Method and system for automatic recognition of echocardiogram image measurement data

Abstract

The application discloses an automatic identification method and system for echocardiogram image measurement data, and belongs to the technical field of medical image processing. The method comprises the following steps: performing segmentation processing on an echocardiogram image, identifying a measurement annotation region, obtaining a measurement type of the measurement annotation region, applying an optical character recognition algorithm to obtain an OCR recognition result, performing semantic analysis on the OCR recognition result and mapping the result to a preset echocardiogram parameter database to obtain standardized measurement data and output the data. The application is suitable for a bedside echocardiogram examination scene and can realize the technical effect that measurement parameters that have been annotated in an image collected by a portable ultrasonic device are quickly and accurately transcribed to a clinical report system.

Description

Technical Field

[0001] This application relates to the field of medical image processing technology, and in particular to an automatic identification method and system for echocardiogram image measurement data. Background Technology

[0002] Echocardiography is a core diagnostic tool for cardiovascular diseases, assessing cardiac structure, function, and hemodynamic status. With the increasing availability of portable ultrasound equipment, bedside echocardiography is being used more and more extensively in intensive care units (ICUs), cardiology wards, and emergency departments.

[0003] The main working principle of portable ultrasound equipment is to acquire ultrasound images from multiple standard sections and perform various measurements (such as left ventricular diameter, ejection fraction, Doppler blood flow velocity, etc.) on the portable device. The measurement results are then displayed on the images with annotations. Doctors can view and annotate each exported image and input each measurement value into the corresponding field in the reporting system.

[0004] However, the above workflow has several technical bottlenecks, such as data format fragmentation, closed reporting system, and substandard efficiency and quality during manual transcription. Therefore, a method capable of recognizing echocardiogram images is needed to solve these problems. Summary of the Invention

[0005] This application provides an automatic identification method and system for echocardiogram image measurement data, which solves the technical defects in the existing echocardiogram image recognition workflow, such as data format fragmentation, closed reporting system, and substandard efficiency and quality during manual transcription. It achieves the technical effect of quickly and accurately transcribing the labeled measurement parameters in images acquired by portable ultrasound equipment into the clinical reporting system.

[0006] In a first aspect, embodiments of this application provide an automatic identification and extraction method for echocardiographic image measurement data, comprising: preprocessing the input echocardiographic image, identifying and segmenting the measurement annotation region based on a text detection algorithm; identifying the measurement type of the current image based on the content features of the measurement annotation region, wherein the measurement type includes at least one of M-mode, two-dimensional, pulsed Doppler, continuous Doppler, and tissue Doppler measurements; applying an optical character recognition algorithm optimized for echocardiographic parameters to the measurement annotation region to identify the parameter name, measurement value, and unit of measurement, thereby obtaining an OCR recognition result; performing semantic analysis on the OCR recognition result to establish the association between the parameter name, measurement value, unit of measurement, and measurement method; mapping the identified parameter names to an echocardiographic parameter terminology system based on a pre-established echocardiographic parameter database, thereby obtaining standardized measurement data; and outputting the standardized measurement data in a structured form, and providing data copying and export functions.

[0007] In conjunction with the first aspect, in one possible implementation, the measurement region is located based on at least one of the following features when the text detection algorithm is used to identify and segment the measurement label region: position feature: detecting a preset measurement value display area in the image; color feature: identifying light-colored text regions on a dark background; layout feature: identifying parameter-value-unit text structures arranged in rows; and the candidate text regions are located through a deep learning text detection model, and the measurement label region is selected by combining the above features.

[0008] In conjunction with the first aspect, in the second possible implementation, the optical character recognition algorithm optimized for echocardiographic parameters includes: image preprocessing optimization targeting the dark background and light-colored text features of ultrasound images; character recognition based on a dedicated character set containing echocardiographic terminology, abbreviations, Greek letters, and special symbols; correction of the preliminary recognition results based on a language model of an echocardiographic parameter corpus; and verification of the rationality of the numerical range and the matching of units of the recognition results.

[0009] In conjunction with the first aspect, in the third possible implementation, the semantic analysis of the OCR recognition results includes: line-level structure parsing, parsing each line of recognition results according to a preset text pattern; derived parameter recognition, automatically calculating derived parameters based on the basic measurement values, the derived parameters including at least the E / A ratio, the E / e' ratio, and the pressure gradient calculated from velocity based on the simplified Bernoulli equation; and measurement method extraction, identifying the specific measurement method from the parameter name suffix or context.

