A data processing method of an image structured report template

Abstract

The application discloses a kind of data processing methods of image structured report template, comprising the following steps: step S1, in the medical image structured report of historical case big data, the organization entity characterized in structured table form is extracted, and the organization entity with higher than general threshold generalization rate is selected as structured main entity;Step S2, in the historical medical image of target object, all organization entities extracted from pathological characterization time series data are eliminated from the structured main entity, and the organization entity with higher than fluctuation threshold fluctuation rate is selected as structured secondary entity in the remaining organization entity.This application forms the image structured report template applied to the target object only, to realize the decentralization of structured main entity, so that the image structured report template is compatible with expert experience and doctor personal experience to improve the expression accuracy of pathological characteristics of image structured report.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image reporting, and in particular to a data processing method of an image structured report template. BACKGROUND

[0002] Compared with the traditional report, the structured report is structured and standardized in content. The traditional report is usually written according to the personal habits of the doctor, and the content is complex and the words are various, so it is difficult to effectively extract the valuable information in the report, and it is impossible to manage and utilize the information. Not only is a large amount of valuable medical history information wasted, but also the omission and errors in report writing are easily caused. When the era of big data and artificial intelligence comes, structured medical history information is the most basic data. Therefore, the popularization and application of structured report are imminent.

[0003] The prior art CN202110892281.6 discloses a prostate MR cancer structured report design method, which includes the following steps: logging into a user interface and providing a plurality of data options on the user interface; filling in the PI_RADS score of the prostate in the structured report template and inserting PACS images, and then uploading to a database; the database receives the uploaded data, and classifies and archives the key information; the database saves the received information; inputs query data request in the user interface and sends it to the database; the database receives the request, matches and filters the data information in the request with the existing information in the database; the database returns all the data to be queried by the user and automatically generates a structured report form.

[0004] Although the above-mentioned prior art can generate a structured report, it is unified in the organization of pathological indicators, so that the generated structured report cannot be personalized and is poor in subjectivity. SUMMARY

[0005] The purpose of the present application is to provide a data processing method of an image structured report template to solve the technical problem that the prior art is unified in the organization of pathological indicators, so that the generated structured report cannot be personalized and is poor in subjectivity.

[0006] To solve the above technical problems, the present application specifically provides the following technical solutions:

[0007] A data processing method of an image structured report template, comprising the following steps:

[0008] Step S1: Extract the tissue entities represented in the form of structured tables from the medical image structured report of historical case big data, and calculate the generality rate of the tissue entities by counting the frequency of use of the tissue entities in the medical image structured report of historical case big data. Then, select the tissue entities with a generality rate higher than the generality threshold from all the tissue entities as the main structured entities. The tissue entities are the organ tissues that represent the pathological features of historical cases in the medical image structured report.

[0009] Step S2: Extract multiple sets of pathological time series data representing the pathological time series status of tissue entities from the historical medical images of the target object, and perform fluctuation analysis on each set of pathological time series data to obtain the fluctuation rate of each set of pathological time series data. Then, extract the corresponding tissue entities from the pathological time series data, and remove the structured main entity from all the tissue entities extracted from the pathological time series data. Among the remaining tissue entities, select tissue entities with fluctuation rates higher than the fluctuation threshold as structured secondary entities.

[0010] Step S3: Using the sigmoid function, determine the entity weights of the structured main entity and structured sub-entities extracted from the pathological characterization time series data based on the volatility. Then, filter the structured main entity and structured sub-entities based on the entity weights. Finally, apply the filtered structured main entity and structured sub-entities to the image structured report template to represent them in a structured table format to form a unique image structured report template applicable to the target object. This achieves decentralization of the structured main entity, making the image structured report template compatible with expert experience and doctors' personal experience, thereby improving the accuracy of the image structured report in expressing pathological features.

[0011] Step S4: While the target object obtains the latest medical images, execute steps S2 and S3 to update the image structured report template of the target object to obtain the latest image structured report template and generate the latest image structured report for the latest medical images.

