Method, apparatus, and product for quality detection of multi-modality medical examination reports

Abstract

This invention discloses a method, apparatus, and product for quality inspection of multimodal medical examination reports. The method includes: acquiring image information of the medical examination report; determining the modality type of the medical examination report based on the image information; determining a target quality control standard set matching the modality type from a quality control standard library; the quality control standard library includes multiple candidate quality control standard sets, with different modality types corresponding to different quality control standard sets, and each candidate quality control standard set including at least an integrity standard determined according to the technical principles and inspection scope corresponding to the modality type; and performing quality inspection on the medical examination report based on the target quality control standard set to determine the quality inspection result. This invention, by calling a matching target quality control standard set from a preset quality control standard library based on the modality type of the medical examination report, and then performing quality inspection on the report, can achieve modality adaptive perception, rule adaptive matching, and quantifiable quality control effects.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, and product for quality inspection of multimodal medical examination reports. Background Technology

[0002] With the rapid development of medical imaging technology, various imaging modalities such as CT, MRI, ultrasound, and X-ray are widely used in clinical diagnosis, resulting in a dramatic increase in the number of medical examination reports. These reports serve as crucial bases for clinical decision-making, and their completeness, standardization, and logical consistency directly impact the quality of diagnosis and treatment and patient safety.

[0003] However, most medical institutions still rely on manual report review, which is not only inefficient but also susceptible to factors such as physician experience and fatigue, making it difficult to guarantee the consistency and comprehensiveness of quality control. For example, report review suffers from the following problems: limited diagnostic physician resources and a heavy review workload, requiring the review of a large number of reports daily, leading to oversights due to fatigue and low efficiency; differences in the professional levels of different physicians result in varying review focuses and standards, making it difficult to achieve homogenized report quality; physicians may make errors in report writing during the diagnostic process, such as selecting the wrong diagnostic template or making text input errors due to factors like input method text associations, examination sites, patient gender, etc., leading to inaccuracies, typos, and even the use of terms like "breast" or "pelvis" in male patient reports, or "chest" in head examination reports. Summary of the Invention

[0004] This invention provides a method, apparatus, and product for quality inspection of multimodal medical examination reports, in order to solve the problems of low inspection efficiency and inconsistent standards for medical examination reports of different modalities.

[0005] According to one aspect of the present invention, a method for quality inspection of multimodal medical examination reports is provided, comprising: Acquire image information from a medical examination report, and determine the modality type of the medical examination report based on the image information; Based on the modality type, a target quality control standard set matching the modality type is determined from the quality control standard library; wherein, the quality control standard library includes multiple candidate quality control standard sets, different modality types correspond to different quality control standard sets, and each candidate quality control standard set includes at least an integrity standard determined according to the technical principles and inspection scope corresponding to the modality type; The medical examination report is subjected to quality testing based on the target quality control standard set, and the quality testing results are determined.

[0006] According to another aspect of the present invention, a quality inspection device for multimodal medical examination reports is provided, comprising: The report modality type determination module is used to acquire image information of medical examination reports and determine the modality type of the medical examination reports based on the image information. The quality control standard set matching module is used to determine a target quality control standard set that matches the modality type from the quality control standard library; wherein, the quality control standard library includes multiple candidate quality control standard sets, different modality types correspond to different quality control standard sets, and each candidate quality control standard set includes at least an integrity standard determined according to the technical principle and inspection scope corresponding to the modality type; The quality inspection module is used to perform quality inspection on the medical examination report according to the target quality control standard set and determine the quality inspection results.

[0007] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the quality inspection method for multimodal medical examination reports according to any embodiment of the present invention.

[0008] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the quality inspection method for multimodal medical examination reports according to any embodiment of the present invention.

[0009] According to another aspect of this application, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the quality inspection method for multimodal medical examination reports as described in any embodiment of this application.

[0010] The technical solution of this invention calls a matching target quality control standard set from a preset quality control standard library based on the modality type of the medical examination report, and then performs quality inspection on the report. This solves the problems of poor modality adaptability, lack of key content evaluation, and low quality control efficiency caused by traditional quality control methods that rely on manual interpretation or use general rules. It achieves quality control effects that can realize modality adaptive perception, rule adaptive matching, and quantifiable quality, and significantly improves the accuracy and automation level of medical examination report review.

[0011] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart of a quality inspection method for multimodal medical examination reports provided by an embodiment of the present invention; Figure 2 This is a flowchart of another method for quality inspection of multimodal medical examination reports provided by an embodiment of the present invention; Figure 3 This is a schematic diagram of a knowledge graph construction process provided by an embodiment of the present invention; Figure 4 This is a flowchart of another method for quality inspection of multimodal medical examination reports provided by an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a quality inspection device for multimodal medical examination reports provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of an electronic device that implements the quality inspection method for multimodal medical examination reports according to embodiments of the present invention. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0015] It should be noted that the terms "candidate," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0016] Figure 1 This invention provides a flowchart of a quality inspection method for multimodal medical examination reports. This embodiment is applicable to situations where unified standard inspections are performed on multimodal medical examination reports. The method can be executed by a multimodal medical examination report quality inspection device, which can be implemented in hardware and / or software and can be configured in a server. Figure 1 As shown, the method includes: S110. Obtain image information from the medical examination report and determine the modality type of the medical examination report based on the image information.

