Automatic batch recognition and processing system and method based on medical record homepage data

Abstract

The present invention relates to the technical field of medical record data processing, and specifically discloses an automated batch recognition and processing system and method based on medical record homepage data. The system acquires attribute data of each medical record homepage image belonging to a medical record homepage image set, thereby analyzing the upload quality assessment value of each medical record homepage image, and adjusting the images that do not meet the upload quality assessment threshold requirement. By automatically batch recognizing and processing each medical record homepage image, analyzing the processing accuracy index of each processed medical record homepage image, each medical record homepage image that needs to be reprocessed is identified, and targeted adjustments are made to it, thereby further improving the accuracy and reliability of medical record homepage data recognition and processing. Finally, feedback is given to each medical record homepage image that needs to be reprocessed. The closed-loop processing improves the convenience and efficiency of the automated batch recognition and processing of medical record homepage data, and provides a more convenient and efficient medical record homepage information management solution for medical institutions and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of medical record data processing, and in particular to an automated batch recognition processing system and method based on medical record homepage data. Background Art

[0002] In the modern medical system, the first page of the medical record is an important record of the patient's hospitalization information, which contains a large amount of key data, such as basic patient information, diagnosis, surgical records, discharge status, etc. However, the traditional medical record data management method mainly relies on manual data entry. Manual data entry is not only time-consuming, but also prone to fatigue and negligence due to the heavy workload when faced with a large number of medical records, resulting in slow data entry. In order to solve the above problems, researchers have gradually begun to explore medical record data processing solutions based on automation technology. In recent years, the rapid development of automation technology has provided a good foundation for batch recognition and processing of medical record first page data. These automation solutions can significantly improve the efficiency and accuracy of data processing, reduce manual intervention, and reduce the error rate, thereby providing strong guarantees for the daily operation and decision-making support of medical institutions.

[0003] For example, the invention patent with announcement number CN113836921B announces a method, device and electronic device for digitizing paper case data. First, the initial recognition data of the paper data is obtained, and then the initial recognition data is processed based on a pre-established sentence boundary detection model to obtain a sentence set corresponding to the initial recognition data; further, the sentences in the sentence set are corrected one by one to obtain the target data corresponding to the paper data.

[0004] For example, the invention patent with publication number CN105608325A discloses a new clinical case data collection system and collection method. The verification module of the identification device further includes an electronic case report form comparison unit. The electronic case report form comparison unit receives the first electronic case report form and the second electronic case report form sent by the first OCR module and the second OCR module, and compares and verifies the electronic case report forms sent by the first OCR module and the second OCR module. It can improve the work efficiency of converting paper case report forms into electronic case report forms, effectively reduce the output of erroneous electronic case report forms, and improve the accuracy and speed of the new clinical case data collection system.

[0005] However, in the process of implementing the embodiments of the present application, it was found that the above-mentioned technology has at least the following technical problems: when performing data identification and processing on paper medical records, the prior art often adopts the method of directly analyzing the uploaded paper documents. During the processing, this method can easily lead to significant deviations in the case data in the initial stage of data identification, which not only affects the accuracy of data processing, but also increases the difficulty of subsequent data analysis and application. Summary of the invention

[0006] In view of the deficiencies of the prior art, the present invention provides an automated batch recognition and processing system and method based on medical record front page data, which can effectively solve the problems involved in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: The first aspect of the present invention provides an automated batch recognition and processing system based on medical record homepage data, including: a medical record homepage upload module, used to receive a batch uploaded medical record homepage image set, obtain the attribute data of each medical record homepage image belonging to the medical record homepage image set, and thereby analyze the upload quality evaluation value of each medical record homepage image; a medical record homepage adjustment module, used to compare the upload quality evaluation value of each medical record homepage image with the upload quality threshold, and mark several medical record homepage images whose upload quality evaluation value is less than the upload quality threshold as medical record homepage images that need to be optimized, match and execute the image adjustment set of each medical record homepage image that needs to be optimized, thereby re-marking each medical record homepage image that has completed the execution of the image adjustment set as Each medical record homepage image, and perform automated batch recognition processing on each medical record homepage image; a medical record homepage processing module, used to obtain the data corresponding to each medical record homepage image that has been identified and processed, marked as the electronic data corresponding to each processed medical record homepage image, determine the processing accuracy index of each processed medical record homepage image, and compare it with the processing accuracy threshold, mark several processed medical record homepage images with a processing accuracy index less than the processing accuracy threshold as medical record homepage images that need to be reprocessed, and adjust the processing of each medical record homepage image that needs to be reprocessed; a medical record homepage feedback module, used to obtain the adjusted electronic data corresponding to each medical record homepage image that needs to be reprocessed, evaluate the adjustment optimization factor of each medical record homepage image that needs to be reprocessed, and thereby provide feedback to each medical record homepage image that needs to be reprocessed.

[0008] As a further solution, the matching and execution of the image adjustment set of each medical record homepage image that needs to be optimized, the specific matching process is: obtaining the upload quality assessment value of each medical record homepage image that needs to be optimized, and sorting them in order from small to large, extracting the first-ranked upload quality assessment value, and matching it with the image adjustment set corresponding to each upload quality assessment value interval stored in the image processing database, marking the matched image adjustment set as the image adjustment set of each medical record homepage image that needs to be optimized, thereby executing the image adjustment set of each medical record homepage image that needs to be optimized.

[0009] As a further solution, the automatic batch recognition processing of the medical record front page images is specifically to automatically batch recognize the medical record front page images through the first image recognition processing set executed by the image processing device.

[0010] As a further solution, the feedback is provided for each medical record homepage image that needs to be reprocessed. The specific analysis feedback is: the adjustment optimization factor of each medical record homepage image that needs to be reprocessed is compared with the adjustment optimization threshold. If there is a medical record homepage image that needs to be reprocessed whose adjustment optimization factor is less than the adjustment optimization threshold, the first image recognition processing adjustment set is automatically adjusted and executed until the adjustment optimization factor of each medical record homepage image that needs to be reprocessed is greater than or equal to the adjustment optimization threshold, thereby providing visual adjustment success feedback for each medical record homepage image that needs to be reprocessed.

[0011] The second aspect of the present invention provides a method for applying to the automatic batch recognition and processing system based on medical record homepage data, comprising: S1. receiving a batch uploaded medical record homepage image set, obtaining the attribute data of each medical record homepage image belonging to the medical record homepage image set, and analyzing the upload quality evaluation value of each medical record homepage image; S2. comparing the upload quality evaluation value of each medical record homepage image with the upload quality threshold, and marking a number of medical record homepage images whose upload quality evaluation value is less than the upload quality threshold as medical record homepage images to be optimized, matching and executing the image adjustment set of each medical record homepage image to be optimized, thereby re-marking each medical record homepage image to be optimized after the image adjustment set has been executed as each medical record homepage image. Image, and perform automated batch recognition processing on each medical record homepage image; S3. Obtain the data corresponding to each medical record homepage image that has been identified and processed, mark it as the electronic data corresponding to each processed medical record homepage image, determine the processing accuracy index of each processed medical record homepage image, and compare it with the processing accuracy threshold, mark several processed medical record homepage images whose processing accuracy index is less than the processing accuracy threshold as medical record homepage images that need to be reprocessed, and adjust the processing of each medical record homepage image that needs to be reprocessed; S4. Obtain the adjusted electronic data corresponding to each medical record homepage image that needs to be reprocessed, evaluate the adjustment optimization factor of each medical record homepage image that needs to be reprocessed, and thereby provide feedback to each medical record homepage image that needs to be reprocessed.

