Report generation method and device, storage medium and electronic equipment

Abstract

The invention provides a report generation method and device, a storage medium and electronic equipment, and relates to the field of medical equipment. The method comprises the following steps: acquiring first lesion information in a to-be-identified medical image by the electronic equipment; determining similar cases from a plurality of historical cases according to the first focus information; and according to the evaluation suggestions in the similar cases, obtaining the evaluation suggestions of the to-be-identified medical image. Therefore, as existing evaluation suggestions exist in the historical cases, and the evaluation suggestions in the to-be-identified medical image are generated by utilizing the matched evaluation suggestions in the similar cases, compared with manual input by a doctor, the efficiency can be remarkably improved.

Description

Technical Field

[0001] The present application relates to the field of medical equipment, and in particular, to a report generation method, device, storage medium and electronic device. Background Art

[0002] Currently, after carefully reviewing medical imaging data, doctors need to manually write imaging assessment reports based on the information extracted from the imaging data and their own professional experience. This process results in the efficiency of report generation being largely dependent on the professional proficiency and personal ability of medical staff. Therefore, this link is not only time-consuming, but the quality of the report is also directly affected by the professional level of medical staff. Summary of the invention

[0003] In order to overcome at least one of the deficiencies in the prior art, the present application provides a report generation method, device, storage medium and electronic device, specifically including:

[0004] In a first aspect, the present application provides a report generation method, the method comprising:

[0005] Acquiring first lesion information in the medical image to be identified;

[0006] Determining similar cases from a plurality of historical cases according to the first lesion information;

[0007] Based on the evaluation suggestions in the similar cases, an evaluation suggestion for the medical image to be identified is obtained.

[0008] In a second aspect, the present application provides a report generating device, the device comprising:

[0009] A lesion recognition module, used to obtain first lesion information in the medical image to be recognized;

[0010] A similar case module, used to determine similar cases from multiple historical cases according to the first lesion information;

[0011] The evaluation suggestion module is used to obtain the evaluation suggestion of the medical image to be identified based on the evaluation suggestions in the similar cases.

[0012] In a third aspect, the present application further provides a storage medium, wherein the storage medium stores a computer program, and the computer program implements the report generation method when executed by a processor.

[0013] In a fourth aspect, the present application further provides an electronic device, comprising a processor and a memory, wherein the memory stores a computer program, and the computer implements the report generation method under the condition that the computer is executed by the processor.

[0014] Compared with the prior art, this application has the following beneficial effects:

[0015] The present application provides a report generation method, device, storage medium and electronic device. Among them, the electronic device obtains the first lesion information in the medical image to be identified; according to the first lesion information, a similar case is determined from multiple historical cases; according to the evaluation suggestions in the similar cases, an evaluation suggestion for the medical image to be identified is obtained. In this way, since there are ready-made evaluation suggestions in historical cases, the evaluation suggestions in the matched similar cases are used to generate the evaluation suggestions in the medical image to be identified, which can significantly improve efficiency compared with manual entry by doctors. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A flowchart of a report generation method provided in an embodiment of the present application;

[0018] Figure 2 One of the detailed schematic diagrams of the report generation method provided in the embodiment of the present application;

[0019] Figure 3 A second detailed schematic diagram of the report generation method provided in an embodiment of the present application;

[0020] Figure 4 A third detailed schematic diagram of the report generation method provided in an embodiment of the present application;

[0021] Figure 5 A schematic diagram of the structure of a report generating device provided in an embodiment of the present application;

[0022] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0024] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0025] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0026] In the description of the present application, it should be noted that the terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance. In addition, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0027] Based on the above statement, as introduced in the background technology, doctors manually write image assessment reports, and this process results in the efficiency of report generation being largely dependent on the professional proficiency and personal ability of medical staff.