[0012] As a preferred embodiment of the present invention, the extraction of organizational entities represented in structured table form from the structured medical image report of historical case big data includes:

[0013] In each structured medical image report, all organizational entities in the structured table are extracted, and the semantic names of the organizational entities are converted into standard semantic names to eliminate dimensional errors in the semantic names of the same organizational entities.

[0014] As a preferred embodiment of the present invention, the step of calculating the generalizability of organizational entities by statistically analyzing their usage frequency in structured medical image reports based on historical case big data includes:

[0015] The total number of structured medical image reports in historical case big data was counted, and the total number of each organizational entity was counted as the usage frequency.

[0016] The generalization rate is defined as the ratio of the total number of each organizational entity to the total number of structured medical image reports. The formula for calculating the generalization rate is as follows:

[0017]

[0018] In the formula, p i The generality of the i-th organizational entity is represented by n. i The number of structures is represented by m, which represents the total number of structured medical image reports. i is a quantitative constant with no substantive meaning.

[0019] As a preferred embodiment of the present invention, organizational entities with a generality rate higher than a generality threshold are selected from all the organizational entities as structured main entities, including:

[0020] A general threshold is set, and the generality rate of the organizational entities is compared with the general threshold, wherein,

[0021] When the generality rate of an organizational entity is greater than or equal to the generality threshold, the corresponding organizational entity is taken as the structured main entity.

[0022] If the generality rate of an organizational entity is less than the generality threshold, the corresponding organizational entity will not be used as a structured main entity.

[0023] Preferably, the method for setting a general threshold includes:

[0024] Step 1: Obtain the maximum and minimum generality rates among the organizational entities, labeled Zmax and Zmin respectively. Set the initial generality threshold as p. k = (Zmax + Zmin) / 2;

[0025] Step 2: Based on the general threshold p k The organizational entities are divided into two groups, and the average generality Zo and Zb of the two groups are calculated respectively, and the generality threshold p is updated. k A new universal threshold p is obtained k+1 = (Zo + Zb) / 2;

[0026] Step 3: Set the new universal threshold p k+1 With the general threshold p k In comparison, among them,

[0027] When p k =p k+1 Then p k+1 This is a general threshold;

[0028] When p k ≠p k+1 Then p k+1 Assign a value to p k Return to step two.

[0029] As a preferred embodiment of the present invention, multiple sets of pathological characterization time-series data representing the temporal status of the tissue entity are extracted from the historical medical images of the target object, including:

[0030] The pathological data of each tissue entity in the historical medical images of the target object are extracted, and the pathological data of each tissue entity are linked according to the time sequence of the historical medical images to obtain the pathological characterization time sequence data, wherein the time sequence attribute in the pathological characterization time sequence data is consistent with the time sequence attribute of the historical medical images.

[0031] As a preferred embodiment of the present invention, the step of sequentially performing fluctuation analysis on each group of pathological characterization time-series data to obtain the fluctuation rate of each group of pathological characterization time-series data includes:

[0032] In each set of pathological characterization time series data, the average growth rate of pathological characterization data at adjacent time series is calculated sequentially. The formula for calculating the average growth rate is:

[0033]

[0034] In the formula, d j,i,i+1 X is represented by the average growth rate of pathological characterization data at the i-th and (i+1)-th time series in the j-th group of pathological characterization time series data. j,i+1 X j,i These are respectively represented as the pathological characterization data at the (i+1)th and ith time points in the j-th group of pathological characterization time series data, t j,i+1 t j,i These are respectively represented as the (i+1)th and ith time series in the j-th group of pathological characterization time series data;

[0035] The volatility of each set of pathological time series data is calculated by summing the absolute values ​​of all average growth rates. The formula for calculating the volatility is as follows:

[0036]

[0037] In the formula, D j L is represented as the volatility of the time series data of the j-th group of pathological characteristics. jThe total number of pathological characterization data in the time series data of the j-th group of pathological characterization is represented by i and j, which are quantitative constants and have no substantial meaning.