[0010] In conjunction with the first aspect, in the fourth possible implementation, the echocardiographic parameter database includes at least: a parameter name mapping table for associating parameter names with standard names from different manufacturers and in different languages; a unit conversion table for unifying different formats of units of measurement into standard units; and a measurement method comparison table for mapping different expressions of measurement methods to standard methods.

[0011] In conjunction with the first aspect, in the fifth possible implementation, the output of standardized measurement data in a structured form includes: a table displaying the identification results in groups of measurement types; a prompt function that visually marks values ​​that exceed the preset normal range; a function that supports copying values ​​and corresponding parameter names individually or in batches; and a function that supports exporting data in multiple formats.

[0012] In conjunction with the first aspect, in the sixth possible implementation, the method further includes a derived parameter verification step: verifying the consistency between the automatically calculated derived parameters and the derived parameters directly identified from the image; if the verification is inconsistent, a prompt is issued to the user.

[0013] Secondly, embodiments of this application provide an automatic identification and extraction system for echocardiographic image measurement data, including an image input module for receiving echocardiographic images to be processed; an image preprocessing and measurement region localization module for preprocessing the image and locating the measurement annotation region; a measurement type identification module for identifying the measurement type of the current image; a dedicated OCR module for performing echocardiographic-optimized optical character recognition on the measurement annotation region; a semantic parsing module for analyzing the OCR results and constructing the association between parameter names, values, units, and methods; a parameter standardization module for mapping the identified parameters to a standard terminology system; a structured output module for displaying the recognition results and providing interactive copy and export functions; and a user interface for interacting with the user, displaying images and results, and receiving user instructions and corrections.

[0014] In conjunction with the second aspect, in one possible implementation, the system operates as standalone software, interacting with external reporting systems via clipboard, and is not dependent on a specific ultrasound equipment brand or reporting system.

[0015] In conjunction with the second aspect, in the second possible implementation, the system further includes a parameter knowledge base management module, a device configuration module, and a history recognition module.

[0016] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: This application can reduce report completion time and improve clinical work efficiency; through automated recognition and semantic understanding, it eliminates errors in manual transcription and improves the quality of echocardiography reports; through multi-layer verification and confidence level prompts, it provides data quality assurance for users; at the same time, this application is designed for the complex parameter system of echocardiography and can accurately handle various measurement types, derived parameters, and display differences between different manufacturers; finally, as an independent system, this application does not require modification of existing equipment or reporting systems, and data transfer can be achieved through simple copy and paste, resulting in low deployment costs. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating an automatic identification and extraction method for echocardiographic image measurement data provided in this application embodiment; Figure 2 This is a block diagram of an automatic identification and extraction system for echocardiographic image measurement data provided in an embodiment of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0020] Figure 1 This is a flowchart of an automatic identification and extraction method for echocardiographic image measurement data provided in an embodiment of this application, such as... Figure 1 As shown, the method includes steps S1 to S6.

[0021] Step S1: Echocardiogram image preprocessing and measurement annotation region localization: The input echocardiogram image is preprocessed, and the measurement annotation region is identified and segmented based on the text detection algorithm.

[0022] The preprocessing operations include: Step S101: Grayscale conversion and adaptive contrast enhancement.

[0023] Step S102: Region segmentation based on echocardiogram image layout features.

[0024] Step S103: Locate the text region using a deep learning text detection model.

[0025] Measurement area localization is based on at least one of the following features: Location characteristics: The preset measurement value display area in the echocardiogram image is usually located in the upper left corner, upper right corner or bottom information bar of the echocardiogram image.

[0026] Color characteristics: Identify light-colored text areas on a dark background. The text is usually white, yellow, or green, and the background is dark gray or black.

[0027] Layout features: Identify parameter-value-unit text structure arranged in rows.

[0028] Step S2: Echocardiographic measurement type identification: Based on the content features of the measurement annotation area, identify the measurement type of the current image.