[0028] In the related art, in the report editing process, a vocabulary can be used to improve efficiency. Specifically, the doctor will select some terms and phrases from the high-frequency vocabulary library, and then combine these words to form a report. In practice, it is found that although this method can speed up the generation of reports to a certain extent, due to the different specific circumstances of each case, mechanically applying the vocabulary library cannot fully adapt to the special needs of each case. Therefore, although this method can improve the reporting efficiency to a certain extent, the improvement effect is not particularly significant.

[0029] Based on the discovery of the above technical problems, the inventors have proposed the following technical solutions to solve or improve the above problems through creative work. It should be noted that the defects in the solutions in the above prior art are the results obtained by the inventors after practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed in the embodiments of the present application for the above problems below should all be the contributions made by the inventors to the present application in the process of invention and creation, and should not be understood as technical contents known to those skilled in the art.

[0030] In view of the above technical problems, an embodiment of the present application (hereinafter referred to as the present embodiment) provides a report generation method. Figure 1 As shown, the method includes:

[0031] S1, obtaining first lesion information in a medical image to be identified.

[0032] S2, determining similar cases from multiple historical cases based on the first lesion information.

[0033] S3, based on the evaluation suggestions in similar cases, obtain the evaluation suggestions for the medical images to be identified.

[0034] In this way, since there are ready-made evaluation suggestions in historical cases, the evaluation suggestions in the matched similar cases can be used to generate evaluation suggestions in the medical images to be identified, which can significantly improve efficiency compared to manual entry by doctors.

[0035] For the report generation method provided in this embodiment, the electronic device implementing the method may be, but is not limited to, a processor in an endoscope device, a mobile terminal, a tablet computer, a laptop computer, a desktop computer, a server, etc. The server may be a single server or a server group. The server group may be centralized or distributed (for example, the server may be a distributed system). In some embodiments, the server may be local or remote relative to the user terminal. In some embodiments, the server may be implemented on a cloud platform; as an example only, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud (Community Cloud), a distributed cloud, an inter-cloud (Inter-Cloud), a multi-cloud (Multi-Cloud), etc., or any combination thereof. In some embodiments, the server may be implemented on an electronic device having one or more components.

[0036] To make the solution provided by this embodiment clearer, the following uses a server as an electronic device to implement the method. Figure 1 However, it should be understood that the operations of the flowchart may not be implemented in order, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art may add one or more other operations to the flowchart, or remove one or more operations from the flowchart, guided by the content of this application. Figure 1 As shown, the method includes:

[0037] S1, obtaining first lesion information in a medical image to be identified.

[0038] In this embodiment, the lesion information in the medical image to be identified is referred to as the first lesion information. In addition, the medical image to be identified may be, but is not limited to, an endoscopic image, an X-ray image, a computed tomography (CT) image, a magnetic resonance imaging (MRI) image, an ultrasound image, etc.

[0039] Taking an endoscopic image as an example, the first lesion information refers to the most important or representative lesion feature in the endoscopic image. In order to obtain these features, the present embodiment may use a variety of image analysis and recognition techniques. For example, a machine learning model or a deep learning network may be applied to automatically identify and extract the first lesion information in the endoscopic image by learning a large amount of annotated data. Among them, the first lesion information may include a variety of lesion sub-features, such as lesion type, morphological features, positional features, color features, texture features, edge features, spatial features, etc. These lesion sub-features are explained below.

[0040] Lesion type refers to the basic category of lesions in endoscopic images, such as polyps, ulcers, etc. The server can analyze the images through pre-trained lesion recognition models to identify and classify different lesion types. In addition, the server can further subdivide polyps into low-risk and high-risk polyps, or classify ulcers into superficial or deep ulcers through other pre-trained image recognition models.

[0041] Morphological features refer to the shape, size and boundary of the lesion. By analyzing the morphological information of the lesion area, the extracted morphological features include area, perimeter, roundness and boundary complexity. Among them, the area represents the number of pixels in the lesion area, indicating the size of the lesion. The perimeter represents the pixel length of the lesion boundary. The roundness represents the degree to which the lesion shape is close to a circle. The boundary complexity represents the irregularity of the edge, which can be calculated by the fractal dimension or the boundary irregularity formula. These features can help doctors assess the malignant risk and progression of the lesion.