[0017] A medical examination report is a structured or semi-structured written document prepared by a professional physician or technician based on the results of a patient's medical examination (such as imaging, laboratory tests, pathological analysis, electrocardiogram, etc.). It is used to record, interpret, and communicate the patient's physiological and pathological state and typically includes textual descriptions, image thumbnails, patient information, and examination parameters. For example, a complete medical examination report includes basic patient information, examination details, findings, and diagnostic opinions or conclusions.

[0018] Modal type refers to the category of imaging or detection technology used in medical examinations. Different modal types acquire medical images or data using different imaging technologies or physical principles, and reflect different physical or physiological characteristics of human tissues. Examples include CT (computed tomography), MRI (magnetic resonance imaging), X-ray, and ultrasound. Different modal types reflect different tissue characteristics, and their report content and image features have significant differences.

[0019] Specifically, the process begins by acquiring the image information of the medical examination report, performing text recognition on the image information, determining the matching modality type based on the examination items in the medical examination report, or identifying the examination image in the medical examination report, extracting its visual features (such as grayscale distribution, texture pattern, anatomical structure morphology, etc.) through image processing or deep learning models, and determining the matching modality type based on the visual features.

[0020] For example, the main image in a medical examination report can be extracted from the image information. If the main image shows high-contrast skeletal structures, clear bone window / lung window boundaries, and typical X-ray attenuation artifacts, then the modality of the report is determined to be CT modality. Conversely, if the image shows abundant soft tissue layers, no bony highlights, and ventricular high signal characteristics under T2 weighting, then the modality of the report is determined to be MRI modality.

[0021] S120. Determine the target quality control standard set that matches the modality type from the quality control standard library.

[0022] Based on the identified modality type, a target quality control standard set that matches the modality type is dynamically and adaptively selected from a pre-set quality control standard library. The quality control standard library includes multiple candidate quality control standard sets, with different quality control standard sets corresponding to different modality types. Each candidate quality control standard set is dedicated to a specific medical imaging modality type (such as CT, MRI, X-ray, ultrasound, etc.).

[0023] Because different modalities differ significantly in their imaging principles and clinical focus, their corresponding quality control requirements also differ. For example, the quality control standard set for CT modalities emphasizes the completeness of bone structure visualization, density description, and multi-window observation, while the quality control standard set for MRI modalities focuses on sequence type labeling, soft tissue signal characteristics, lesion boundary clarity, and consistency of multiple parameters. Therefore, after identifying a report as belonging to the target modality, the system automatically retrieves a quality control standard set specifically designed for that modality from the quality control standard library. Through a modality-driven adaptive matching mechanism, quality control is ensured to be both accurate and clinically relevant, avoiding misjudgments or missed detections caused by a one-size-fits-all approach to quality control.

[0024] Each candidate quality control standard set includes at least one integrity standard determined based on the technical principles and examination scope corresponding to the modality type. The technical principles determine the type of tissue information that the modality type can present and the types of imaging parameters that must be included. For example, CT, based on X-ray attenuation imaging, can clearly display high-density structures; therefore, its integrity standard requires the report to describe key findings at different window widths and levels, such as bone windows and lung windows. MRI relies on the relaxation properties of hydrogen protons in a magnetic field and requires multiple sequences such as T1, T2, and DWI to reflect tissue state; therefore, its integrity standard requires a clear list of the sequences used and the signal characteristics of each sequence. The examination scope refers to the anatomical regions or target structures that the modality type should cover in a specific clinical setting. For example, the integrity standard for head CT examinations requires coverage of the brain parenchyma, ventricles, skull, and midline structures; while liver MRI examinations require a systematic description of liver lobe segments, bile ducts, portal veins, and lesions. If a report omits key areas or structures that should be assessed within the examination scope, it is considered to have failed the integrity standard.

[0025] In one feasible embodiment, the integrity criterion includes a first element condition for describing key imaging features based on the technical principles corresponding to the modality type, and a second element condition for the pathological structure coverage content determined based on the examination range corresponding to the modality type.

[0026] The first element condition refers to the requirement for describing key imaging features that must be included in the report based on the imaging technology principles upon which the modality type relies. Since different modalities reflect tissue characteristics through different physical mechanisms, their corresponding reports must reflect the unique information dimensions of that modality type. For example, CT, based on X-ray attenuation, can clearly show density differences; therefore, the first element condition for the CT modality type requires the report to describe complete information about the skull under bone windows. MRI, based on tissue relaxation characteristics, requires the report to include information such as whether lesions show high signal intensity on T2-weighted images, whether contrast-enhanced scanning was performed, and the enhancement pattern.