[0012] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:

[0013] (1) The present invention provides an automated batch recognition and processing system and method based on medical record homepage data. By acquiring the attribute data of each medical record homepage image belonging to the medical record homepage image set, the upload quality assessment value of each medical record homepage image is analyzed, and the images that do not meet the upload quality assessment threshold requirements are adjusted, thereby ensuring that the medical record homepage image set maintains a high quality during the upload process. By performing automated batch recognition and processing on each medical record homepage image, the processing accuracy index of each processed medical record homepage image is analyzed, thereby identifying each medical record homepage image that needs to be reprocessed, and making targeted adjustments to it, thereby further improving the accuracy and reliability of the medical record homepage data recognition and processing. Finally, feedback is provided for each medical record homepage image that needs to be reprocessed. The closed-loop processing improves the convenience and efficiency of the automated batch recognition and processing of medical record homepage data, and provides a more convenient and efficient medical record homepage information management solution for medical institutions and the like.

[0014] (2) The present invention analyzes the attribute data of each medical record homepage image in the medical record homepage image set to quantify the upload quality assessment value of each medical record homepage image. The present invention can timely adjust the medical record homepage images that do not meet the upload quality assessment threshold requirements, optimize the image attributes, and ensure that the quality of the medical record homepage images meets the requirements of subsequent processing and analysis.

[0015] (3) The present invention determines the electronic data corresponding to each processed medical record front page image and obtains the processing accuracy index of each processed medical record front page image, thereby reflecting the possible problems in the recognition processing process. As a result, the recognition processing flow and method can be timely improved and optimized in a targeted manner, thereby improving the efficiency and accuracy of the recognition processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0017] Figure 1 It is a schematic diagram of system module connection of the present invention.

[0018] Figure 2 The figure is a schematic flow chart of the method steps of the present invention. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0020] Reference Figure 1 As shown, the first aspect of the present invention provides an automated batch recognition and processing system based on medical record front page data, including: a medical record front page upload module, a medical record front page adjustment module, a medical record front page processing module and a medical record front page feedback module.

[0021] The first aspect of the present invention provides an automated batch recognition and processing system based on medical record homepage data, which also includes an image processing database, wherein the image processing database is used to store image contrast boundary values, edge density reference values, total offset distance boundary values, image adjustment sets corresponding to each upload quality assessment value interval, first text filling reference byte numbers corresponding to each first processing and recognition time interval, first field vacancy rate boundary values, first processing and recognition reference time, first recognition error rate boundary values, weight factors corresponding to upload quality assessment values, first image recognition processing adjustment sets corresponding to each processing precision index interval, weight factors corresponding to processing precision index deviation values, influencing factors corresponding to second field vacancy rate deviation value unit values, influencing factors corresponding to second recognition error ratio deviation value unit values, each sampling centroid reference position point, upload quality threshold, processing precision threshold, and adjustment optimization threshold.

[0022] The medical record homepage upload module is connected to the medical record homepage adjustment module, the medical record homepage adjustment module is connected to the medical record homepage processing module, the medical record homepage processing module is connected to the medical record homepage feedback module, and the medical record homepage upload module, the medical record homepage adjustment module, the medical record homepage processing module and the medical record homepage feedback module are all connected to the image processing database.

[0023] The medical record homepage upload module is used to receive a batch of uploaded medical record homepage image sets, obtain attribute data of each medical record homepage image belonging to the medical record homepage image set, and thereby analyze the upload quality evaluation value of each medical record homepage image.

[0024] In a specific embodiment, the present invention quantifies the upload quality assessment value of each medical record homepage image by analyzing the attribute data of each medical record homepage image belonging to the medical record homepage image set, and can timely adjust the medical record homepage images that do not meet the upload quality assessment threshold requirements, optimize the image attributes, and ensure that the quality of the medical record homepage images meets the requirements of subsequent processing and analysis.

[0025] Specifically, the upload quality assessment value of each medical record homepage image is analyzed, and the specific analysis process is as follows:

[0026] The average brightness of the background area of ​​each medical record homepage image and the average brightness of the text area of ​​each medical record homepage image are extracted from the attribute data of each medical record homepage image, and the image contrast of each medical record homepage image is obtained by data processing; the average brightness of the background area of ​​each medical record homepage image and the average brightness of the text area of ​​each medical record homepage image can be obtained by first dividing the medical record homepage image into a background area and a text area through image processing technology (such as Canny edge detection algorithm), and then using functions in an image processing library (such as OpenCV or Pillow) to identify and calculate the average brightness of the background area and the average brightness of the text area; the image contrast of the above-mentioned medical record homepage images, the specific data processing process is: the average brightness of the background area of ​​each medical record homepage image and the corresponding average brightness of the text area are accumulated, and at the same time, the average brightness of the background area of ​​each medical record homepage image and the corresponding average brightness of the text area are differenced and absolute value processed, and the absolute value processing result is compared with the corresponding accumulation processing result, thereby obtaining the image contrast of each medical record homepage image.

[0027] The total number of edge pixels of each medical record homepage image and the total number of image pixels of each medical record homepage image are extracted from the attribute data of each medical record homepage image, and the total number of edge pixels of each medical record homepage image and the total number of image pixels of each medical record homepage image are ratio-processed to obtain the edge density of each medical record homepage image; the total number of edge pixels of each medical record homepage image is obtained by identifying and extracting the edge information in each medical record homepage image through an edge detection algorithm (such as Canny edge detection), and the edge information is usually represented in the form of pixel points. The total number of edge pixels can be obtained by counting the number of edge pixel points obtained after edge detection; the total number of image pixels of each medical record homepage image is obtained by reading the attributes of each medical record homepage image by an image processing device.

[0028] The sampling centroid position points of each medical record homepage image are extracted from the attribute data of each medical record homepage image, and the reference position points of each sampling centroid are obtained. The offset distances of the sampling centroid position points of each medical record homepage image are obtained through distance analysis, and the accumulated processing is performed, and the accumulated processing results are marked as the total offset distances of the sampling centroid position points of each medical record homepage image; the above-mentioned sampling centroid position points refer to a number of sub-areas randomly arranged in each medical record homepage image, and these sub-areas are closely distributed along the edge of the medical record homepage image. The pixel coordinates in the several sub-areas are averaged through image processing libraries such as OpenCV, and the average processing results are marked as the total offset distances of the sampling centroid position points of each medical record homepage image. Sampling centroid position points; the above-mentioned sampling centroid reference position points are extracted from the image processing database; it should be explained that the images of the first page of each medical record are uploaded to the image processing equipment in a unified and standardized manner to ensure their consistency in spatial position. Based on this unified upload mechanism, a series of standardized centroid reference position points can be established as the benchmark for subsequent analysis. On this basis, the reference position points of each sampling centroid are accurately marked and positioned by reviewing the historical upload records and combining the professional knowledge of image processing. These reference position points not only reflect the spatial layout of the image when it is uploaded, but also provide an important reference basis for the subsequent sampling centroid calculation and analysis.

[0029] Thus, the analysis method for comprehensively obtaining the upload quality evaluation value of each medical record homepage image is used. In this embodiment, the upload quality evaluation value of each medical record homepage image is obtained by comprehensively analyzing the image contrast, edge density and the total offset distance of the sampling centroid position point of each medical record homepage image, and is used to determine the value of the image quality of each medical record homepage image after uploading. The specific analysis method is as follows:

[0030] ;

[0031] It is the image contrast of the a-th medical record homepage image, which refers to the absolute value of the difference between the average brightness of the background area and the average brightness of the text area in the a-th medical record homepage image, and the ratio of the average brightness of the background area to the average brightness of the text area in the a-th medical record homepage image, reflecting the contrast of the brightness and darkness of the medical record homepage image in visual presentation.

[0032] is the edge density of the a-th medical record homepage image, which refers to the ratio of the pixel value of the edge of the a-th medical record homepage image (i.e., the area of ​​the text and table frame in the a-th medical record homepage image) to the overall number of pixels of the a-th medical record homepage image. For the a-th medical record homepage image, its edge density reflects the richness of the detail information in the image.