[0042] Positional features refer to the specific location of the lesion in the image. The server can calculate the image coordinates of the center point of the lesion to determine the exact location of the lesion in the image. It should be understood that positional features are very important for comparative analysis of multiple lesions, because lesions in different locations may have different risk levels and progression trends.

[0043] Color features reflect the color distribution of the lesion in the image. The color features extracted by the server include average color value, color distribution, abnormal color ratio and other features. Among them, the average color value is used to characterize the mean of the RGB channel or HSV channel in the lesion area. Color distribution is used to characterize the uniformity and concentration of color distribution in the lesion area. Abnormal color ratio is used to characterize the proportion of abnormal color (such as red, white) pixels in the lesion area. These color features can help doctors determine the nature and progression of the lesion.

[0044] Texture features reflect the delicacy of tissue structure. The server can use a variety of methods to extract texture features. For example, the server can use the Gray-Level Co-occurrence Matrix (GLCM) method to reflect the depth difference, regularity, smoothness and complexity of the texture through indicators such as contrast, energy and entropy. In addition, the server can also analyze the local texture pattern of the image through the Local Binary Patterns (LBP) method to generate a histogram to describe the overall texture characteristics. And the server decomposes the lesion area into different frequency components through the wavelet transform method to capture delicate texture features. It should be understood that these texture features help doctors assess the malignancy risk and progression of lesions.

[0045] Edge features mainly involve the clarity of the lesion boundary. The server can use Canny edge detection or Sobel operator to evaluate the sharpness of the lesion boundary. In addition, the gradient change near the boundary can be calculated to quantify the smoothness of the edge transition. Blurred boundaries may indicate that the lesion has an invasive growth tendency.

[0046] Spatial features describe the relationship between the lesion and other tissues. The server can analyze the distance, color difference and other relationships between the lesion and the adjacent healthy tissues, and provide doctors with more information about the location and environment of the lesion. For example, the center location of the lesion can be represented by image coordinates, while the association with adjacent tissues can be quantitatively analyzed by distance and color difference. These spatial features help doctors better understand the environmental background of the lesion and formulate a reasonable treatment plan.

[0047] Based on the description of the first lesion information in step S1 in the above embodiment, the following is a description of the first lesion information in step S1. Figure 1 Step S2 in the description is as follows:

[0048] S2, determining similar cases from multiple historical cases based on the first lesion information.

[0049] It should be understood that in this embodiment, the lesion information recorded in the historical cases is the second lesion information. In practice, it is found that a large number of historical cases have been accumulated in the hospital database. If the second lesion information in each historical case is matched in turn, it will take a lot of time. Figure 1 Based on the flow chart, Figure 2 As shown, this embodiment also provides the following optional implementation of step S2:

[0050] S2-1, according to the first lesion information, determine multiple candidate cases from multiple historical cases through a first matching algorithm.

[0051] In this embodiment, in order to improve the matching efficiency, it is necessary to quickly screen out candidate cases whose recorded lesion information is close to the first lesion information from a large number of historical cases through a first matching algorithm. In contrast, a matching algorithm based on an index tree can be adopted as the first matching algorithm. Therefore, as an optional implementation method, the server can construct an index tree of multiple historical cases based on the second lesion information in each historical case; use the first lesion information to index the index tree, and index multiple candidate cases from multiple historical cases.

[0052] It should be understood that, compared with the first lesion information, the second lesion information in each historical case includes the same type of lesion sub-features as the first lesion information, which can be, but not limited to, lesion type, morphological features, positional features, color features, texture features, edge features, spatial features, etc. The server will first quantify the second lesion information in these historical cases and convert it into the form of an index tree. In this process, the system classifies and sorts each historical case according to the difference in the second lesion information, thereby constructing a hierarchical and orderly arranged data structure. Therefore, the lesion information in the historical cases can be efficiently stored and retrieved using the index tree.