[0027] The second element condition refers to the standard examination scope of this modality in a specific clinical application, i.e., the anatomical regions or target structures that should be covered. This determines the pathological structures or anatomical regions that must be covered in the report. In other words, the second element condition includes the necessary assessment area information contained in the report content. For example, for the head MRI modality, the second element condition requires coverage of structures such as the cerebral hemispheres, basal ganglia, brainstem, cerebellum, and pituitary gland, and descriptions of abnormal signals or space-occupying lesions within them. For the chest CT modality, the report must assess the lung parenchyma, mediastinum, pleura, and ribs, and cannot only describe the lungs while omitting mediastinal lymph nodes. Furthermore, the modality not only reflects the type of imaging technology or physical principle used, but also needs to reflect the examination site information. For example, modalities include chest CT modalities, abdominal CT modalities, and head CT modalities, each with a corresponding set of candidate quality control standards.

[0028] S130. Conduct quality testing on medical examination reports based on the target quality control standard set, and determine the quality testing results.

[0029] After determining the target quality control standard set that matches the report modality type, the text content of the report is automatically analyzed and compared according to the various testing standards (such as the integrity standard) in the quality control standard set, and each item is checked to see if it meets the preset quality requirements. For example, it checks whether key anatomical structures are omitted, whether the terminology is used correctly, and whether the diagnostic conclusions contradict the image description. Finally, based on the test results, a clear quality test result is generated. The quality test result includes results such as pass, partial omission, or logical error. Furthermore, partial omission also corresponds to the specific content of the missing items determined according to the integrity standard.

[0030] In one feasible embodiment, S130 includes: The textual diagnosis content is determined based on the medical examination report, and the textual diagnosis content is identified to obtain the diagnostic object and the corresponding diagnostic description elements; The first matching result is determined based on the degree of matching between the first element condition and the diagnostic description element; The second matching result is determined based on the degree of matching between the second element conditions and the diagnostic subjects; The integrity test result is determined based on the first and second matching results, and is used as the quality test result.

[0031] The textual diagnostic content is located and extracted from medical examination reports. Named entity recognition and semantic parsing technologies are used to identify and decompose the textual diagnostic content, yielding the diagnostic object and corresponding diagnostic descriptive elements. The diagnostic object refers to the specific anatomical structure, organ, tissue, or lesion being diagnosed, representing the subject of the lesion or observation. The diagnostic descriptive elements describe the qualitative or quantitative characteristics of the diagnostic object, including pathological properties, morphological features, size, density / signal, enhancement pattern, and functional state. By decomposing unstructured diagnostic statements into a structured form of "object + descriptive elements," a standardized data foundation can be provided for subsequent quality control, archiving, retrieval, or clinical decision support.

[0032] The diagnostic description elements are compared item by item with the first element conditions preset in the target quality control standard set. The first element conditions are a set of key image feature description items that must be included in the report, defined based on the modality imaging principle. For example, for CT modality, the first element conditions include describing lesion density, indicating whether the boundary is clear, and indicating the maximum diameter; for MRI modality, they include indicating T1 / T2 signal characteristics, whether enhancement is performed, and the enhancement mode. A semantic matching algorithm (such as keyword rule matching or deep learning classifier) ​​is used to determine whether the extracted diagnostic description elements cover all the necessary description items specified in the first element conditions. If they cover all items, the first matching result is considered satisfied; if there are any uncovered items, the first matching result is considered unsatisfactory. Furthermore, the missing specific elements are generated based on the first matching result.

[0033] The diagnostic object is spatially anatomically matched with the second-factor conditions pre-defined in the target quality control standards. The second-factor conditions are a list of anatomical regions or pathological structures that must be evaluated, defined based on the clinical examination scope corresponding to the modality type. For example, the second-factor conditions for head CT include brain parenchyma, ventricular system, skull, midline structures, etc.; for abdominal enhanced CT, they include segments of the liver, spleen, pancreas, both kidneys, retroperitoneum, etc. The identification of the diagnostic object is determined by site standardization mapping to determine whether it completely covers all the structures listed in the second-factor conditions. If all are covered, the second-factor matching result is considered satisfactory; if there are structures that are not mentioned or evaluated, the second-factor matching result is considered unsatisfactory. Furthermore, the missing anatomical regions are generated based on the second-factor matching results.

[0034] Finally, when both the first and second matching results are satisfied, the integrity test result of the medical examination report is deemed qualified, and the quality test result is output as passed; otherwise, the integrity test result of the medical examination report is deemed unqualified, and the quality test result is output as failed, and specific quality control feedback is generated in combination with the missing items.

[0035] This embodiment retrieves the corresponding target quality control standard set from a preset quality control standard library based on the modality type of the medical examination report. Then, based on the first element condition determined by technical principles and the second element condition determined by the examination scope within this target quality control standard set, it performs refined matching with the diagnostic description elements and diagnostic objects extracted from the report. This achieves dual-dimensional automated detection of report integrity, solving problems such as poor modality adaptability and low quality control accuracy caused by using uniform quality control rules in existing technologies. It achieves the beneficial effect of accurately, efficiently, and quantifiably assessing the integrity of medical examination reports of different modalities, significantly improving the pertinence, compliance, and intelligence level of report quality control.