[0033] It is the total offset distance of the sampling centroid position points belonging to the a-th medical record homepage image, which refers to the total deviation distance of all the sampling centroid position points in the a-th medical record homepage image relative to the corresponding sampling centroid reference position points. Analyzing the total offset distance of the sampling centroid position points can also help evaluate the degree of rotation or translation of the image.

[0034] e is a natural constant, a is the number of the image on the front page of each medical record, , c is the total number of images on the medical record homepage, It is the image contrast limit value preset in the image processing database, indicating the minimum allowable image contrast value of each medical record front page image.

[0035] It is the edge density reference value preset in the image processing database, which represents the reference value of the edge density of each medical record front page image.

[0036] It is the total offset distance limit value preset in the image processing database, which represents the maximum allowable total offset distance of the sampling centroid position point of each medical record homepage image.

[0037] is the upload quality assessment value of the ath medical record homepage image. When the contrast of the medical record homepage image is large, the brightness difference between the text and the background in the medical record homepage image becomes more obvious, which makes the edge (i.e., the boundary between the text and the background) easier to identify. In the high-contrast medical record homepage image, the text edge may be clearer, so that the text edge density is less different from the reference edge density. If the image contrast is abnormal or the image contrast is low, the text edge in the medical record homepage image may be over-sharpened, resulting in unnecessary edges or noise, thereby affecting the rationality of the edge density. At the same time, the offset of the centroid position may be related to the image blur or distortion caused by insufficient contrast. In images with low contrast, the edges of text elements may become blurred, making the determination of the centroid position inaccurate. This offset may affect the layout and typesetting of the image, and then affect the readability and overall quality of the medical record homepage image. This shows that the image quality of the medical record homepage image is obviously insufficient when it is uploaded, and further image optimization is needed. Therefore, by analyzing the above parameters, the medical record homepage image with image quality problems during the upload process can be identified, so that precise optimization measures can be taken to obtain higher quality medical record homepage images and more accurate upload results.

[0038] In this exemplary embodiment, the table showing the changes in the upload quality assessment value of the ath medical record homepage image and its corresponding parameters is shown in Table 1:

[0039] Table 1 The upload quality evaluation value of the ath medical record homepage image and its corresponding parameter changes

[0040] Image contrast / percentage Edge Density / Percentage Total offset distance / mm Upload quality rating / percentage 30 4.6 5.3 251 25 5.2 5.7 231 22 3.8 5.9 221

[0041] In this example embodiment, the image contrast limit value is set to 20%, the edge density reference value is set to 4.5%, and the total offset distance limit value is set to 6 mm. It can be seen from Table 1 that when the image contrast is large (30%), the edge density is slightly different from the edge density reference value (4.6% and 4.5% only differ by 0.1%). At this time, the total offset distance is also small (5.3 mm). These work together to make the upload quality of the a-th medical record homepage image relatively good, which is reflected in the large upload quality assessment value of 251%.

[0042] The medical record homepage adjustment module is used to compare the upload quality assessment value of each medical record homepage image with the upload quality threshold, and mark several medical record homepage images whose upload quality assessment value is less than the upload quality threshold as medical record homepage images that need to be optimized, match and execute the image adjustment set of each medical record homepage image that needs to be optimized, thereby re-marking each medical record homepage image that has completed the execution of the image adjustment set as each medical record homepage image, and performing automatic batch recognition processing on each medical record homepage image.

[0043] The above-mentioned automatic batch recognition processing is performed on the medical record homepage images, wherein each medical record homepage image includes a number of medical record homepage images with an upload quality assessment value greater than or equal to the upload quality threshold, and also includes a number of medical record homepage images that need to be optimized after being adjusted by the image adjustment set; the above-mentioned upload quality threshold represents the minimum value of a reasonable range of the upload quality assessment value of each medical record homepage image, which is extracted from the image processing database. For a number of medical record homepage images with an upload quality assessment value greater than or equal to the upload quality threshold, it indicates that they can meet the upload quality requirements and do not need to perform image optimization adjustment, that is, there is no need to optimize the image adjustment set, and image recognition processing can be directly performed.

[0044] Specifically, the automated batch recognition processing of each medical record front page image is specifically automated batch recognition processing of each medical record front page image through the first image recognition processing set executed by the image processing device; the above-mentioned first image recognition processing set is the default image recognition processing flow in the image processing device, which is formulated based on widely recognized professional knowledge and practical experience in the industry, has a high degree of universality and stability, and can provide reliable technical support for various image recognition tasks; wherein the first image recognition processing set includes: using an edge detection algorithm based on machine learning to identify the table area in the uploaded medical record front page image, and after successfully identifying the table area, using the contour extraction algorithm in the image processing technology to extract the coordinate information of each cell in the table area, and then using optical character recognition The technology is used to recognize text information and post-process the optical character recognition results, such as correcting typos and merging similar characters, and the post-processing results, that is, the text information, are matched with the corresponding cell coordinates to determine the cell to which each character belongs. Finally, a programming language (such as Python) and related data processing libraries (such as Pandas) are used to automatically fill in the spreadsheet corresponding to the medical record front page image. Specifically, a script written in Python reads the text information and the corresponding cell coordinate information corrected by the optical character recognition technology, uses the Pandas library to create and operate a spreadsheet, and fills the corrected text information into the spreadsheet corresponding to the medical record front page image according to the predetermined format and position, thereby completing the automated batch recognition processing of each medical record front page image.

[0045] The first image recognition processing set is the default image recognition processing flow in the image processing device. Therefore, during the actual image recognition processing process, it may be interfered by the performance of the image processing device and the characteristics of the medical record front page image set. This may cause some problems when the image processing device executes the first image recognition processing set to perform automated batch recognition processing on each medical record front page image. Therefore, further analysis and adjustment are needed.

[0046] It should be explained that the core functions of image processing equipment include but are not limited to: Image preprocessing: It can perform preprocessing operations on the input raw image, such as denoising, contrast enhancement, brightness adjustment, color distortion correction and geometric distortion correction, etc., to optimize image quality and lay a solid foundation for subsequent processing steps; Feature extraction: Using existing algorithms and technologies to automatically extract key feature information from the image, such as edges, textures, shapes and colors, which is crucial for subsequent image recognition, classification and understanding; Image recognition and classification: Based on advanced technologies such as deep learning and machine learning, image processing equipment can accurately recognize and classify objects, text and symbols in the image, providing strong support for the understanding and application of image content; Image transformation and enhancement: It can perform various transformation operations on the image, such as scaling, rotating and cropping, to meet the needs of different application scenarios, and at the same time, it can enhance the visual effect of the image through enhancement technology, making it clearer and easier to observe; Image analysis and understanding: On the basis of extracting features and identifying objects, further in-depth analysis and understanding of the image content is carried out, such as detecting anomalies in the image and identifying relationships in the scene, providing decision-making basis for advanced image applications.