[0053] Then, the server extracts the first lesion information from the medical image to be identified. This information is similar to the information in the historical cases extracted previously, and is the key information used to describe the lesion. Finally, the server uses the extracted first lesion information to query the constructed index tree. Due to the efficiency of the index tree, the system can quickly find historical cases similar to the first lesion information. In this embodiment, these historical cases matched by the index tree are called candidate cases.

[0054] The above-mentioned index tree can be implemented using a variety of mature technical methods. Exemplarily, since the first lesion information and the second lesion information in the present embodiment are both multidimensional features, the index tree that the first matching algorithm relies on can be implemented using a KD tree (K-dimensional Tree). Among them, the KD tree is an index structure suitable for multidimensional data. The KD tree can efficiently store and retrieve high-dimensional data, and is very suitable for storing and searching multidimensional lesion sub-features in historical cases. In addition, the index tree can also be implemented using a Ball tree, which is also an efficient index structure, particularly suitable for approximate nearest neighbor searches of high-dimensional data. Both technologies can construct an index tree by recursively dividing the space, thereby achieving efficient data retrieval.

[0055] In addition, the Locality Sensitive Hashing (LSH) algorithm can be used to speed up the search for similar cases. The LSH algorithm maps high-dimensional data into a low-dimensional space by randomly projecting and hashing the data, thereby achieving approximate nearest neighbor search at a lower computational complexity. Therefore, by combining the above methods, multiple candidate cases with similar information to the first lesion can be found in a shorter time.

[0056] Based on the multiple candidate cases obtained in the above embodiments, continue to refer to Figure 2 , let's continue Figure 2 Step S2-2 in the following is described:

[0057] S2-2, according to the first lesion information, determine similar cases from multiple candidate cases through a second matching algorithm.

[0058] In order to improve the matching accuracy, it is necessary to accurately determine similar cases that meet the similarity conditions with the first lesion information from the candidate cases of the screening vehicle through the second matching algorithm. In contrast, a matching algorithm that can measure the similarity between lesion information can be used as the second matching algorithm. Therefore, as an optional implementation, the server obtains the second lesion information in each candidate case; according to the similarity between the second lesion information in each candidate case and the first lesion information, similar cases that meet the similarity conditions are determined from multiple candidate cases.

[0059] It can be understood that the server obtains the second lesion information in each candidate case. These features are usually extracted from historical cases through previous processing steps, and contain various key information of the lesion, such as the size, shape, color, etc. These features will be stored in a structured data format to facilitate subsequent comparison and analysis.

[0060] Then, the server will quantify the second lesion information in these candidate cases and the first lesion information in the medical image to be identified, and then perform similarity calculation. Among them, the similarity calculation method relied on by the second matching algorithm can adopt some mature algorithms, such as cosine similarity, Euclidean distance, etc. These algorithms can quantify the similarity between two features. For example, cosine similarity can measure the angle between two vectors, thereby reflecting their similarity; Euclidean distance directly calculates the straight-line distance between two feature vectors.

[0061] Finally, the server calculates the similarity between the second lesion information and the first lesion information in each candidate case, thereby determining similar cases that meet specific similarity conditions. These similarity conditions may be pre-defined thresholds, and only when the similarity exceeds the threshold will the corresponding candidate case be considered a similar case.

[0062] In summary, the first matching algorithm has a high matching efficiency and can efficiently screen out candidate cases; then, the second matching algorithm with higher matching accuracy is used to more accurately identify similar cases, thereby ensuring the reliability and accuracy of similar cases.

[0063] Based on the above implementation, for similar cases in step S2, see Figure 1 , let's continue Figure 1 Step S3 in the following is described:

[0064] S3, based on the evaluation suggestions in similar cases, obtain the evaluation suggestions for the medical images to be identified.

[0065] As an optional implementation, the number of similar cases can be multiple. Figure 2 On the basis of Figure 3 As shown, step S3 may include:

[0066] S3-1A, display evaluation suggestions for multiple similar cases in the display interface.