[0038] As a preferred embodiment of the present invention, organizational entities with volatility higher than a volatility threshold are selected from the remaining organizational entities as structured sub-entities, including:

[0039] The volatility corresponding to the remaining organizational entities is compared with the volatility threshold, where,

[0040] When the volatility of the organizational entity is higher than the volatility threshold, the corresponding organizational entity is treated as a structured sub-entity.

[0041] If the volatility of the organizational entity is lower than or equal to the volatility threshold, the corresponding organizational entity will not be treated as a structured sub-entity.

[0042] As a preferred embodiment of the present invention, the determination of entity weights for the structured primary and structured secondary entities extracted from the pathological characterization time-series data based on the volatility using the sigmoid function includes:

[0043] The initial entity weight of each of the structured main entities is set to 1. R represents the total number of structured main entities, and the initial entity weight of each structured sub-entity is set to 0.

[0044] The volatility corresponding to the structured sub-entities is classified using the sigmoid function, where...

[0045] When the volatility of the structured sub-entity is higher than sigmoid(0), the entity weight of the structured sub-entity is updated to (initial entity weight + volatility / R).

[0046] If the volatility of the structured sub-entity is less than or equal to sigmoid(0), then the entity weight of the structured sub-entity is maintained at the initial entity weight.

[0047] The volatility corresponding to the structured principal entities is classified using the sigmoid function, where...

[0048] When the volatility of the structured principal entity is higher than sigmoid(0), the entity weight of the structured principal entity is maintained at the initial entity weight.

[0049] When the volatility of the structured principal entity is less than or equal to sigmoid(0), the entity weight of the structured principal entity is updated to (initial entity weight - volatility / R);

[0050] As a preferred embodiment of the present invention, the entity filtering of structured main entities and structured sub-entities based on entity weights includes:

[0051] The entity weights of the structured main entity and structured sub-entities are normalized, and the normalized entity weights of the structured main entity and structured sub-entities are compared with the weight threshold. Structured main entities and structured sub-entities whose normalized entity weights are less than the weight threshold are removed.

[0052] As a preferred embodiment of the present invention, the execution steps S2 and S3 update the image structured report template of the target object to obtain the latest image structured report template, including:

[0053] The latest medical images of the target object are added to the target object's historical medical images. Steps S2 and S3 are then executed to obtain the latest structured main entity and structured sub-entity.

[0054] The latest structured main entity and structured sub-entities are applied to the image structured report template to represent it in a structured table format, thereby updating the image structured report template of the target object to obtain the latest image structured report template.

[0055] Compared with the prior art, the present invention has the following advantages:

[0056] This invention extracts structured primary entities from medical image structured reports based on historical case big data, and extracts structured secondary entities from historical medical images of the target object. The structured primary and secondary entities are then applied to an image structured report template to represent them in a structured tabular form, forming a unique image structured report template applicable to the target object. This achieves decentralization of the structured primary entities, making the image structured report template compatible with expert experience and physician's personal experience, thereby improving the accuracy of the image structured report in expressing pathological features. Attached Figure Description

[0057] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0058] Figure 1 A flowchart illustrating the data processing method for the image structured report template provided in this embodiment of the invention. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] like Figure 1 As shown, image structured report templates typically employ pathological features of organs and tissues that most doctors focus on during construction, quantifying these features into structured tables for easy retrieval. This approach summarizes the personal experience of most doctors, forming expert experience. Based on this expert experience, standardized report templates are created using these frequently used organ and tissue pathological features. The report then only reflects these organ and tissue pathological features to generate the target object's image structured report. However, this approach ignores the individual differences of the target object. There may be pathological features of organs and tissues that require attention but are not commonly used in the report template. These features are not represented in the structured report template, potentially leading to misdiagnosis of the target object's disease condition. Furthermore, a standardized image structured report template suffers from organ and tissue centralization, focusing only on commonly used organs and tissues summarized by expert experience while ignoring others. Therefore, this invention provides a data processing method for image structured report templates that effectively decentralizes expert experience by incorporating doctors' personal experience, adding flexibility to the standardized diagnosis and treatment of the target object.