[0047] Furthermore, the image adjustment set of each medical record front page image to be optimized is matched and executed, and the specific matching process is:

[0048] Obtain the upload quality assessment value of each medical record homepage image that needs to be optimized, and sort them in ascending order, extract the first-ranked upload quality assessment value, and match them with the image adjustment sets corresponding to each upload quality assessment value interval stored in the image processing database, mark the matched image adjustment set as the image adjustment set of each medical record homepage image that needs to be optimized, and thus execute the image adjustment set of each medical record homepage image that needs to be optimized; it should be explained that each medical record homepage image that needs to be optimized belongs to each medical record homepage image, so the upload quality assessment value of each medical record homepage image that needs to be optimized can be obtained from the upload quality assessment value of each medical record homepage image; in an example embodiment, assuming that the above-mentioned first-ranked upload quality assessment value is G, in the upload quality assessment value interval [G-45%, G+35%] stored in the image processing database, the image adjustment set corresponding to the upload quality assessment value interval [G-45%, G+35%] includes: histogram equalization, Gamma transform, Laplace transform, bilinear interpolation and Gaussian filter Wave algorithm, first use histogram equalization and Gamma transform to adjust the grayscale distribution and brightness of the image to enhance the contrast of the image, then use the second-order differential operation in Laplace transform to highlight the edge of the image, enhance the details and clarity of the image, finally use bilinear interpolation to improve the resolution of the image to further enhance the image details and use Gaussian filtering algorithm to eliminate the noise in the image to further improve the image clarity; therefore, the image adjustment set that needs to optimize the image of the medical record homepage includes: histogram equalization, Gamma transform, Laplace transform, bilinear interpolation and Gaussian filtering algorithm, first use histogram equalization and Gamma transform to adjust the grayscale distribution and brightness of the image to enhance the contrast of the image, then use the second-order differential operation in Laplace transform to highlight the edge of the image, enhance the details and clarity of the image, finally use bilinear interpolation to improve the resolution of the image to further enhance the image details and use Gaussian filtering algorithm to eliminate the noise in the image to further improve the image clarity.

[0049] It should also be explained that the image adjustment set is matched using the minimum value of the uploaded quality assessment values ​​of each medical record homepage image that needs to be optimized. For the medical record homepage images that need to be optimized whose uploaded quality assessment values ​​are less than the uploaded quality assessment threshold but greater than or equal to the minimum of the uploaded quality assessment values, the image adjustment set is also applicable and may even bring better recognition processing effects, significantly reducing the complexity of matching the uploaded quality assessment values ​​of each medical record homepage image that needs to be optimized one by one, and improving the efficiency of image optimization. The image adjustment set matches a personalized adjustment plan based on the characteristics and requirements of each medical record homepage image to be optimized. Therefore, each medical record homepage image that needs to be optimized after adjustment by the image adjustment set can meet the requirements of the uploaded quality threshold, and therefore can be re-marked as each medical record homepage image.

[0050] The medical record homepage processing module is used to obtain data corresponding to each medical record homepage image that has been identified and processed, mark it as electronic data corresponding to each processed medical record homepage image, determine the processing accuracy index of each processed medical record homepage image, and compare it with the processing accuracy threshold, mark several processed medical record homepage images whose processing accuracy index is less than the processing accuracy threshold as medical record homepage images that need to be reprocessed, and process and adjust each medical record homepage image that needs to be reprocessed.

[0051] The above-mentioned processing accuracy threshold represents the minimum value of the reasonable range of the processing accuracy index of each processed medical record front page image, which is extracted from the image processing database. For several processed medical record front page images whose processing accuracy index is greater than or equal to the processing accuracy threshold, it indicates that their recognition meets the requirements of the processing accuracy threshold, that is, they can be accurately identified and processed and filled into the corresponding electronic data table, so no processing adjustment is required.

[0052] In a specific embodiment, the present invention determines the electronic data corresponding to each processed medical record front page image to obtain the processing accuracy index of each processed medical record front page image, thereby reflecting the possible problems in the recognition processing process, thereby timely and targetedly improving and optimizing the recognition processing flow and method, thereby improving the efficiency and accuracy of the recognition processing.

[0053] Specifically, the image of each medical record homepage that needs to be reprocessed is processed and adjusted, and the specific adjustment process is:

[0054] Obtain the processing accuracy index of each medical record homepage image that needs to be reprocessed, and sort them in ascending order, extract the processing accuracy index ranked first, and match them with the first image recognition processing adjustment set corresponding to each processing accuracy index interval stored in the image processing database, thereby marking the obtained first image recognition processing adjustment set as the first image recognition processing adjustment set of each medical record homepage image that needs to be reprocessed; thereby, the image processing device processes and adjusts each medical record homepage image that needs to be reprocessed through the first image recognition processing adjustment set of each medical record homepage image that needs to be reprocessed; the above-mentioned medical record homepage images that need to be reprocessed refer to the partial images that need to be adjusted in each processed medical record homepage image, so the processing accuracy index of each medical record homepage image that needs to be reprocessed can be obtained from the processing accuracy index of each processed medical record homepage image; in an example embodiment, assuming that the above-mentioned processing accuracy index ranked first is H, in the processing accuracy index interval [H-41%, H+29%] stored in the image processing database, the first image recognition processing adjustment set corresponding to the processing accuracy index interval [H-41%, H+29%] includes: using Canny The edge detection algorithm is used to identify the table area in the uploaded medical record front page image. After the table area is successfully identified, the contour extraction algorithm in the image processing technology is used to extract the coordinate information of each cell in the table area. Then, the optical character recognition technology is used to recognize the text information, and the recognition parameters of the optical character recognition engine (such as page segmentation mode, character recognition threshold, etc.) are adjusted to optimize the recognition effect. The optical character recognition results are post-processed, such as correcting typos, merging similar characters, etc., and the post-processing results, that is, the text information, are matched with the corresponding cell coordinates to determine the cell to which each character belongs. Finally, a programming language (such as Python) and related data processing libraries (such as Pandas) are used to automatically fill in the electronic form corresponding to the medical record front page image. Specifically, the text information and the corresponding cell coordinate information corrected by the optical character recognition technology are read through a script written in Python, and the Pandas library is used to create and operate an electronic form, and the corrected text information is filled into the electronic form corresponding to the medical record front page image according to the predetermined format and position, thereby completing the automatic batch recognition processing of each medical record front page image;The first image recognition processing adjustment set for each medical record homepage image that needs to be reprocessed includes: using the Canny edge detection algorithm to identify the table area in the uploaded medical record homepage image, after successfully identifying the table area, using the contour extraction algorithm in the image processing technology to extract the coordinate information of each cell in the table area, then using the optical character recognition technology to recognize the text information, and adjusting the recognition parameters of the optical character recognition engine (such as page segmentation mode, character recognition threshold, etc.) to optimize the recognition effect, post-processing the optical character recognition results, such as correcting typos, merging similar characters, etc., and matching the post-processing results, that is, the text information with the corresponding cell coordinates to determine the cell to which each text belongs, and finally using a programming language (such as Python) and a related data processing library (such as Pandas) to automatically fill in the electronic form corresponding to the medical record homepage image, specifically: reading the text information and the corresponding cell coordinate information corrected by the optical character recognition technology through a script written in Python, creating and operating an electronic form using the Pandas library, and filling the corrected text information into the electronic form corresponding to the medical record homepage image according to a predetermined format and position, thereby completing the automatic batch recognition processing of each medical record homepage image that needs to be reprocessed. ;

[0055] It needs to be explained that, in the process of adjusting the first image recognition processing set, the parameters in the OCR technology, the upload quality assessment value of each medical record homepage image and the image adjustment set are mainly adjusted, and the optimal recognition conditions are ensured after each medical record homepage image is uploaded. Therefore, when the processing accuracy of the medical record homepage image is improved, the performance of the image recognition processing technology, that is, the parameters in the OCR technology, is mainly improved, and efforts are made to achieve high-precision recognition of text and tables in the medical record homepage image; it needs to be explained that the minimum value of the processing accuracy index of each medical record homepage image that needs to be reprocessed is used to match the first image recognition processing adjustment set. For medical record homepage images whose processing accuracy index is less than the processing accuracy threshold but greater than or equal to the minimum value of the processing accuracy index, the first image recognition processing adjustment set is also applicable, and may even bring better recognition processing effects, which significantly reduces the complexity of matching the processing accuracy index of each medical record homepage image that needs to be reprocessed one by one, and improves the recognition processing efficiency.

[0056] The above-mentioned processing and adjustment of the medical record front page images that need to be reprocessed also requires first clearing the electronic data corresponding to the medical record front page images that need to be reprocessed, and then re-recognizing, processing and filling.