[0067] S3-2A, receiving and responding to a user's selection operation of evaluation suggestions for multiple similar cases, and using the evaluation suggestion selected by the user as the evaluation suggestion for the medical image to be identified.

[0068] It can be understood that doctors can access the server through the processor in the endoscope device or the desktop computer. The server will provide a display interface in the display of the endoscope, and list the evaluation suggestions for multiple similar cases in the display interface, so that users can intuitively see the content of each suggestion. The user can select one of the evaluation suggestions as the final diagnosis suggestion based on his or her own experience and judgment. For example, the evaluation suggestions for multiple similar cases can be selected through the arrow keys on the keyboard or the mouse. The server will respond to the user's selection operation and use the evaluation suggestion selected by the user as the final evaluation suggestion for the medical image to be identified. In this way, not only a variety of reference opinions are provided to doctors, but also they are given the final decision-making power to ensure that the diagnosis suggestions are both reliable and in line with the actual situation.

[0069] As another optional implementation, the number of similar cases is multiple. Figure 2 On the basis of Figure 4 As shown, step S3 may include:

[0070] S3-1B, fill the evaluation suggestions and the first lesion information in multiple similar cases into the reserved position of the prompt word template to obtain the target prompt word.

[0071] S3-2B, input the target prompt word into the pre-trained large language model for processing to obtain evaluation suggestions for the medical image to be identified.

[0072] It can be understood that the server fills the evaluation suggestions from multiple similar cases and the first lesion information of the medical image to be identified into the reserved position of the prompt word template to generate the target prompt word. Specifically, the system will extract evaluation suggestions from multiple similar cases, and fill these evaluation suggestions and the lesion information (such as location, size, shape, etc.) in the medical image to be identified into the preset prompt word template. This template contains some specific reserved positions for inserting the above information. The generated target prompt word contains all relevant lesion information and evaluation suggestions to form a complete description.

[0073] The server then inputs the target prompt word into a pre-trained large language model for processing. It should be understood that the large language model is a neural network system that can understand and generate natural language. It generates evaluation suggestions for the medical images to be identified by analyzing the information in the target prompt word. Therefore, the server can use the large language model to generate more comprehensive and accurate evaluation suggestions to help doctors make more scientific and reasonable diagnostic decisions. In this way, not only the accuracy of the evaluation suggestions is improved, but also the existing medical data resources are fully utilized to achieve efficient intelligent auxiliary diagnosis.

[0074] Exemplarily, the prompt word template may include a medical image description and a similar case evaluation suggestion section, and the structure of each section is as follows:

[0075] Medical imaging description:

[0076] Lesion location: [lesion location]

[0077] Lesion size: [Lesion size]

[0078] Lesion shape: [Lesion shape]

[0079] Other features: [Other features]

[0080] Similar case evaluation suggestions:

[0081] Similar case 1 evaluation suggestions: [Similar case 1 evaluation suggestions]

[0082] Similar case 2 evaluation suggestions: [Similar case 2 evaluation suggestions]

[0083] Similar case 3 evaluation suggestions: [Similar case 3 evaluation suggestions] ...

[0085] Comprehensive evaluation recommendations:

[0086] Comprehensive evaluation suggestions: [Comprehensive evaluation suggestions generated based on similar case evaluation suggestions]

[0087] The symbol "[]" in the above prompt word template indicates a reserved filling position, and the content in "[]" indicates the content to be filled. In this way, the server fills the lesion information of the medical image and the similar case evaluation suggestions into the above reserved position of the prompt word template to obtain the target prompt word, so that the large language model generates the evaluation suggestions of the medical image to be identified according to the indication of the target prompt word.

[0088] In this embodiment, the evaluation suggestions for the medical images to be identified may include the risk level, progression trend, and decision suggestions of the medical images to be identified. For example, if the risk level in similar cases is similar to that of typical early-stage cancer, it will prompt the doctor that the lesion has a high risk; similarly, further examination or intervention is recommended based on the progression trend in similar cases. As for the decision suggestions, the following decision suggestions will be generated based on the decision suggestions in similar cases:

[0089] (1) If the risk of lesions is high, the doctor may recommend imaging tests such as CT or MRI, or a tissue biopsy.