[0061] A data processing method for an image structured report template includes the following steps:

[0062] Step S1: Extract the tissue entities represented in the form of structured tables from the structured medical image report of historical case big data, and calculate the generality rate of the tissue entities by counting the frequency of use of the tissue entities in the structured medical image report of historical case big data. Then, select the tissue entities with a generality rate higher than the generality threshold from all tissue entities as the main structured entities. The tissue entities are the organs and tissues that represent the pathological features of historical cases in the structured medical image report.

[0063] The organizational entities represented in structured tabular form are extracted from the structured medical image reports of historical case big data, including:

[0064] In each structured medical image report, all organizational entities in the structured table are extracted, and the semantic names of the organizational entities are converted into standard semantic names to eliminate dimensional errors in the semantic names of the same organizational entities.

[0065] The general usage rate of organizational entities is obtained by statistically analyzing the frequency of use of structured medical image reports based on historical case big data, including:

[0066] The total number of structured medical image reports in historical case big data was counted, and the total number of each organizational entity was counted as the usage frequency.

[0067] The generalization rate is the ratio of the total number of each organizational entity to the total number of structured medical image reports. The formula for calculating the generalization rate is as follows:

[0068]

[0069] In the formula, p i The generality of the i-th organizational entity is represented by n. i The number of structures is represented by m, which represents the total number of structured medical image reports. i is a quantitative constant with no substantive meaning.

[0070] Organizational entities with a generality rate higher than the generality threshold are selected from all organizational entities as structured master entities, including:

[0071] A general threshold is set, and the generality rate of organizational entities is compared with the general threshold.

[0072] When the generality rate of an organizational entity is greater than or equal to the generality threshold, the corresponding organizational entity is taken as the structured main entity.

[0073] If the generality rate of an organizational entity is less than the generality threshold, the corresponding organizational entity will not be used as a structured main entity.

[0074] Preferably, the method for setting a general threshold includes:

[0075] Step 1: Obtain the maximum and minimum generality rates among the organizational entities, labeled Zmax and Zmin respectively. Set the initial generality threshold as p. k = (Zmax + Zmin) / 2;

[0076] Step 2: Based on the general threshold p k The organizational entities are divided into two groups, and the average generality Zo and Zb of the two groups are calculated respectively, and the generality threshold p is updated. k A new universal threshold p is obtained k+1 = (Zo + Zb) / 2;

[0077] Step 3: Set the new universal threshold p k+1 With the general threshold p k In comparison, among them,

[0078] When p k =p k+1 Then p k+1 This is a general threshold;

[0079] When p k ≠p k+1 Then p k+1 Assign a value to p k Return to step two.

[0080] This method extracts high-profile tissue entities from structured medical image reports, summarizing expert experience to identify pathological characteristics of tissue entities that require attention for a given condition. This leads to standardized indicators for image diagnosis and treatment. For example, when constructing a structured image report template for pneumonia, the extraction of expert experience—specifically, calculating the commonality of various tissue entities—reveals that exceed the commonality threshold as the left lung, right lung, left upper lobe, right upper lobe, left lower lobe, right lower lobe, left pleural cavity, and right pleural cavity. Therefore, the left lung, right lung, left upper lobe, right upper lobe, left lower lobe, right lower lobe, left pleural cavity, and right pleural cavity are selected as the main structured entities.

[0081] Step S2: Extract multiple sets of pathological time series data representing the pathological time series status of tissue entities from the historical medical images of the target object, and perform fluctuation analysis on each set of pathological time series data to obtain the fluctuation rate of each set of pathological time series data. Then, extract the corresponding tissue entities from the pathological time series data, and remove the structured main entities from all the tissue entities extracted from the pathological time series data. Among the remaining tissue entities, select tissue entities with fluctuation rates higher than the fluctuation threshold as structured secondary entities.