[0036] In one feasible embodiment, the method further includes updating the target quality control standard set based on the quality inspection results of the medical examination report.

[0037] After completing the quality inspection of the medical examination report and obtaining the quality inspection result, the system uses the result as feedback information. If the quality inspection result is passed, the medical examination report is used as the standard report to update the knowledge graph of the target quality control standard set; if the quality inspection result is failed, the medical examination report is corrected, and the knowledge graph of the target quality control standard set is updated based on the corrected medical examination report.

[0038] For example, the effectiveness and comprehensiveness of the current target quality control standard set in practical applications can be evaluated based on the quality inspection results. If the test results repeatedly show that a certain type of problem has not been identified by the existing standards, such as multiple physicians omitting "mediastinal lymph node description" in CT reports but all being judged as qualified, it indicates that there are blind spots in the completeness or standardization of the target quality control standard set. The system can incrementally update the target quality control standard set based on the content of the blind spot, such as adding new mandatory elements, adjusting matching thresholds, or introducing new logical constraint rules.

[0039] This embodiment achieves the transformation of quality control standards from static preset to dynamic evolution by continuously integrating quality control feedback from real reports, thereby continuously improving their clinical relevance, detection sensitivity and adaptability.

[0040] The technical solution of this embodiment calls a matching target quality control standard set from a preset quality control standard library based on the modality type of the medical examination report, and then performs quality inspection on the report. This solves the problems of poor modality adaptability, lack of key content evaluation, and low quality control efficiency caused by traditional quality control methods that rely on manual interpretation or use general rules. It achieves quality control effects that can realize modality adaptive perception, rule adaptive matching, and quantifiable quality, and significantly improves the accuracy and automation level of medical examination report review.

[0041] Figure 2This is a flowchart illustrating another quality inspection method for multimodal medical examination reports provided by an embodiment of the present invention. This embodiment further refines the content of the quality control standard set in the above embodiments. For example... Figure 2 As shown, the method includes: S210. Obtain image information from the medical examination report and determine the modality type of the medical examination report based on the image information.

[0042] S220. Determine the target quality control standard set that matches the modality type from the quality control standard library.

[0043] The quality control standard library includes multiple candidate quality control standard sets. Different modal types correspond to different quality control standard sets. Each candidate quality control standard set includes at least one integrity standard determined according to the technical principles and inspection scope corresponding to the modal type.

[0044] In this embodiment, each candidate quality control standard set also includes a knowledge graph determined according to the standard report of the corresponding modality type. The knowledge graph includes multiple medical entity nodes corresponding to the modality type, as well as edges representing the semantic association between nodes. The medical entity nodes include at least body structure, pathological manifestations, examination parameters and report elements under the modality type.

[0045] Each candidate quality control standard set is associated with a knowledge graph constructed for a specific modality (such as CT, MRI, etc.). This knowledge graph organizes medical knowledge in the form of a structured graph, and displays the relationships between report content under that modality. Nodes in the knowledge graph represent key medical entities under that modality, including at least four categories: body structures, such as the upper lobe of the lung, white matter of the brain, and the S8 segment of the liver; pathological features, such as ground-glass nodules, T2 high-signal lesions, and cortical bone interruption; examination parameters, such as slice thickness 1mm, T1-weighted sequence, and bone window width / level; and report elements, such as nodule size, boundary clarity, and whether enhancement is present. Edges between any two nodes in the knowledge graph represent the semantic relationships between the corresponding node entities, for example, lung nodule—requires description—size and density, T2 high signal—commonly seen in—white matter lesions, bone window—used for observation—cortical bone integrity.

[0046] Specifically, such as Figure 3The diagram illustrates the knowledge graph construction process, which includes: First, collecting a large number of standard reports corresponding to the modality type, reviewed by experts, as the corpus source; second, anonymizing the text of the standard reports and extracting the above four types of entities and their co-occurrence or logical relationships using natural language processing techniques (including Chinese word segmentation, medical terminology recognition, named entity extraction, and relation classification); then, combining radiology guidelines, clinical consensus, and imaging textbooks, further having experts verify and supplement entity definitions and association rules; finally, storing the verified entities and relations in the form of triples (head entity – relation – tail entity) into the graph database, forming a structured knowledge system that supports semantic reasoning and quality control matching.

[0047] S230. Determine the textual diagnostic content based on the medical examination report, and identify the textual diagnostic content to obtain the medical entity content and the corresponding diagnostic description elements.

[0048] The textual diagnostic content in medical examination reports is located and extracted. Natural language processing (NLP) technology is used to semantically analyze this textual diagnostic content, identifying the medical entities and corresponding diagnostic descriptive elements contained within. The medical entities refer to specific objects explicitly identified in the diagnosis, such as anatomical structures, organs, tissues, lesions, or pathological states, like the right upper lobe of the lung, liver segment S6, or adrenal nodules in the brain white matter. Diagnostic descriptive elements are qualitative or quantitative descriptions of the characteristic manifestations of these medical entities on imaging, including morphology, size, density or signal characteristics, boundaries, and enhancement patterns.