[0090] (2) Based on the severity and progression of the lesions, specific treatment options may be recommended, such as endoscopic resection or follow-up observation.

[0091] (3) For low-risk but progressive lesions, a higher follow-up frequency may be recommended to monitor disease progression in a timely manner.

[0092] It is not difficult to see from the above embodiments that the accuracy of the evaluation suggestion of the medical image to be identified is heavily dependent on the recognition accuracy of the first lesion information. After detecting and classifying the lesions in the medical image to be identified, the model will give a confidence score to the identified lesions. This confidence score is the maximum value among the many probabilities output by the model. The confidence score is a probability value between 0 and 1, and the larger the number, the more confident the model is in the lesion identification result. It can be understood that the confidence score is like a confidence index, which is used to inform the doctor how much trust the doctor has in a lesion identification result. In this way, the doctor can judge whether to fully trust the recognition result of the model based on this score.

[0093] The first lesion information is not limited to the type of lesion, but also includes lesion sub-features such as morphological features, position features, color features, texture features, edge features, spatial features, etc. These lesion sub-features may be obtained by recognizing the medical image to be recognized through multiple recognition models.

[0094] Therefore, as an optional implementation, the server may also obtain an evaluation report including evaluation recommendations; and obtain a comprehensive confidence based on the confidence levels of multiple lesion sub-features, wherein the comprehensive confidence level is used to prompt the user of the reference value of the evaluation report.

[0095] It can be understood that, in some examples, the comprehensive confidence can be obtained according to the confidence of multiple lesion sub-features and the weight of each lesion sub-feature. For example, assuming that there are 4 recognition models, the confidence of the 4 recognition models is a value between 0 and 1, which is 0.7, 0.6, 0.98, and 0.72 respectively. These confidences can be weighted by preset weights to obtain comprehensive execution. If the respective weights are 0.25, the comprehensive confidence is 0.75. The higher the comprehensive confidence, the higher the credibility of the evaluation report, and vice versa, the lower the credibility of the evaluation report. In addition, it should also be understood that the evaluation report may include other information in addition to the evaluation suggestion, for example, it may also include the first lesion information.

[0096] Based on the same inventive concept as the report generation method provided in this embodiment, this embodiment also provides a report generation device, which includes at least one software function module that can be stored in a memory or fixed in an electronic device in the form of software. The processor in the electronic device is used to execute the executable module stored in the memory. For example, the software function module and computer program included in the device. Please refer to Figure 5 , functionally speaking, the device may include:

[0097] The lesion recognition module 11 is used to obtain first lesion information in the medical image to be recognized;

[0098] A similar case module 12 is used to determine similar cases from multiple historical cases according to the first lesion information;

[0099] The evaluation suggestion module 13 is used to obtain evaluation suggestions for the medical image to be identified based on evaluation suggestions in similar cases.

[0100] In this embodiment, the lesion identification module 11 is used to implement Figure 1 In step S1, the similar case module 12 is used to implement Figure 1 In step S2, the evaluation and suggestion module 13 is used to implement Figure 1 Therefore, for the detailed description of the above modules, please refer to the specific implementation of the corresponding steps.

[0101] In addition, since the invention concept is the same as that of the report generating method, the report generating device can also implement other steps or sub-steps of the method through the above modules.

[0102] Optionally, there are multiple similar cases, and the evaluation and suggestion module 13 is further specifically used for:

[0103] Displaying evaluation suggestions for multiple similar cases in the display interface;

[0104] Receive and respond to a user's selection operation of evaluation suggestions for multiple similar cases, and use the evaluation suggestion selected by the user as the evaluation suggestion for the medical image to be identified.