[0082] Multiple sets of pathological time-series data characterizing the temporal status of tissue pathology were extracted from the historical medical images of the target object, including:

[0083] The pathological data of each tissue entity in the historical medical images of the target object are extracted, and the pathological data of each tissue entity are linked according to the time sequence of the historical medical images to obtain the pathological characterization time sequence data. The time sequence attributes in the pathological characterization time sequence data are consistent with the time sequence attributes of the historical medical images.

[0084] Fluctuation analysis was performed on each group of pathological characterization time-series data to obtain the volatility of each group of pathological characterization time-series data, including:

[0085] In each set of pathological characterization time series data, the average growth rate of pathological characterization data at adjacent time series is calculated sequentially. The formula for calculating the average growth rate is:

[0086]

[0087] In the formula, d j,i,i+1 X is represented by the average growth rate of pathological characterization data at the i-th and (i+1)-th time series in the j-th group of pathological characterization time series data. j,i+1 X j,i These are respectively represented as the pathological characterization data at the (i+1)th and ith time points in the j-th group of pathological characterization time series data, t j,i+1 t j,i These are respectively represented as the (i+1)th and ith time series in the j-th group of pathological characterization time series data;

[0088] The volatility of each set of pathological time series data is calculated by summing the absolute values ​​of all average growth rates. The formula for calculating volatility is:

[0089]

[0090] In the formula, D j L is represented as the volatility of the time series data of the j-th group of pathological characteristics. j The total number of pathological characterization data in the time series data of the j-th group of pathological characterization is represented by i and j, which are quantitative constants and have no substantial meaning.

[0091] From the remaining organizational entities, those with volatility higher than the volatility threshold are selected as structured sub-entities, including:

[0092] The volatility corresponding to the remaining organizational entities is compared with the volatility threshold, where,

[0093] When the volatility of an organizational entity is higher than the volatility threshold, the corresponding organizational entity is treated as a structured sub-entity.

[0094] If the volatility of an organizational entity is lower than or equal to the volatility threshold, the corresponding organizational entity will not be treated as a structured sub-entity.

[0095] The attending physician of the target patient makes a diagnosis of the target patient based on their own personal experience using medical images, thereby obtaining the pathological features of the organs and tissues of interest to the attending physician. This is the multiple sets of pathological characterization time-series data in this embodiment, such as the pathological characterization time-series data of the left upper lobe, right upper lobe, left lower lobe, right lower lobe, left pleural cavity, right pleural cavity, trachea, bronchus, and pleura. The corresponding tissue entities are obtained, including: left upper lobe, right upper lobe, left lower lobe, right lower lobe, left pleural cavity, right pleural cavity, trachea, bronchus, and pleura. Among them, the left upper lobe, right upper lobe, left lower lobe, right lower lobe, left pleural cavity, and right pleural cavity are the main structural entities, belonging to the part of expert experience, while the trachea, bronchus, and pleura are the secondary structural entities, belonging to the part of the attending physician's personal experience.

[0096] Step S3: Using the sigmoid function, the structured main entity and structured sub-entities extracted from the pathological characterization time series data are weighted based on volatility. The structured main entity and structured sub-entities are then screened based on their weights. The screened structured main entity and structured sub-entities are then applied to the image structured report template to represent them in a structured table format, forming a unique image structured report template applicable to the target object. This decentralizes the structured main entity, making the image structured report template compatible with expert experience and physician's personal experience, thereby improving the accuracy of the image structured report in expressing pathological features.

[0097] The sigmoid function is used to determine the entity weights of structured principal entities and structured secondary entities extracted from pathological characterization time-series data based on volatility. This includes:

[0098] The initial entity weight of each structured main entity is set to 1. R represents the total number of structured main entities, and the initial entity weight of each structured sub-entity is set to 0.