[0057] Furthermore, the specific determination process of determining the processing accuracy index of each processed medical record front page image is as follows:

[0058] Perform automated batch recognition processing on the front page images of each medical record to monitor, monitor and obtain data corresponding to each medical record front page image that has been recognized and processed within the first monitoring cycle, and mark them as electronic data corresponding to each processed medical record front page image; the above-mentioned first monitoring cycle refers to the time period in which all medical record front page images have completed recognition processing, and the specific duration is extracted from the timestamp of the image processing device; the electronic data corresponding to each processed medical record front page image specifically refers to the electronic data fields that need to be filled or completed after the uploaded medical record front page images have been recognized and processed. These electronic data fields usually contain key data recognized and extracted from the image, such as basic patient information, diagnosis information, treatment records, etc. In short, these electronic data are a digital presentation of the content of the processed medical record front page images, ensuring the accuracy and traceability of the information.

[0059] It should be explained that the first and second prefixes in the following parameters are used to clearly distinguish different time periods. Specifically, for example, the first field vacancy rate is the field vacancy rate within the first monitoring period, and the second field vacancy rate is the field vacancy rate within the second monitoring period. This naming method is intended to clearly define and distinguish the relevant indicators or status of each time period to facilitate more accurate analysis and comparison.

[0060] The number of first filled fields belonging to each processed medical record homepage image and the number of first fields to be filled belonging to each processed medical record homepage image are extracted from the electronic data corresponding to each processed medical record homepage image, and the number of first filled fields belonging to each processed medical record homepage image and the number of first fields to be filled corresponding to each processed medical record homepage image are processed, and the processing result is marked as the vacancy rate of the first field belonging to each processed medical record homepage image; the number of first filled fields belonging to each processed medical record homepage image and the number of first fields to be filled belonging to each processed medical record homepage image are extracted from the processing log of the image processing device; the specific data processing process of the vacancy rate of the first field is: the number of the first filled fields and the first number of fields to be filled are processed by ratio, and then the ratio processing result is subtracted from the unit "1", which is the vacancy rate of the first field.

[0061] The first processing recognition time of each processed medical record homepage image is extracted from the electronic data corresponding to each processed medical record homepage image, and matched with the first text filling reference byte number corresponding to each first processing recognition time interval stored in the image processing database, thereby obtaining the first text filling reference byte number belonging to each processed medical record homepage image, and at the same time, the first text filling byte number belonging to each processed medical record homepage image is extracted from the electronic data corresponding to each processed medical record homepage image; the first processing recognition time of each processed medical record homepage image is extracted from the timestamp log of the image processing device; the first text filling byte number belonging to each processed medical record homepage image is extracted from the processing log of the image processing device; the first text filling reference byte number belonging to each processed medical record homepage image, the specific matching process is: query the first processing recognition time interval corresponding to the first processing recognition time of each processed medical record homepage image in the image processing database, then the first text filling reference byte number corresponding to the first processing recognition time interval belonging to the first processing recognition time of each processed medical record homepage image is the first text filling reference byte number belonging to each processed medical record homepage image.

[0062] The number of first sampled recognition word errors of each processed medical record homepage image is extracted from the electronic data corresponding to each processed medical record homepage image, and the total number of first sampled recognition words of each processed medical record homepage image is obtained at the same time. The number of first sampled recognition word errors of each processed medical record homepage image and the total number of first sampled recognition words corresponding to each processed medical record homepage image are ratio processed, and the ratio processing result is marked as the first recognition error ratio of each processed medical record homepage image. The specific acquisition process of the number of first sampled recognition word errors of each processed medical record homepage image is as follows: words are extracted from the electronic data corresponding to each processed medical record homepage image through a bag of words model, and the total number of first sampled recognition words of each processed medical record homepage image is counted. Each extracted word is accurately compared with the words in the standard proper noun library through an exact matching algorithm. If exactly the same words are found, the match is considered successful; otherwise, the match is considered failed and the word is marked as a sampled recognition word error.

[0063] Obtain the upload quality assessment value of each processed medical record homepage image, thereby comprehensively obtaining the processing accuracy index of each processed medical record homepage image. In this embodiment, the processing accuracy index of each processed medical record homepage image is obtained by comprehensively analyzing the upload quality assessment value of each processed medical record homepage image, the vacancy rate of the first field, the first processing recognition time, the number of bytes of the first text filling, and the first recognition error rate. It is used to determine the numerical value of the recognition processing accuracy of each processed medical record homepage image. The specific determination method is:

[0064] ;

[0065] is the upload quality assessment value of the pth processed medical record homepage image. The processed medical record homepage image is the name of the medical record homepage image after recognition processing. Therefore, the upload quality assessment value of each medical record homepage image corresponds to the upload quality assessment value of each processed medical record homepage image. For example, the upload quality assessment value of the 5th medical record homepage image is the upload quality assessment value of the 5th processed medical record homepage image. High-quality image upload can ensure the accuracy of subsequent processing steps such as medical record homepage image recognition processing, reduce recognition errors caused by blurred medical record homepage images, and thus lay the foundation for a high level of accurate medical record homepage image processing. Therefore, the upload quality assessment value is incorporated into the processing accuracy index.

[0066] It is the vacancy rate of the first field of the pth processed medical record front page image, which refers to the proportion of vacant (i.e., not filled in or unrecognizable) fields in the electronic data field corresponding to the pth processed medical record front page image. A high vacancy rate indicates that there is a lot of missing information.

[0067] The first processing recognition time of the pth processed medical record front page image refers to the time taken from the start of the recognition process to the completion of the recognition process of all fields of the pth medical record front page image.

[0068] The number of bytes filled in the first text of the pth processed medical record front page image refers to the total number of bytes filled with characters (including text, letters, numbers, symbols, etc.) in the electronic data field corresponding to the pth processed medical record front page image.

[0069] It is the first recognition error rate of the pth processed medical record front page image, which refers to the proportion of words with errors (such as inconsistent with standard terms, etc.) in the words filled in the corresponding electronic data field of the pth processed medical record front page image.

[0070] The reference number of bytes filled in the first text of the p-th processed medical record homepage image represents the reference value of the number of bytes filled in the first text of the p-th processed medical record homepage image.

[0071] It is the first field vacancy rate limit value preset in the image processing database, indicating the maximum allowable vacancy rate of the first field of each processed medical record front page image.

[0072] It is the first processing recognition reference time preset in the image processing database, and represents the reference value of the first processing recognition time of each processed medical record front page image.

[0073] It is a first recognition error ratio limit value preset in the image processing database, indicating the maximum allowable first recognition error ratio of each processed medical record front page image.

[0074] e is a natural constant, p is the number of each processed medical record front page image, , t is the total number of processed medical record front page images, where t is equal to c.

[0075] It is the weight factor corresponding to the uploaded quality assessment value preset in the image processing database, which indicates the proportion of the uploaded quality assessment value to the processing precision index. When in use, the weight factor corresponding to the uploaded quality assessment value can be directly obtained from the image processing database, and the corresponding relationship can be a preset mapping relationship. For example, the image contrast, edge density, and the total offset distance of the sampling centroid position point form a mapping set with the weight factor corresponding to the uploaded quality assessment value preset in the image processing database. The real-time image contrast, edge density, and the total offset distance of the sampling centroid position point are input into the mapping set to obtain the weight factor corresponding to the uploaded quality assessment value. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0076] in, is the processing accuracy index of the pth processed medical record front page image. In the recognition process of the medical record front page image, the field vacancy rate is an important indicator to measure the quality of the image recognition processing performance. When the field vacancy rate is high, it directly reflects that there is an obvious performance deficiency in the recognition and processing of the text information of the medical record front page image. This performance defect not only leads to the omission of a large amount of key medical information, directly leads to an increase in the recognition error rate, but also makes the entire recognition process time-consuming significantly increased, significantly deviating from the preset reference recognition processing time. Since the reference value of the number of text filling bytes is set based on the actual recognition processing time, when the actual processing time is greatly extended due to the poor image recognition processing performance, the actual number of text bytes filled will significantly deviate from the reference number of text filling bytes. This significant deviation further highlights the insufficiency of the image recognition processing performance and further reduces the accuracy of the recognition processing. Therefore, the above parameters are intertwined and influence each other, which together lead to a significant decrease in the overall accuracy of the medical record front page image recognition processing.