[0029] The measurement type identification is based on preset rules, as shown in Table 1: Table 1:

[0030] Step S3: Echocardiography-specific OCR recognition: Apply an optical character recognition algorithm optimized for echocardiography parameters to the measurement and annotation area to recognize the parameter name, measurement value, and unit of measurement, and obtain the OCR recognition result.

[0031] Specifically, the optical character recognition algorithm optimized for echocardiography parameters includes the following optimizations: Step S301: Image preprocessing optimization: adaptive binarization for light text on dark background, noise reduction based on ultrasound image characteristics, and contrast enhancement of text regions.

[0032] Step S302: Character recognition optimization: Establish a dedicated character set for echocardiography (usually including common English abbreviations, Chinese parameter names, Greek letters, special symbols, superscripts and subscripts, etc.) and correct the language model based on echocardiography.

[0033] Step S303: Verification of recognition results: numerical range verification, unit matching verification, context consistency verification, etc.

[0034] Step S4: Semantic analysis of echocardiogram parameters: Perform semantic analysis on the OCR recognition results to establish the relationship between parameter names, measurement values, units of measurement, and measurement methods.

[0035] Semantic analysis is performed on the OCR recognition results, including steps S401 to S403: Step S401: Line-level structure parsing. Each line of recognition results is parsed according to a preset text pattern (usually parameter name-value-unit, or parameter name-method-value-unit). In step S401, when parameter names and values ​​are displayed on separate lines, this case should be prioritized. When multiple values ​​are measured consecutively, the case of recognizing multi-valued parameters should also be considered.

[0036] Step S402: Derived parameter identification. Derived parameters are automatically calculated based on the basic measurement values. The derived parameters include at least the E / A ratio, the E / e' ratio, and the pressure gradient calculated from the velocity based on the simplified Bernoulli equation.

[0037] Step S403: Measurement method extraction, inferring the specific measurement method from the parameter name suffix or context.

[0038] Step S5: Echocardiographic parameter standardization mapping: Based on the pre-established echocardiographic parameter database, the identified parameter names are mapped to the echocardiographic parameter terminology system to obtain standardized measurement data.

[0039] The echocardiographic parameter database includes at least: A parameter name mapping table used to associate parameter names with standard names from different manufacturers and languages.

[0040] A unit conversion table used to unify different formats of units of measurement into standard units.

[0041] Measurement method reference table used to map measurement methods with different formulations to standard methods.

[0042] Taking the parameter name mapping table as an example, the same parameter may be displayed in different forms on different devices and in different language environments, as shown in Table 2.

[0043] Table 2:

[0044] Step S6: Structured Data Output: Output the standardized measurement data in a structured format and provide data copying and export functions.

[0045] The step of outputting the standardized measurement data in a structured format includes: The measurement type group displays the identification results in a table, showing the standard name of the parameter, the identification value, the unit, and the measurement method.

[0046] A visual alert function that marks values ​​that exceed the preset normal range.

[0047] It supports copying values, corresponding parameter names, and units, either individually or in batches.

[0048] It supports data export in multiple formats (such as CSV, text, JSON, etc.) to facilitate table processing, program processing, or text processing.

[0049] The automatic identification method for echocardiographic image measurement data proposed in this application embodiment further includes step S7: derived parameter verification. The derived parameters obtained by automatic calculation are compared with the derived parameters directly identified from the image to obtain the confidence level after verification. If the confidence level is low, the low confidence level result is marked. At the same time, this application embodiment also supports users to manually correct recognition errors / supplement missing parameters through the user interface.

[0050] If the verification fails, a prompt will be sent to the user.

[0051] This application segments echocardiogram images and identifies the measurement annotation regions to determine the measurement type of each region. An optical character recognition (OCR) algorithm is then applied to obtain the OCR recognition results. Semantic analysis of the OCR results is performed and mapped to a pre-defined echocardiogram parameter database to obtain standardized measurement data, which is then output. This application is applicable to bedside echocardiography examinations, enabling the rapid and accurate transcription of labeled measurement parameters from images acquired by portable ultrasound equipment into a clinical reporting system.

[0052] While this application provides the method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-inventive labor. The order of steps listed in this embodiment is merely one possible execution order among many and does not represent the only execution order. In actual device or client product execution, the methods shown in this embodiment or the accompanying drawings can be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment).