[0099] The volatility corresponding to the structured sub-entities is classified using the sigmoid function, where...

[0100] When the volatility of the structured sub-entity is higher than sigmoid(0), the entity weight of the structured sub-entity is updated to (initial entity weight + volatility / R).

[0101] If the volatility of the structured sub-entity is less than or equal to sigmoid(0), then the entity weight of the structured sub-entity is maintained at the initial entity weight.

[0102] The volatility corresponding to the structured principal entities is classified using the sigmoid function, where...

[0103] When the volatility of the structured principal entity is higher than sigmoid(0), the entity weight of the structured principal entity is maintained at the initial entity weight.

[0104] When the volatility of the structured principal entity is less than or equal to sigmoid(0), the entity weight of the structured principal entity is updated to (initial entity weight - volatility / R);

[0105] The sigmoid function is commonly used in data classification. It can effectively divide entity weights into two update methods. Among them, sigmoid(0) = 0.5, which is used as the volatility threshold, can also be customized according to the user's actual needs. And (initial entity weight - volatility / R) can also be customized according to the user's actual needs.

[0106] To eliminate the centralization problem of structured main entities that rely on expert experience, this embodiment uses the sigmoid function to determine the direction of the update and decrease of entity weights for structured main entities and structured sub-entities. To achieve decentralization, the structured main entity with low volatility (meaning it has a low impact on the target object's condition changes and is not worth paying attention to) has its entity weight reduced accordingly. Conversely, the structured main entity with high volatility (meaning it has a high impact on the target object's condition changes and is worth paying attention to) has its entity weight maintained at its initial value. Similarly, the structured sub-entities with low volatility (meaning they have a low impact on the target object's condition changes and are not worth paying attention to) have their entity weight maintained at its initial value. Conversely, the structured sub-entities with high volatility (meaning they have a high impact on the target object's condition changes and are worth paying attention to) have their entity weight increased accordingly. This process transfers the entity weights of the structured main entities to the structured sub-entities, achieving the goal of decentralization.

[0107] Entity filtering is performed on structured main entities and structured sub-entities based on entity weights, including:

[0108] The entity weights of the structured main entity and structured sub-entities are normalized, and the normalized entity weights of the structured main entity and structured sub-entities are compared with the weight threshold. Structured main entities and structured sub-entities whose normalized entity weights are less than the weight threshold are removed.

[0109] This embodiment provides an example where the left upper lobe, right upper lobe, left lower lobe, right lower lobe, left pleural cavity, and right pleural cavity are considered primary structural entities, with an initial entity weight of 1 / 6. The trachea, bronchus, and pleura are considered secondary structural entities, with an initial entity weight of 0. Calculations show that the left upper lobe, right upper lobe, right lower lobe, left pleural cavity, bronchus, and pleura are ultimately retained. Therefore, the organ tissues represented in the structured table format of the image structured report template for this target object are the left upper lobe, right upper lobe, right lower lobe, left pleural cavity, bronchus, and pleura, which is compatible with both expert experience and physician's personal experience.

[0110] Step S4: While the target object obtains the latest medical images, execute steps S2 and S3 to update the target object's image structured report template to obtain the latest image structured report template and generate the latest image structured report for the latest medical images.

[0111] Steps S2 and S3 are executed to update the image structured report template of the target object to obtain the latest image structured report template, including:

[0112] The latest medical images of the target object are added to the target object's historical medical images. Steps S2 and S3 are then executed to obtain the latest structured main entity and structured sub-entity.

[0113] The latest structured main entity and structured sub-entities are applied to the image structured report template to represent it in a structured table format, thereby updating the image structured report template of the target object to obtain the latest image structured report template.

[0114] This invention extracts structured primary entities from medical image structured reports based on historical case big data, and extracts structured secondary entities from historical medical images of the target object. The structured primary and secondary entities are then applied to an image structured report template to represent them in a structured tabular form, forming a unique image structured report template applicable to the target object. This achieves decentralization of the structured primary entities, making the image structured report template compatible with expert experience and physician's personal experience, thereby improving the accuracy of the image structured report in expressing pathological features.

