of the response case and an entity identifier of the candidate entity, and the response summary packet is generated by combining the response image summary feature, the structured field summary, and the image feature similarity.

4. The information query method of claim 3, wherein, The first entity, in response to receiving a plurality of the response summary packets, sequentially performs similarity verification on the plurality of the response summary packets, filters out one or more reference cases, and takes the corresponding candidate entity as a second entity, including: The first entity parses the plurality of the response summary packets received to obtain a plurality of the identification codes and corresponding response image summary features, structured field summaries, and image feature similarities; Based on the image feature similarity, the plurality of the identification codes are sorted to generate a verification sequence; and one or more reference cases are filtered out from the plurality of the response summary packets by sequentially performing similarity verification on the plurality of the response summary packets, and the corresponding candidate entity is taken as a second entity. performing image pyramid feature extraction under the second scale on the image examination data of the target patient to generate an image verification vector; based on the verification sequence, sequentially performing multi-layer matching analysis on a plurality of the response image summary features and the image verification vector to calculate a second scale image verification similarity; aligning the structure field summary with the disease feature data by field, and calculating a structure field matching degree according to a preset field weight rule; based on a preset joint scoring function, jointly calculating the second scale image verification similarity and the structure field matching degree to obtain a comprehensive similarity corresponding to the identification code; in response to the comprehensive similarity being greater than or equal to a preset available threshold, determining that the corresponding response case is the reference case, recording the corresponding response image summary feature and the structure field summary as the reference case summary information, and determining the corresponding candidate entity as the second entity.

5. The information query method of claim 4, wherein, The first entity generates a key according to the interaction parameters between the query request package and the response summary package fed back by the second entity, establishes a data encrypted transmission channel with the second entity, and obtains full case information of the reference case stored by the second entity, including: The first entity extracts the sending timestamp of the query request package and the negotiated random factor, and combines the identification code and the receiving timestamp of the response summary package to construct an interaction parameter set; inputting the interaction parameter set into a preset elliptic curve key agreement algorithm to generate a one-time communication key; based on the one-time communication key, establishing a temporary encrypted communication channel, and completing the identity verification and handshaking confirmation operation in the channel initialization process.

6. An information inquiry apparatus characterized by comprising: The device for implementing the method of any one of claims 1-5 comprises: a query generation module configured to extract disease feature data from case information of a target patient by a first entity, and generate a query request package in combination with image feature data; a candidate entity determination module configured to determine a plurality of candidate entities from a pre-stored searchable entity set by the first entity based on the query request package through a constructed target entity prediction model, and send the query request package; a response receiving module configured to generate a response summary package according to a local case library and feed back to the first entity in response to receiving the query request package by the candidate entity; a similarity verification module configured to sequentially perform similarity verification on a plurality of the response summary packages by the first entity in response to receiving a plurality of the response summary packages, filter out one or more reference cases, and take the corresponding candidate entity as a second entity; a key agreement communication module configured to generate a key according to the interaction parameters between the query request package and the response summary package fed back by the second entity by the first entity, establish a data encrypted transmission channel with the second entity, and obtain full case information of the reference case stored by the second entity.

7. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 5.

Images