[0049] By breaking down unstructured diagnostic statements into structured pairs of medical entities and descriptive elements, a standardized and computable data foundation can be provided for subsequent quality control matching, knowledge graph reasoning, and clinical decision support.

[0050] S240. Match the medical entity content and the corresponding diagnostic description elements with the nodes in the knowledge graph, and based on the semantic association between the nodes in the knowledge graph, determine the third matching result of the medical examination report and the corresponding medical entity node under the modality type, as well as the logical judgment result of the description logic of the diagnostic description elements and the semantic association between the corresponding nodes in the knowledge graph.

[0051] After extracting the medical entity content and its corresponding diagnostic description elements from the report, these elements are compared in a structured manner with the knowledge graph corresponding to the modality type, that is, the medical examination report is compared with the standard report of the corresponding modality type.

[0052] First, the identified medical entity content is semantically matched with medical entity nodes in the knowledge graph to determine whether it belongs to the standard anatomical or pathological categories that should be covered under this modality, and whether it includes all the standard anatomical or pathological categories that should be covered under this modality, thereby generating a third matching result. For example, if the similarity between the medical entity content and all medical entity nodes in the corresponding knowledge graph is greater than a preset similarity threshold, the third matching result is considered to be complete coverage; otherwise, it is considered to be missing key entities. Second, it is determined whether the combination logic between diagnostic description elements and their association with medical entities conforms to the preset semantic association relationships in the knowledge graph. For example, the graph defines lung nodules as requiring description of size, density, and margins, and ground-glass opacity as typically not accompanied by significant enhancement. Through rule engines or graph neural network reasoning, it is determined whether the semantic association relationships represented by the edges of the corresponding nodes in the medical examination report match. Based on the matching results, a logical judgment result is generated, such as whether the logical description is compliant or whether there is an unreasonable combination.

[0053] S250. Determine the standardization test result based on the third matching result and the logical judgment result, and use it as the quality test result.

[0054] The third matching result reflects whether the medical entities mentioned in the medical examination report belong to the standard anatomical structures or pathological categories defined by the knowledge graph under the current modality type. It is used to assess whether the terminology is accurate and whether the object reference conforms to clinical conventions. The logical judgment result is used to assess whether the combination of diagnostic description elements and their association with medical entities conforms to the preset medical semantic rules in the knowledge graph, in order to identify misuse of terminology, contradictory features, or descriptions that do not conform to pathophysiological laws. Only when the medical entity is correctly identified and its description logic is consistent with the professional relationships in the knowledge graph, the system determines that the standardization is qualified; otherwise, if there are non-standard entity naming (such as using non-standard abbreviations), mismatch between description elements and entities (such as describing "bone tissue" as "T2 high signal"), or combinations that violate medical common sense, it is judged as non-standard. This comprehensive judgment result is the standardization test result and serves as an important part of the overall quality test result.

[0055] For example, if the third matching result is complete coverage and the logical judgment result is compliant with the description logic, then the standardization test result is determined to be standard; otherwise, the standardization test result is removed as non-standard. Furthermore, adjustment feedback information is output based on the third matching result and the logical judgment result to prompt the doctor to adjust the medical examination report accordingly, improving the efficiency of medical examination report modification and avoiding doctors receiving vague, low-quality results without knowing the location and method of modification. This embodiment determines the adjustment location and method through adjustment feedback information, improving the accuracy of the feedback results.

[0056] Furthermore, the quality inspection results are determined and the feedback information is adjusted based on the results of the standardization inspection and the integrity inspection.

[0057] The technical solution in this embodiment uses a knowledge graph determined by the standard report to perform quality inspection on medical examination reports, improving the accuracy of the inspection and ensuring that the reports meet the requirements of the standard report. Furthermore, it can determine adjustment feedback information based on the matching results with nodes and edges of the knowledge graph, further improving the efficiency of report modification and avoiding the need to review the entire report during the review process; only the positions corresponding to the adjustment feedback information need to be reviewed, thus improving review efficiency.

[0058] Figure 4 This is a flowchart illustrating another quality inspection method for multimodal medical examination reports provided by an embodiment of the present invention. This embodiment further refines the content of the quality control standard set in the above embodiments. For example... Figure 4 As shown, the method includes: S410. Obtain image information from the medical examination report and determine the modality type of the medical examination report based on the image information.

[0059] S420. Determine the target quality control standard set that matches the modality type from the quality control standard library.

[0060] The quality control standard library includes multiple candidate quality control standard sets. Different modal types correspond to different quality control standard sets. Each candidate quality control standard set includes at least one integrity standard determined according to the technical principles and inspection scope corresponding to the modal type.

[0061] In this embodiment, each candidate quality control standard set also includes a text-image consistency standard and / or a logical consistency standard; the text-image consistency standard is determined based on the degree of matching between the report text and the report image, and the logical consistency standard is determined based on the degree of logical matching between different parts of the report.