[0115] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.

Claims

1. A data processing method for an image structured report template, characterized in that, Includes the following steps: Step S1: Extract the tissue entities represented in the form of structured tables from the medical image structured report of historical case big data, and calculate the generality rate of the tissue entities by counting the frequency of use of the tissue entities in the medical image structured report of historical case big data. Then, select the tissue entities with a generality rate higher than the generality threshold from all the tissue entities as the main structured entities. The tissue entities are the organ tissues that represent the pathological features of historical cases in the medical image structured report. Step S2: Extract multiple sets of pathological time series data representing the pathological time series status of tissue entities from the historical medical images of the target object, and perform fluctuation analysis on each set of pathological time series data to obtain the fluctuation rate of each set of pathological time series data. Then, extract the corresponding tissue entities from the pathological time series data, and remove the structured main entity from all the tissue entities extracted from the pathological time series data. Among the remaining tissue entities, select tissue entities with fluctuation rates higher than the fluctuation threshold as structured secondary entities. Step S3: Using the sigmoid function, determine the entity weights of the structured main entity and structured sub-entities extracted from the pathological characterization time series data based on the volatility. Then, filter the structured main entity and structured sub-entities based on the entity weights. Finally, apply the filtered structured main entity and structured sub-entities to the image structured report template to represent them in a structured table format to form a unique image structured report template applicable to the target object. This achieves decentralization of the structured main entity, making the image structured report template compatible with expert experience and doctors' personal experience, thereby improving the accuracy of the image structured report in expressing pathological features. Step S4: While the target object obtains the latest medical images, execute steps S2 and S3 to update the image structured report template of the target object to obtain the latest image structured report template and generate the latest image structured report for the latest medical images.

2. The data processing method for an image structured report template according to claim 1, characterized in that: The extraction of organizational entities represented in structured tabular form from structured medical image reports based on historical case big data includes: In each structured medical image report, all tissue entities in the structured table are extracted, and the semantic names of the tissue entities are converted into standard semantic names to eliminate dimensional errors in the semantic names of the same tissue entities.

3. The data processing method for an image structured report template according to claim 2, characterized in that: The method of calculating the generalizability of organizational entities by statistically analyzing their usage frequency in structured medical image reports based on historical case big data includes: The total number of structured medical image reports in historical case big data was counted, and the total number of each organizational entity was counted as the usage frequency. The generalization rate is defined as the ratio of the total number of each organizational entity to the total number of structured medical image reports. The formula for calculating the generalization rate is as follows: ； In the formula, Characterized by the generality rate of the i-th organizational entity. Represented by the total number of the i-th organizational entities, The total number of structured medical image reports is represented by 'i', which is a quantitative constant with no substantive meaning.

4. The data processing method for an image structured report template according to claim 3, characterized in that: From all the organizational entities, those with a generality rate higher than the generality threshold are selected as the structured main entities, including: A general threshold is set, and the generality rate of the organizational entities is compared with the general threshold, wherein, When the generality rate of an organizational entity is greater than or equal to the generality threshold, the corresponding organizational entity is taken as the structured main entity. If the generality rate of an organizational entity is less than the generality threshold, the corresponding organizational entity will not be used as a structured main entity. Methods for setting general thresholds include: Step 1: Obtain the maximum and minimum generality rates among the organizational entities, labeled Zmax and Zmin respectively, and set the initial generality threshold as follows: ; Step 2: Based on the general threshold The organizational entities are divided into two groups, and the average generality Zo and Zb of the two groups are calculated respectively, and the generality threshold is updated. Obtain the new general threshold ; Step 3: Set the new universal threshold With general threshold In comparison, among them, when ,but This is a general threshold; when Then Assign to Return to step two.