[0053] like Figure 2 As shown in the illustration, this application also provides an automatic identification and extraction system for echocardiographic image measurement data. This system includes the following modules: Image input module: Used to receive echocardiogram images to be processed. It supports both single echocardiogram image input and batch echocardiogram image input, and supports import path, import method, import echocardiogram image format, etc.

[0054] Image preprocessing and measurement area localization module: used to preprocess the image and locate the measurement annotation area to enable image format conversion and size normalization.

[0055] Measurement type identification module: used to identify the measurement type of the current image. The identification methods include, but are not limited to, rule-based identification based on keywords, auxiliary judgment based on image features, or measurement type classifier based on a large deep learning model.

[0056] Dedicated OCR module: Utilizes a text recognition engine optimized for ultrasound images to perform optical character recognition on the measurement annotation area, optimized for echocardiography, and performs post-processing and verification on the recognition results.

[0057] Semantic parsing module: used to analyze OCR results and build relationships between parameter names, values, units, and methods.

[0058] Parameter standardization module: This module is used to map the identified parameters to a standard terminology system. It includes multiple echocardiographic parameter knowledge bases, multilingual and multi-vendor parameter name mapping tables, unit conversion tables, and measurement method comparison tables.

[0059] Structured output module: Used to display recognition results and provide interactive copy and export functions. The output of recognition results includes, but is not limited to, table format, text format, and image format.

[0060] User interface: Used to interact with users, display images and results, and receive user commands and corrections.

[0061] Parameter Knowledge Base Management Module: Used to maintain and update the mapping rules for echocardiography parameters.

[0062] Device configuration module: Used to configure the parameter display features of devices from different manufacturers.

[0063] Historical Recognition Module: Used to record and query historical recognition results.

[0064] It should be noted that the automatic identification and extraction system for echocardiographic image measurement data proposed in this application embodiment runs as independent software and can operate without being based on a specific ultrasound equipment brand and reporting system. The system can interact with external reporting systems through the clipboard or other functions that enable data interaction.

[0065] Some modules in the apparatus described in this application can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0066] The apparatus or module described in the above embodiments can be implemented by a computer chip or physical entity, or by a product with a certain function. For ease of description, the above apparatus is described by dividing it into various modules according to their functions. When implementing the embodiments of this application, the functions of each module can be implemented in one or more software and / or hardware. Of course, a module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0067] Example 1 Taking the M-mode echocardiogram image acquired by the Philips CX50 portable ultrasound device as an example, this application illustrates the specific implementation process of the automatic identification and extraction method for echocardiogram image measurement data.

[0068] For example, the input echocardiogram image features include: image size of 800×600 pixels; image format of JPG; measurement area located in the upper left corner of the echocardiogram image; measurement type of M-mode measurement; and display language of English parameter name + Chinese unit.

[0069] Step S1: The system preprocesses and analyzes the input echocardiogram images.

[0070] Step S101: Convert the echocardiogram image into a grayscale image and apply CLAHE (contrast-limited adaptive histogram equalization) to adaptively enhance the contrast of the grayscale image.

[0071] Step S102: Region segmentation based on echocardiogram image layout features.

[0072] Step S103: Use the EAST text detection model to detect text regions in the image. Determine the measurement annotation region based on the position features (top left corner) and content features (including measurement parameters). Measurement region boundary: x=10-200, y=10-150 (pixel coordinates).

[0073] Step S2: Echocardiographic measurement type identification. The system analyzes the text content within the measurement annotation area, detects the keyword "MM" appearing in multiple parameter names, and detects typical M-mode measurement parameters such as "LVIDd", "LVIDs", and "EF". Simultaneously, the main image displays M-mode time-motion curve characteristics. The measurement type is determined to be M-mode measurement.

[0074] Step S3: Perform OCR recognition on the measured and marked area using an optical character recognition algorithm optimized for echocardiography parameters to obtain the OCR recognition result.

[0075] Step S4: Perform semantic analysis on the OCR recognition results and establish the association between parameter names, measurement values, units of measurement, and measurement methods.

[0076] Step S5: Parameter standardization. The parameter names from step S4 are standardized and mapped to obtain standard names, standard units, and standard methods.

[0077] Step S6: Structured output, the system displays the recognition results in tabular form.