[0062] The consistency standard between text and image refers to the quality requirements used to assess whether the textual description in a medical examination report matches the accompanying imaging findings. Its core is to determine whether the location, size, density / signal characteristics of the lesion in the report are verified in the corresponding image slice or region. For example, if the textual report of a medical examination report includes the description of a 10mm solid nodule in the lower lobe of the right lung, then a matching high-density shadow should exist in the corresponding anatomical location in the image. If the textual description does not correspond to the image, it is considered that the text and image are inconsistent.

[0063] Logical consistency standards refer to quality requirements used to detect semantic contradictions or inference conflicts between different textual sections within a medical examination report. For example, determining the logical relationship between examination findings and diagnostic conclusions. For instance, if the examination findings describe homogeneous liver parenchyma without any space-occupying lesions, while the diagnostic opinion includes consideration of hepatocellular carcinoma, this constitutes a logical contradiction. Alternatively, if the descriptions of different body parts in the medical examination report are medically impossible to simultaneously occur, this also constitutes a logical contradiction. For example, according to medical common sense, the size of the first body part is smaller than the second body part, but the size of the first body part in the medical examination report is larger than the second body part, this constitutes a logical contradiction.

[0064] This embodiment ensures that the report accurately reflects the imaging evidence through the consistency standard between images and text, and ensures that the conclusions are reasonably derived through the consistency standard between logic and text. The two can be included in the candidate quality control standard set individually or in combination to comprehensively improve the clinical credibility and professional standardization of medical examination reports.

[0065] S430. Determine the textual diagnostic content based on the medical examination report, identify the textual diagnostic content, and determine the first diagnostic result; determine the image diagnostic content based on the medical examination report, identify the image diagnostic content, and determine the second diagnostic result; based on the image-text consistency standard, determine the image-text consistency detection result based on the matching result of the first diagnostic result and the second diagnostic result.

[0066] The textual diagnostic content of medical examination reports is extracted and structured for analysis to identify key information such as the location, nature, size, density / signal characteristics of lesions, forming a first diagnostic result. Simultaneously, by performing computer vision analysis, deep learning model inference, or combining radiologist annotations on the medical images attached to the medical examination report, the actual abnormal areas and their imaging characteristics in the images are automatically identified to generate a corresponding second diagnostic result. This second diagnostic result is objective diagnostic information based on image evidence.

[0067] Finally, based on the preset consistency criteria for images and text, the first and second diagnostic results are matched and compared semantically and spatially. The consistency test result is determined based on the matching results. The consistency criteria require that the diagnostic conclusion described in the text must be supported by corresponding manifestations in the imaging. For example, it is necessary to determine whether the 8mm ground-glass nodule in the left upper lobe mentioned in the text has a lesion with the same shape, size, and density at the corresponding anatomical location in the imaging. If the two are highly consistent in key elements (such as location, quantity, nature, and extent), the consistency test result is considered passed. If there is no imaging evidence for the textual description, such as the report stating "liver mass" but no abnormalities seen in the imaging, or if the imaging abnormalities are not mentioned in the text, resulting in a missed diagnosis, the consistency test result is considered inconsistent. Further adjustments to the feedback information are made based on the matching results to determine the inconsistency. This test result, as an important component of quality control, effectively prevents clinical risks such as inconsistencies between images and text or missed diagnoses and misreporting, significantly improving the objectivity and reliability of the report.

[0068] S440. Extract the description information of the examination findings and the diagnostic conclusion information from the medical examination report respectively; based on the logical consistency standard, determine the logical consistency test result according to the conflict judgment result between the diagnostic conclusion information and the description information of the examination findings.

[0069] The description of findings is an objective observation record of imaging or test results, such as a 6mm ground-glass nodule with clear borders seen in the upper lobe of the right lung; the diagnostic conclusion information is a clinical judgment or opinion based on the findings, such as conclusive statements such as considering the possibility of early adenocarcinoma or no obvious abnormalities seen.

[0070] First, the description of examination findings and the diagnostic conclusions are extracted from the medical examination report using techniques such as semantic recognition or location positioning. Then, based on a pre-defined logical consistency standard—requiring that the diagnostic conclusions be reasonably derived from the examination findings and that the two must not contradict each other in key elements such as anatomical location, lesion nature, quantity, and extent—a semantic comparison and conflict analysis is performed on both parts. For example, if the examination findings include a description of a space-occupying lesion while the diagnostic conclusions include "no abnormalities," or if the examination findings include multiple metastatic lesions while the diagnostic conclusions mention a single lesion, a logical conflict is identified, and the logical consistency test result is deemed unsuccessful. If the examination findings and the diagnostic conclusions are logically consistent, the logical consistency test result is deemed successful.

[0071] The system uses natural language understanding models or rule engines to identify inconsistencies in the descriptive and diagnostic information as defined by logical consistency standards, and generates a logical consistency test result accordingly. This logical consistency test result, as a core dimension of quality control, effectively prevents internal contradictions in reports caused by typos, copy-paste errors, or reasoning mistakes, ensuring the rigor and clinical credibility of diagnostic opinions.