5. The data processing method for an image structured report template according to claim 4, characterized in that: Multiple sets of pathological time-series data characterizing the temporal status of tissue pathology were extracted from the historical medical images of the target object, including: The pathological data of each tissue entity in the historical medical images of the target object are extracted, and the pathological data of each tissue entity are linked according to the time sequence of the historical medical images to obtain the pathological characterization time sequence data, wherein the time sequence attribute in the pathological characterization time sequence data is consistent with the time sequence attribute of the historical medical images.

6. The data processing method for an image structured report template according to claim 5, characterized in that: The step of sequentially performing fluctuation analysis on each group of pathological characterization time-series data to obtain the volatility of each group of pathological characterization time-series data includes: In each set of pathological characterization time series data, the average growth rate of pathological characterization data at adjacent time series is calculated sequentially. The formula for calculating the average growth rate is: ； In the formula, This is represented by the average growth rate of the pathological characterization data at the i-th and (i+1)-th time series positions in the j-th group of pathological characterization time series data. , These are respectively represented as the pathological characterization data at the (i+1)th and ith time points in the j-th group of pathological characterization time series data. , These are respectively represented as the (i+1)th and ith time series in the j-th group of pathological characterization time series data; The volatility of each set of pathological time series data is calculated by summing the absolute values ​​of all average growth rates. The formula for calculating the volatility is as follows: ； In the formula, The volatility of the time series data of the pathological characteristics of the j-th group is represented. The total number of pathological characterization data in the time series data of the j-th group of pathological characterization is represented by i and j, which are quantitative constants and have no substantial meaning.

7. The data processing method for an image structured report template according to claim 6, characterized in that: From the remaining organizational entities, those with volatility higher than the volatility threshold are selected as structured sub-entities, including: The volatility corresponding to the remaining organizational entities is compared with the volatility threshold, where, When the volatility of the organizational entity is higher than the volatility threshold, the corresponding organizational entity is treated as a structured sub-entity. If the volatility of the organizational entity is lower than or equal to the volatility threshold, the corresponding organizational entity will not be treated as a structured sub-entity.

8. The data processing method for an image structured report template according to claim 7, characterized in that: The determination of entity weights for the structured primary and secondary entities extracted from the pathological characterization time-series data using the sigmoid function based on the volatility includes: The initial entity weight of each of the structured main entities is set to 1. R represents the total number of structured main entities, and the initial entity weight of each structured sub-entity is set to 0. The volatility corresponding to the structured sub-entities is classified using the sigmoid function, where... When the volatility of the structured sub-entity is higher than sigmoid(0), the entity weight of the structured sub-entity is updated to (initial entity weight + volatility / R). If the volatility of the structured sub-entity is less than or equal to sigmoid(0), then the entity weight of the structured sub-entity is maintained at the initial entity weight. The volatility corresponding to the structured principal entities is classified using the sigmoid function, where... When the volatility of the structured principal entity is higher than sigmoid(0), the entity weight of the structured principal entity is maintained at the initial entity weight. When the volatility of the structured principal entity is less than or equal to sigmoid(0), the entity weight of the structured principal entity is updated to (initial entity weight - volatility / R).

9. The data processing method for an image structured report template according to claim 8, characterized in that: The entity filtering based on entity weights for structured main entities and structured sub-entities includes: The entity weights of the structured main entity and structured sub-entities are normalized, and the normalized entity weights of the structured main entity and structured sub-entities are compared with the weight threshold. Structured main entities and structured sub-entities whose normalized entity weights are less than the weight threshold are removed.

10. The data processing method for an image structured report template according to claim 9, characterized in that: The execution steps S2 and S3 update the image structured report template of the target object to obtain the latest image structured report template, including: The latest medical images of the target object are added to the target object's historical medical images. Steps S2 and S3 are then executed to obtain the latest structured main entity and structured sub-entity. The latest structured main entity and structured sub-entities are applied to the image structured report template to represent it in a structured table format, thereby updating the image structured report template of the target object to obtain the latest image structured report template.

Images