[0078] The various embodiments described in this specification are presented in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. All or part of this application can be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, mobile communication terminals, multiprocessor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0079] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of this application.

Claims

1. A method for automatic identification and extraction of echocardiographic image measurement data, characterized in that, include: The input echocardiogram image is preprocessed, and the measurement annotation area is identified and segmented based on a text detection algorithm; Based on the content characteristics of the measurement annotation area, the measurement type of the current image is identified. The measurement type includes at least one of M-mode, two-dimensional, pulse Doppler, continuous Doppler and tissue Doppler measurement. An optical character recognition algorithm optimized for echocardiographic parameters is applied to the measurement and annotation area to identify the parameter name, measurement value, and unit of measurement, and to obtain the OCR recognition result; Perform semantic analysis on the OCR recognition results to establish the association between parameter names, measurement values, units of measurement, and measurement methods; Based on a pre-established echocardiographic parameter database, the identified parameter names are mapped to an echocardiographic parameter terminology system to obtain standardized measurement data. The standardized measurement data is output in a structured format, and data copying and export functions are provided.

2. The method according to claim 1, characterized in that, The text detection algorithm is used to identify and segment the measurement annotation area, and the measurement area is located based on at least one of the following features: Location features: Detects a pre-defined area in the image where measurement values ​​are displayed; Color features: Identify light-colored text areas on a dark background; Layout features: Identify parameter-value-unit text structure arranged in rows; The deep learning text detection model is used to locate candidate text regions, and the measurement and annotation regions are selected by combining the above features.

3. The method according to claim 1, characterized in that, The optical character recognition algorithm optimized for echocardiographic parameters includes: Image preprocessing optimization targeting the features of dark backgrounds and light text in ultrasound images; Character recognition based on a dedicated character set that includes echocardiography terminology, abbreviations, Greek letters, and special symbols; The language model based on the echocardiogram parameter corpus was used to correct the preliminary recognition results; The identification results are verified for the reasonableness of the numerical range and the matching of units.

4. The method according to claim 1, characterized in that, The semantic analysis of the OCR recognition results includes: Line-level structure parsing: Parse the recognition results of each line according to the preset text pattern; Derived parameter identification: Derived parameters are automatically calculated based on the basic measurement values. The derived parameters include at least the E / A ratio, the E / e' ratio, and the pressure gradient calculated from the velocity based on the simplified Bernoulli equation. Measurement method extraction identifies the specific measurement method from parameter name suffixes or context.

5. The method according to claim 1, characterized in that, The echocardiographic parameter database includes at least: A parameter name mapping table used to associate parameter names with standard names from different manufacturers and languages; A unit conversion table used to unify different formats of units of measurement into standard units; Measurement method reference table used to map measurement methods with different formulations to standard methods.

6. The method according to claim 1, characterized in that, The step of outputting the standardized measurement data in a structured format includes: A table showing the identification results grouped by measurement type; A visual alert function that marks values ​​that exceed the preset normal range; Supports copying values ​​and corresponding parameter names individually or in batches; It supports data export in multiple formats.

7. The method according to claim 1, characterized in that, The method also includes a derived parameter verification step: The automatically calculated derived parameters are compared with the derived parameters directly identified from the image for consistency verification. If the verification fails, a prompt will be sent to the user.

8. An automatic identification and extraction system for echocardiographic image measurement data, characterized in that, include: Image input module: used to receive echocardiogram images to be processed; Image preprocessing and measurement area localization module: used to preprocess images and locate measurement and annotation areas; Measurement type identification module: used to identify the measurement type of the current image; Dedicated OCR module: used for optical character recognition in the measurement annotation area optimized for echocardiography; Semantic parsing module: used to analyze OCR results and build relationships between parameter names, values, units, and methods; Parameter standardization module: used to map the identified parameters to a standard terminology system; Structured output module: used to display recognition results and provide interactive copy and export functions; User interface: Used to interact with users, display images and results, and receive user commands and corrections.

9. The system according to claim 8, characterized in that, The system operates as standalone software, interacting with external reporting systems via the clipboard, and is not dependent on any specific ultrasound equipment brand or reporting system.

10. The system according to claim 8, characterized in that, The system also includes a parameter knowledge base management module, a device configuration module, and a recognition history module.

Images