[0077] The medical record homepage feedback module is used to obtain the adjusted electronic data corresponding to each medical record homepage image that needs to be reprocessed, evaluate the adjustment optimization factor of each medical record homepage image that needs to be reprocessed, and thus provide feedback for each medical record homepage image that needs to be reprocessed.

[0078] Specifically, the adjustment optimization factors of each medical record front page image that needs to be reprocessed are evaluated, and the specific evaluation process is as follows:

[0079] Based on the processing and adjustment of each medical record front page image that needs to be reprocessed, monitor and obtain the adjusted electronic data corresponding to each medical record front page image that needs to be reprocessed within the second monitoring cycle; the above-mentioned second monitoring cycle refers to the time period when all medical record front page images that need to be reprocessed have completed the re-identification processing, and the specific duration is extracted from the timestamp of the image processing device.

[0080] The vacancy rate of the second field of each medical record front page image that needs to be reprocessed and the second recognition error rate of each medical record front page image that needs to be reprocessed are extracted from the adjusted electronic data corresponding to each medical record front page image that needs to be reprocessed; the vacancy rate of the second field of each medical record front page image that needs to be reprocessed is consistent with the process of obtaining the vacancy rate of the first field of each processed medical record front page image; the second recognition error rate of each medical record front page image that needs to be reprocessed is consistent with the process of obtaining the first recognition error rate of each processed medical record front page image.

[0081] The vacancy rate of the first field of each medical record homepage image that needs to be reprocessed and the first recognition error rate of each medical record homepage image that needs to be reprocessed are obtained, and at the same time, the processing accuracy index of each medical record homepage image that needs to be reprocessed is combined to obtain an evaluation method for the adjustment optimization factor of each medical record homepage image that needs to be reprocessed; it should be explained that each medical record homepage image that needs to be reprocessed refers to a part of the medical record homepage image that needs to be adjusted, so the vacancy rate of the first field of each medical record homepage image that needs to be reprocessed, the first recognition error rate of each medical record homepage image that needs to be reprocessed, and the processing accuracy index of each medical record homepage image that needs to be reprocessed can all be obtained from the parameters of each processed medical record homepage image.

[0082] Specifically, the adjustment optimization factor of each medical record homepage image that needs to be reprocessed, in this embodiment, is obtained by comprehensive analysis of the processing accuracy index, processing accuracy threshold, second field vacancy rate, second recognition error rate, first field vacancy rate and first recognition error rate of each medical record homepage image that needs to be reprocessed, and is used to determine the value of the adjustment optimization degree of each medical record homepage image that needs to be reprocessed. The specific evaluation method is as follows:

[0083] ;

[0084] It is the processing precision threshold preset in the image processing database. is the processing accuracy index of the sth medical record homepage image that needs to be reprocessed. If the processing accuracy index of the sth medical record homepage image that needs to be reprocessed is significantly different from the processing accuracy threshold, it means that the vacancy rate of the second field needs to be significantly reduced relative to the corresponding vacancy rate of the first field, and the second recognition error rate needs to be significantly reduced relative to the corresponding first recognition error rate. The adjustment range needs to meet the difference requirement between the processing accuracy index and the processing accuracy threshold.

[0085] It is the weight factor corresponding to the processing accuracy index deviation value preset in the image processing database, which indicates the proportion of the processing accuracy index deviation value in the adjustment optimization factor. When used, the weight factor corresponding to the processing accuracy index deviation value can be directly obtained from the image processing database, and the corresponding relationship can be a preset mapping relationship. For example, the processing accuracy index deviation value and the weight factor corresponding to the processing accuracy index deviation value preset in the image processing database form a mapping set, and the real-time processing accuracy index deviation value is input into the mapping set to obtain the weight factor corresponding to the processing accuracy index deviation value. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0086] is the vacancy rate of the second field of the sth medical record front page image that needs to be reprocessed, is the second recognition error rate of the sth medical record front page image that needs to be reprocessed, is the vacancy rate of the first field of the sth medical record front page image that needs to be reprocessed, is the first recognition error rate of the sth medical record front page image that needs to be reprocessed. It needs to be explained that the field vacancy rate and the recognition error rate are two important indicators of the medical record front page image recognition processing performance, and they are also a direct reflection of the medical record front page image recognition processing performance. The level of these two indicators is directly related to the optimization degree of the recognition processing flow and the practical value of the medical record front page image recognition processing. Therefore, by analyzing the field vacancy rate and the recognition error rate, the optimization degree of the medical record front page image recognition processing performance can be more effectively evaluated.

[0087] It is the impact factor corresponding to the unit value of the second field vacancy rate deviation value preset in the image processing database, which represents the numerical value of the influence of the unit value of the deviation value of the second field vacancy rate relative to the first field vacancy rate on the adjustment optimization factor. When used, the impact factor corresponding to the unit value of the second field vacancy rate deviation value can be directly obtained from the image processing database, and the corresponding relationship can be a preset mapping relationship. For example, the difference between the second field vacancy rate and the first field vacancy rate and the impact factor corresponding to the unit value of the second field vacancy rate deviation value preset in the image processing database form a mapping set, and the real-time difference between the second field vacancy rate and the first field vacancy rate is input into the mapping set to obtain the impact factor corresponding to the unit value of the second field vacancy rate deviation value. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0088] It is the influence factor corresponding to the unit value of the second recognition error ratio deviation value preset in the image processing database, which represents the influence degree of the unit value of the deviation value of the second recognition error ratio relative to the first recognition error ratio on the adjustment optimization factor. When used, the influence factor corresponding to the unit value of the second recognition error ratio deviation value can be directly obtained from the image processing database, and the corresponding relationship can be a preset mapping relationship. For example, the difference between the second recognition error ratio and the first recognition error ratio and the influence factor corresponding to the unit value of the second recognition error ratio deviation value preset in the image processing database form a mapping set, and the real-time difference between the second recognition error ratio and the first recognition error ratio is input into the mapping set to obtain the influence factor corresponding to the unit value of the second recognition error ratio deviation value. The mapping relationship can be one-to-one or many-to-one. In this example, its value range is [0, 1].

[0089] e is a natural constant, s is the number of the medical record front page image that needs to be reprocessed, , w is the total number of medical record front page images that need to be reprocessed, where w is less than or equal to t.

[0090] in, is the adjustment optimization factor of the sth medical record homepage image that needs to be reprocessed. Assuming that the processing accuracy index of a medical record homepage image that needs to be reprocessed deviates significantly from the processing accuracy threshold, it indicates that the medical record homepage image that needs to be reprocessed has many problems after being processed and recognized by the default first image recognition processing set. These problems include but are not limited to the lack of field information (i.e., high field vacancy rate) and frequent recognition errors (i.e., high recognition error rate). Therefore, in order to make the medical record homepage image that needs to be reprocessed meet the quality requirements of subsequent use or analysis, after adjustment, the field vacancy rate and recognition error rate of the medical record homepage image that needs to be reprocessed in the second monitoring cycle must be significantly reduced relative to the field vacancy rate and recognition error rate in the first monitoring cycle. The degree of adjustment must be large enough to make the adjustment optimization factor of the medical record homepage image that needs to be reprocessed meet the requirements of the adjustment optimization threshold, so that the recognition of the medical record homepage image that needs to be reprocessed The processing can meet the standards and requirements of actual use, and then be effectively applied to subsequent medical information management. Assuming that the processing accuracy index of a medical record homepage image that needs to be reprocessed is not much different from the processing accuracy threshold, the field vacancy rate and recognition error rate of the medical record homepage image that needs to be reprocessed in the second monitoring cycle are less reduced than the field vacancy rate and recognition error rate in the first monitoring cycle. Although the room for improvement may be relatively limited, through appropriate adjustment measures, the recognition processing quality of the adjusted image can still be effectively improved, thereby meeting the standards and requirements of actual use. Therefore, by analyzing the parameters in the adjustment optimization factor and comparing the degree of change before and after the adjustment, it is possible to intuitively evaluate whether the medical record homepage image that needs to be reprocessed has met the standards and requirements of actual use. At the same time, it can also reduce the time to obtain additional parameters and reduce the analysis errors caused by improper parameter settings or interactions between parameters.