[0072] S450. Determine the quality inspection result based on the consistency test results of the text and images and / or the logical consistency test results.

[0073] This embodiment includes an implementation scheme that determines the image-text consistency detection result according to step 430 and determines the corresponding quality detection result according to the image-text consistency detection result; an implementation scheme that determines the logical consistency detection result according to step 440 and determines the corresponding quality detection result according to the logical consistency detection result; and an implementation scheme that determines the image-text consistency detection result according to step 430, determines the logical consistency detection result according to step 440, and determines the quality detection result according to the image-text consistency detection result and the logical consistency detection result.

[0074] Specifically, if the image and text consistency test result is passed, the quality test result is determined to be passed; if the image and text consistency test result is inconsistent, the quality test result is determined to be failed; further, the image and text inconsistency adjustment feedback information is determined based on the image and text matching result, and is also fed back to the user as part of the quality test result.

[0075] Alternatively, if the logical consistency test result is passed, the quality test result is determined to be passed; if the logical consistency test result is failed, the quality test result is determined to be failed; further, the logical inconsistency adjustment feedback information is determined based on the logical matching result and is also fed back to the user as part of the quality test result.

[0076] Alternatively, if both the image / text consistency test and the logical consistency test result are passed, the quality test result is determined to be passed; otherwise, the quality test result is determined to be failed. Further adjustments to the feedback information based on the image / text matching result and the logical matching result are also given to the user as part of the quality test result.

[0077] Furthermore, by combining the integrity test results and / or the standardization test results, and based on the text-image consistency test results and / or the logical consistency test results, the quality test results are determined.

[0078] The technical solution of this embodiment improves the accuracy of report quality inspection by refining the quality inspection in multiple dimensions. Furthermore, it refines the adjustment feedback information through the multi-dimensional inspection results. When the report quality inspection fails, the adjustment feedback information determined by the refined dimensions helps to improve the efficiency of report modification and accurately locate the error in the report.

[0079] Figure 5 This is a schematic diagram of a quality inspection device for multimodal medical examination reports provided in an embodiment of the present invention. Figure 5 As shown, the device includes: The report modality type determination module 510 is used to acquire image information of a medical examination report and determine the modality type of the medical examination report based on the image information. The quality control standard set matching module 520 is used to determine a target quality control standard set that matches the modality type from the quality control standard library; wherein, the quality control standard library includes multiple candidate quality control standard sets, different modality types correspond to different quality control standard sets, and each candidate quality control standard set includes at least an integrity standard determined according to the technical principle and inspection scope corresponding to the modality type; The quality inspection module 530 is used to perform quality inspection on the medical examination report according to the target quality control standard set and determine the quality inspection result.

[0080] The technical solution of this embodiment calls a matching target quality control standard set from a preset quality control standard library based on the modality type of the medical examination report, and then performs quality inspection on the report. This solves the problems of poor modality adaptability, lack of key content evaluation, and low quality control efficiency caused by traditional quality control methods that rely on manual interpretation or use general rules. It achieves quality control effects that can realize modality adaptive perception, rule adaptive matching, and quantifiable quality, and significantly improves the accuracy and automation level of medical examination report review.

[0081] Optionally, the integrity criteria include a first element condition for describing key image features based on the technical principles corresponding to the modality type, and a second element condition for the pathological structure coverage content determined based on the examination range corresponding to the modality type.

[0082] Optional, the quality inspection module includes an integrity inspection unit, specifically used for: The textual diagnosis content is determined based on the medical examination report, and the textual diagnosis content is identified to obtain the diagnostic object and the corresponding diagnostic description elements; The first matching result is determined based on the degree of matching between the first element condition and the diagnostic description element; The second matching result is determined based on the degree of matching between the second element condition and the diagnostic object; The integrity detection result is determined based on the first matching result and the second matching result, and is used as the quality detection result.

[0083] Optionally, each candidate quality control standard set also includes a knowledge graph determined according to the standard report of the corresponding modality type. The knowledge graph includes multiple medical entity nodes corresponding to the modality type, and edges representing semantic relationships between nodes. The medical entity nodes include at least body structures, pathological manifestations, examination parameters, and report elements under the modality type.

[0084] Optional, the quality inspection module includes a standardized inspection unit, specifically used for: The textual diagnostic content is determined based on the medical examination report, and the textual diagnostic content is identified to obtain medical entity content and corresponding diagnostic description elements; The medical entity content and the corresponding diagnostic description elements are matched with the nodes in the knowledge graph. Based on the semantic association between the nodes in the knowledge graph, the third matching result of the medical examination report and the corresponding medical entity node under the modality type is determined, as well as the logical judgment result of the semantic association between the description logic of the diagnostic description elements and the corresponding nodes in the knowledge graph. The standardization test result is determined based on the third matching result and the logical judgment result, and is used as the quality test result.

[0085] Optionally, each of the candidate quality control standard sets may also include a text-image consistency standard and / or a logical consistency standard; the text-image consistency standard is determined based on the degree of matching between the report text and the report image, and the logical consistency standard is determined based on the degree of logical matching between different parts of the report.