[0105] Optionally, there are multiple similar cases, and the evaluation and suggestion module 13 is further specifically used for:

[0106] Fill the evaluation suggestions and the first lesion information in multiple similar cases into the reserved position of the prompt word template to obtain the target prompt word;

[0107] The target prompt words are input into a pre-trained large language model for processing to obtain evaluation suggestions for the medical images to be identified.

[0108] Optionally, the similar case module 12 is further specifically used for:

[0109] According to the first lesion information, a plurality of candidate cases are determined from a plurality of historical cases by using a first matching algorithm;

[0110] Based on the first lesion information, similar cases are determined from multiple candidate cases through a second matching algorithm, wherein the matching efficiency of the first matching algorithm is higher than the matching efficiency of the second matching algorithm, and the matching accuracy of the second matching algorithm is higher than the matching accuracy of the first matching algorithm.

[0111] Optionally, the similar case module 12 is further specifically used for:

[0112] Constructing an index tree of multiple historical cases according to the second lesion information in each historical case;

[0113] The index tree is indexed using the first lesion information, and multiple candidate cases are indexed from multiple historical cases.

[0114] Optionally, the similar case module 12 is further specifically used for:

[0115] Obtaining the second lesion information in each candidate case;

[0116] According to the similarity between the second lesion information and the first lesion information in each candidate case, similar cases satisfying similarity conditions are determined from multiple candidate cases.

[0117] Optionally, the first lesion information includes lesion sub-features identified by multiple recognition models from the medical image to be identified, and the similar case module 12 is further used for:

[0118] Obtain an evaluation report including evaluation recommendations;

[0119] According to the confidence of multiple lesion sub-features, a comprehensive confidence is obtained, wherein the comprehensive confidence is used to prompt the user of the reference value of the evaluation report.

[0120] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0121] It should also be understood that if the above implementation is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the various embodiments of the present application.

[0122] Therefore, this embodiment further provides a storage medium, which is a computer-readable storage medium. The storage medium stores a computer program, and when the computer program is executed by a processor, the report generation method provided in this embodiment is implemented. The storage medium can be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0123] This embodiment provides an electronic device for implementing a report generation method. Figure 6 As shown, the electronic device may include a processor 22 and a memory 21. In addition, the memory 21 stores a computer program, and the processor implements the report generation method provided in this embodiment by reading and executing the computer program corresponding to the above implementation manner in the memory 21.

[0124] Continue to see Figure 6 The electronic device further includes a communication unit 23. The memory 21, the processor 22 and the communication unit 23 are electrically connected to each other directly or indirectly through a system bus 24 to achieve data transmission or interaction.

[0125] The memory 21 may be an information recording device based on any electronic, magnetic, optical or other physical principle, used to record execution instructions, data, etc. In some embodiments, the memory 21 may be, but is not limited to, a volatile memory, a non-volatile memory, a storage drive, etc.

[0126] In some embodiments, the volatile memory may be a random access memory (RAM); in some embodiments, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), a flash memory, etc.; in some embodiments, the storage drive may be a disk drive, a solid-state drive, any type of storage disk (such as a CD, a DVD, etc.), or a similar storage medium, or a combination thereof, etc.

[0127] The communication unit 23 is used to send and receive data through a network. In some embodiments, the network may include a wired network, a wireless network, a fiber optic network, a telecommunication network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a wide area network (WAN), a public switched telephone network (PSTN), a Bluetooth network, a ZigBee network, or a near field communication (NFC) network, or any combination thereof. In some embodiments, the network may include one or more network access points. For example, the network may include a wired or wireless network access point, such as a base station and / or a network switching node, through which one or more components of the service request processing system may be connected to the network to exchange data and / or information.

[0128] The processor 22 may be an integrated circuit chip having a signal processing capability, and the processor may include one or more processing cores (e.g., a single-core processor or a multi-core processor). By way of example only, the processor may include a central processing unit (CPU), an application specific integrated circuit (ASIC), an application specific instruction set processor (ASIP), a graphics processing unit (GPU), a physical processing unit (PPU), a digital signal processor (DSP), a field programmable gate array (FPGA), a programmable logic device (PLD), a controller, a microcontroller unit, a reduced instruction set computer (RISC), or a microprocessor, or any combination thereof.