[0091] Furthermore, the feedback for each medical record front page image that needs to be reprocessed is specifically analyzed and fed back as follows:

[0092] The adjustment optimization factor of each medical record homepage image that needs to be reprocessed is compared with the adjustment optimization threshold. If there is a medical record homepage image that needs to be reprocessed whose adjustment optimization factor is less than the adjustment optimization threshold, the first image recognition processing adjustment set is automatically adjusted and executed until the adjustment optimization factor of each medical record homepage image that needs to be reprocessed is greater than or equal to the adjustment optimization threshold, thereby performing visual adjustment success feedback for each medical record homepage image that needs to be reprocessed; the above-mentioned adjustment optimization threshold refers to the minimum value of the reasonable range of the adjustment optimization factor of each medical record homepage image that needs to be reprocessed, which is extracted from the image processing database; the above-mentioned automatic adjustment of the first image recognition processing adjustment set refers to the image processing device automatically making slight adjustments to the first image recognition processing adjustment set, and re-recognizing and processing a number of medical record homepage images that need to be reprocessed corresponding to the adjustment optimization factor that is less than the adjustment optimization threshold; It should be explained that the first image recognition processing adjustment set can adjust most (such as 97%) of the medical record homepage images that need to be reprocessed to an adjustment optimization factor greater than or equal to the adjustment optimization threshold. However, since the network may fluctuate during the actual execution process, there may be a very small number of medical record homepage images that need to be reprocessed whose processing results fail to meet this standard. In order to meet this challenge, the image processing device has a built-in automatic adjustment mechanism, which automatically triggers the automatic adjustment mechanism to adjust and execute the first image recognition processing adjustment set, and re-identify and process several medical record homepage images that need to be reprocessed whose adjustment optimization factor is less than the adjustment optimization threshold. Through this automatic adjustment process, the image processing device can process these images in a targeted manner until the adjustment optimization factors of all medical record homepage images that need to be reprocessed are increased to a level greater than or equal to the adjustment optimization threshold.

[0093] The above-mentioned re-identification processing of several medical record front page images that need to be reprocessed and whose adjustment optimization factor is less than the adjustment optimization threshold requires first clearing the electronic data corresponding to the several medical record front page images that need to be reprocessed and whose adjustment optimization factor is less than the adjustment optimization threshold, and then re-identification processing and filling.

[0094] The above-mentioned successful feedback of visual adjustment of the front page image of each medical record that needs to be reprocessed refers to sending a success log and visually displaying it on the image processing device page.

[0095] In a specific embodiment, the present invention provides an automated batch recognition and processing system based on medical record homepage data. By acquiring the attribute data of each medical record homepage image belonging to the medical record homepage image set, the upload quality assessment value of each medical record homepage image is analyzed, and the images that do not meet the upload quality assessment threshold requirements are adjusted, thereby ensuring that the medical record homepage image set maintains a high quality during the upload process. By performing automated batch recognition processing on each medical record homepage image, the processing accuracy index of each processed medical record homepage image is analyzed, thereby identifying each medical record homepage image that needs to be reprocessed, and making targeted adjustments to it, thereby further improving the accuracy and reliability of the medical record homepage data recognition processing. Finally, feedback is provided for each medical record homepage image that needs to be reprocessed. The closed-loop processing improves the convenience and efficiency of the automated batch recognition processing of the medical record homepage data, and provides a more convenient and efficient medical record homepage information management solution for medical institutions and the like.

[0096] Reference Figure 2 As shown, the second aspect of the present invention provides a method applied to the automated batch recognition and processing system based on medical record front page data, comprising: S1. receiving a batch uploaded medical record front page image set, obtaining the attribute data of each medical record front page image belonging to the medical record front page image set, and thereby analyzing the upload quality assessment value of each medical record front page image.

[0097] S2. Compare the uploaded quality assessment value of each medical record homepage image with the uploaded quality threshold, and mark several medical record homepage images whose uploaded quality assessment value is less than the uploaded quality threshold as medical record homepage images that need to be optimized, match and execute the image adjustment set of each medical record homepage image that needs to be optimized, thereby re-marking each medical record homepage image that has completed the execution of the image adjustment set as each medical record homepage image, and perform automatic batch recognition processing on each medical record homepage image.

[0098] S3. Obtain the data corresponding to each medical record homepage image that has been identified and processed, mark it as the electronic data corresponding to each processed medical record homepage image, determine the processing accuracy index of each processed medical record homepage image, and compare it with the processing accuracy threshold, mark several processed medical record homepage images whose processing accuracy index is less than the processing accuracy threshold as medical record homepage images that need to be reprocessed, and adjust the processing of each medical record homepage image that needs to be reprocessed.

[0099] S4. Obtain the adjusted electronic data corresponding to each medical record front page image that needs to be reprocessed, evaluate the adjustment optimization factor of each medical record front page image that needs to be reprocessed, and thereby provide feedback for each medical record front page image that needs to be reprocessed.

[0100] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the present invention, they should all fall within the protection scope of the present invention.

Claims

1. An automated batch recognition and processing system based on the medical record front page data, characterized in that: include: The medical record homepage upload module is used to receive a batch of uploaded medical record homepage image sets, obtain the attribute data of each medical record homepage image belonging to the medical record homepage image set, and analyze the upload quality evaluation value of each medical record homepage image; The medical record homepage adjustment module is used to compare the upload quality assessment value of each medical record homepage image with the upload quality threshold, and mark a number of medical record homepage images whose upload quality assessment value is less than the upload quality threshold as medical record homepage images that need to be optimized, match and execute the image adjustment set of each medical record homepage image that needs to be optimized, thereby re-marking each medical record homepage image that has completed the execution of the image adjustment set as each medical record homepage image, and automatically batch-recognize each medical record homepage image; The medical record homepage processing module is used to obtain the data corresponding to each medical record homepage image that has been identified and processed, mark it as the electronic data corresponding to each processed medical record homepage image, determine the processing accuracy index of each processed medical record homepage image, and compare it with the processing accuracy threshold, mark several processed medical record homepage images whose processing accuracy index is less than the processing accuracy threshold as medical record homepage images that need to be reprocessed, and process and adjust each medical record homepage image that needs to be reprocessed; The medical record homepage feedback module is used to obtain the adjusted electronic data corresponding to each medical record homepage image that needs to be reprocessed, evaluate the adjustment optimization factor of each medical record homepage image that needs to be reprocessed, and thus provide feedback on each medical record homepage image that needs to be reprocessed; The specific evaluation process of evaluating the adjustment optimization factors of each medical record front page image that needs to be reprocessed is as follows: Based on processing and adjusting each medical record front page image that needs to be reprocessed, monitoring and acquiring adjusted electronic data corresponding to each medical record front page image that needs to be reprocessed within a second monitoring period; Extracting the vacancy rate of the second field of each medical record front page image that needs to be reprocessed and the second recognition error rate of each medical record front page image that needs to be reprocessed from the adjusted electronic data corresponding to each medical record front page image that needs to be reprocessed; Obtain the vacancy rate of the first field of each medical record front page image that needs to be reprocessed and the first recognition error rate of each medical record front page image that needs to be reprocessed, and at the same time, comprehensively analyze the processing accuracy index of each medical record front page image that needs to be reprocessed to obtain an evaluation method for the adjustment optimization factor of each medical record front page image that needs to be reprocessed; The specific evaluation method for adjusting and optimizing the factors of the medical record front page images that need to be reprocessed is as follows: ;in, is the adjustment optimization factor for the sth medical record front page image that needs to be reprocessed, It is the processing precision threshold preset in the image processing database. is the processing accuracy index of the sth medical record front page image that needs to be reprocessed, is the weight factor corresponding to the processing precision index deviation value preset in the image processing database, is the vacancy rate of the second field of the sth medical record front page image that needs to be reprocessed, is the second recognition error rate of the sth medical record front page image that needs to be reprocessed, is the vacancy rate of the first field of the sth medical record front page image that needs to be reprocessed, is the first recognition error rate of the sth medical record front page image that needs to be reprocessed, e is a natural constant, The impact factor corresponding to the unit value of the second field vacancy rate deviation value preset in the image processing database, is the impact factor corresponding to the second recognition error ratio deviation value unit value preset in the image processing database, s is the number of each medical record front page image that needs to be reprocessed, , w is the total number of medical record front page images that need to be reprocessed.