[0129] Understandably, Figure 6The structure shown is for illustration only. The electronic device may also have Figure 6 More or fewer components than shown, or with Figure 6 Different configurations are shown. Figure 6 The components shown may be implemented in hardware, software or a combination thereof.

[0130] It should be understood that the apparatus and method disclosed in the above-mentioned embodiments can also be implemented in other ways. The apparatus embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the apparatus, methods and computer program products according to the multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of boxes in the block diagram and / or the flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0131] The above are only various implementations of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A report generation method, characterized in that: The method comprises: Acquiring first lesion information in the medical image to be identified; Determining similar cases from a plurality of historical cases according to the first lesion information; Based on the evaluation suggestions in the similar cases, an evaluation suggestion for the medical image to be identified is obtained.

2. The report generation method according to claim 1, characterized in that: The number of the similar cases is multiple, and obtaining the evaluation suggestions of the medical image to be identified based on the evaluation suggestions in the similar cases includes: Displaying evaluation suggestions for a plurality of similar cases in a display interface; Receiving and responding to a user's selection operation of the evaluation suggestions for the plurality of similar cases, and using the evaluation suggestion selected by the user as the evaluation suggestion for the medical image to be identified.

3. The report generation method according to claim 1, characterized in that: The number of the similar cases is multiple, and obtaining the evaluation suggestions of the medical image to be identified according to the evaluation suggestions in the similar cases includes: Filling the evaluation suggestions in the plurality of similar cases and the first lesion information into the reserved position of the prompt word template to obtain a target prompt word; The target prompt word is input into a pre-trained large language model for processing to obtain an evaluation suggestion for the medical image to be identified.

4. The report generation method according to claim 1, characterized in that: Determining similar cases from a plurality of historical cases according to the first lesion information includes: Determine a plurality of candidate cases from the plurality of historical cases by using a first matching algorithm according to the first lesion information; Based on the first lesion information, the similar case is determined from the multiple candidate cases by a second matching algorithm, wherein the matching efficiency of the first matching algorithm is higher than the matching efficiency of the second matching algorithm, and the matching accuracy of the second matching algorithm is higher than the matching accuracy of the first matching algorithm.

5. The report generation method according to claim 4, characterized in that: Determining a plurality of candidate cases from the plurality of historical cases by a first matching algorithm according to the first lesion information includes: Constructing an index tree of the plurality of historical cases according to the second lesion information in each of the historical cases; The index tree is indexed using the first lesion information, and the multiple candidate cases are indexed from the multiple historical cases.

6. The report generation method according to claim 4, characterized in that: Determining the similar case from the multiple candidate cases by a second matching algorithm according to the first lesion information includes: Acquiring the second lesion information in each of the candidate cases; According to the similarity between the second lesion information in each candidate case and the first lesion information, similar cases that meet similarity conditions are determined from the multiple candidate cases.

7. The report generation method according to any one of claims 1 to 6, characterized in that: The first lesion information includes lesion sub-features identified by multiple recognition models from the medical image to be identified, and the method further includes: obtaining an evaluation report including the evaluation recommendations; A comprehensive confidence is obtained according to the confidences of the plurality of lesion sub-features, wherein the comprehensive confidence is used to prompt the user of the reference value of the evaluation report.

8. A report generating device, characterized in that: The device comprises: A lesion recognition module, used to obtain first lesion information in the medical image to be recognized; A similar case module, used to determine similar cases from multiple historical cases according to the first lesion information; The evaluation suggestion module is used to obtain the evaluation suggestion of the medical image to be identified based on the evaluation suggestions in the similar cases.

9. A storage medium, characterized in that: The storage medium stores a computer program, and the computer program, when executed by a processor, implements the report generation method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: The electronic device includes a processor and a memory, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the report generation method according to any one of claims 1 to 7.

Images