2. The automated batch recognition and processing system based on the medical record front page data according to claim 1 is characterized by: The upload quality assessment value of each medical record homepage image is analyzed, and the specific analysis process is as follows: Extracting the average brightness of the background area of ​​each medical record homepage image and the average brightness of the text area of ​​each medical record homepage image from the attribute data of each medical record homepage image, and obtaining the image contrast of each medical record homepage image through data processing; Extracting the total number of edge pixels of each medical record homepage image and the total number of image pixels of each medical record homepage image from the attribute data of each medical record homepage image, performing ratio processing on the total number of edge pixels of each medical record homepage image and the total number of image pixels of each medical record homepage image to obtain the edge density of each medical record homepage image; The sampling centroid position points of each medical record homepage image are extracted from the attribute data of each medical record homepage image, and the reference position points of each sampling centroid are obtained. The offset distances of each sampling centroid position points of each medical record homepage image are obtained through distance analysis, and the accumulated distances are added up. The accumulated results are marked as the total offset distances of the sampling centroid position points of each medical record homepage image, thereby comprehensively obtaining the upload quality assessment value of each medical record homepage image.

3. The automated batch recognition and processing system based on the medical record front page data according to claim 1 is characterized by: The matching and execution of the image adjustment set for each medical record front page image to be optimized, the specific matching process is: Obtain the upload quality assessment value of each medical record front page image that needs to be optimized, and sort them in ascending order, extract the first-ranked upload quality assessment value, and match it with the image adjustment set corresponding to each upload quality assessment value interval stored in the image processing database, mark the matched image adjustment set as the image adjustment set of each medical record front page image that needs to be optimized, and thus execute the image adjustment set of each medical record front page image that needs to be optimized.

4. The automated batch recognition and processing system based on medical record front page data according to claim 1 is characterized in that: The automatic batch recognition processing of the first page images of each medical record is specifically to automatically batch recognize the first page images of each medical record through the first image recognition processing set executed by the image processing device.

5. The automated batch recognition and processing system based on medical record front page data according to claim 1 is characterized in that: The specific determination process of determining the processing accuracy index of each processed medical record front page image is as follows: Performing automated batch recognition processing on the front page images of each medical record to monitor and obtain data corresponding to each front page image of each medical record that has been recognized and processed within the first monitoring cycle, and marking them as electronic data corresponding to each processed front page image of the medical record; Extracting the number of first filled fields belonging to each processed medical record homepage image and the number of first fields to be filled belonging to each processed medical record homepage image from the electronic data corresponding to each processed medical record homepage image, and performing data processing on the number of first filled fields belonging to each processed medical record homepage image and the number of first fields to be filled corresponding to each processed medical record homepage image, and marking the processing result as the vacancy rate of the first field belonging to each processed medical record homepage image; Extracting the first processing recognition duration of each processed medical record homepage image from the electronic data corresponding to each processed medical record homepage image, and matching it with the first text filling reference byte number corresponding to each first processing recognition duration interval stored in the image processing database, thereby obtaining the first text filling reference byte number belonging to each processed medical record homepage image, and at the same time extracting the first text filling byte number belonging to each processed medical record homepage image from the electronic data corresponding to each processed medical record homepage image; The number of first sampled recognition word errors belonging to each processed medical record homepage image is extracted from the electronic data corresponding to each processed medical record homepage image, and the total number of first sampled recognition words belonging to each processed medical record homepage image is obtained at the same time. The number of first sampled recognition word errors belonging to each processed medical record homepage image and the total number of first sampled recognition words corresponding to each processed medical record homepage image are ratio processed, and the ratio processing result is marked as the first recognition error ratio of each processed medical record homepage image, thereby comprehensively obtaining a method for determining the processing accuracy index of each processed medical record homepage image.

6. The automated batch recognition and processing system based on the medical record front page data according to claim 1 is characterized by: The above-mentioned adjustment process of processing and adjusting the front page images of each medical record that needs to be reprocessed is as follows: Obtaining the processing accuracy index of each medical record front page image that needs to be reprocessed, and sorting them in ascending order, extracting the processing accuracy index ranked first, and matching them with the first image recognition processing adjustment set corresponding to each processing accuracy index interval stored in the image processing database, thereby marking the obtained first image recognition processing adjustment set as the first image recognition processing adjustment set of each medical record front page image that needs to be reprocessed; Therefore, the image processing device processes and adjusts each medical record front page image that needs to be reprocessed through the first image recognition processing adjustment set of each medical record front page image that needs to be reprocessed.

7. The automated batch recognition and processing system based on the medical record front page data according to claim 1 is characterized by: The feedback on the front page images of each medical record that needs to be reprocessed is analyzed and fed back as follows: The adjustment optimization factor of each medical record homepage image that needs to be reprocessed is compared with the adjustment optimization threshold. If the adjustment optimization factor of a medical record homepage image that needs to be reprocessed is less than the adjustment optimization threshold, the first image recognition processing adjustment set is automatically adjusted and executed until the adjustment optimization factor of each medical record homepage image that needs to be reprocessed is greater than or equal to the adjustment optimization threshold, thereby providing visual adjustment success feedback for each medical record homepage image that needs to be reprocessed.

8. A method applied to the automatic batch recognition and processing system based on medical record front page data as claimed in any one of claims 1 to 7, characterized in that: include: S1. Receive a batch of uploaded medical record homepage image sets, obtain attribute data of each medical record homepage image in the medical record homepage image set, and thereby analyze the upload quality assessment value of each medical record homepage image; S2. Compare the upload quality assessment value of each medical record homepage image with the upload quality threshold, and mark a number of medical record homepage images whose upload quality assessment value is less than the upload quality threshold as medical record homepage images that need to be optimized, match and execute the image adjustment set of each medical record homepage image that needs to be optimized, thereby re-marking each medical record homepage image that has completed the image adjustment set as each medical record homepage image, and perform automatic batch recognition processing on each medical record homepage image; S3. Obtain the data corresponding to each medical record homepage image that has been identified and processed, mark it as the electronic data corresponding to each processed medical record homepage image, determine the processing accuracy index of each processed medical record homepage image, and compare it with the processing accuracy threshold, mark several processed medical record homepage images whose processing accuracy index is less than the processing accuracy threshold as medical record homepage images that need to be reprocessed, and process and adjust each medical record homepage image that needs to be reprocessed; S4. Obtain the adjusted electronic data corresponding to each medical record front page image that needs to be reprocessed, evaluate the adjustment optimization factor of each medical record front page image that needs to be reprocessed, and thereby provide feedback for each medical record front page image that needs to be reprocessed.

Images