[0086] Optional, a quality inspection module, including other inspection units, is specifically used for: Based on the medical examination report, determine the textual diagnostic content and identify the textual diagnostic content to determine a first diagnostic result; based on the medical examination report, determine the image diagnostic content and identify the image diagnostic content to determine a second diagnostic result; based on the image-text consistency standard, determine the image-text consistency detection result based on the matching result of the first diagnostic result and the second diagnostic result; and / or Extract the description of examination findings and the diagnostic conclusion information from the medical examination report; based on the logical consistency standard, determine the logical consistency test result according to the conflict judgment result between the diagnostic conclusion information and the description of examination findings. The quality inspection result is determined based on the image-text consistency detection result and / or the logical consistency detection result.

[0087] Optionally, the device also includes a standard set update module for: The target quality control standard set is updated based on the quality inspection results of the medical examination report.

[0088] The quality inspection device for multimodal medical examination reports provided in this embodiment of the invention can execute the quality inspection method for multimodal medical examination reports provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0089] The acquisition, storage, use, and processing of data in this application comply with relevant national laws and regulations and do not violate public order and good morals.

[0090] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0091] Figure 6 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0092] like Figure 6 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0093] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0094] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods described above, such as the quality inspection method for multimodal medical examination reports.

[0095] In some embodiments, the quality inspection method for multimodal medical examination reports may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the quality inspection method for multimodal medical examination reports described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the quality inspection method for multimodal medical examination reports by any other suitable means (e.g., by means of firmware).

[0096] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific reference products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0097] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0098] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0099] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0100] The systems and technologies described herein can be implemented in computing systems that include back-end components (e.g., as data servers), or computing systems that include switching components (e.g., application servers), or computing systems that include front-end components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such back-end, switching, or front-end components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0101] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0102] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication unit 19, or installed from storage unit 18, or installed from ROM 12. When the computer program is executed by processor 11, it performs the functions defined in the methods of the embodiments of the present invention.

[0103] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the quality inspection method for multimodal medical examination reports as provided in any embodiment of this application.

[0104] In implementing the computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0105] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0106] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method of quality detection of a multi-modality medical examination report, characterized in that, The method includes: Acquire image information from a medical examination report, and determine the modality type of the medical examination report based on the image information; Based on the modality type, a target quality control standard set matching the modality type is determined from the quality control standard library; wherein, the quality control standard library includes multiple candidate quality control standard sets, different modality types correspond to different quality control standard sets, and each candidate quality control standard set includes at least an integrity standard determined according to the technical principles and inspection scope corresponding to the modality type; The medical examination report is subjected to quality testing based on the target quality control standard set, and the quality testing results are determined.

2. The method according to claim 1, characterized in that, in, The integrity criteria include a first element condition for describing key image features based on the technical principles corresponding to the modality type, and a second element condition for the pathological structure coverage content determined based on the examination range corresponding to the modality type.

3. The method according to claim 2, characterized in that, The medical examination report is subjected to quality inspection based on the target quality control standard set, and the quality inspection results are determined, including: The textual diagnosis content is determined based on the medical examination report, and the textual diagnosis content is identified to obtain the diagnostic object and the corresponding diagnostic description elements; The first matching result is determined based on the degree of matching between the first element condition and the diagnostic description element; The second matching result is determined based on the degree of matching between the second element condition and the diagnostic object; The integrity detection result is determined based on the first matching result and the second matching result, and is used as the quality detection result.

4. The method according to claim 1, characterized in that, in, Each candidate quality control standard set also includes a knowledge graph determined according to the standard report of the corresponding modality type. The knowledge graph includes multiple medical entity nodes corresponding to the modality type, and edges representing the semantic relationships between nodes. The medical entity nodes include at least body structures, pathological manifestations, examination parameters, and report elements under the modality type.

5. The method according to claim 4, characterized in that, The medical examination report is subjected to quality inspection based on the target quality control standard set, and the quality inspection results are determined, including: The textual diagnostic content is determined based on the medical examination report, and the textual diagnostic content is identified to obtain medical entity content and corresponding diagnostic description elements; The medical entity content and the corresponding diagnostic description elements are matched with the nodes in the knowledge graph. Based on the semantic association between the nodes in the knowledge graph, the third matching result of the medical examination report and the corresponding medical entity node under the modality type is determined, as well as the logical judgment result of the semantic association between the description logic of the diagnostic description elements and the corresponding nodes in the knowledge graph. The standardization test result is determined based on the third matching result and the logical judgment result, and is used as the quality test result.

6. The method according to claim 1, characterized in that, in, Each of the candidate quality control standard sets also includes a text-image consistency standard and / or a logical consistency standard; the text-image consistency standard is determined based on the degree of matching between the report text and the report images, and the logical consistency standard is determined based on the degree of logical matching between different parts of the report.

7. The method according to claim 6, characterized in that, The medical examination report is subjected to quality inspection based on the target quality control standard set, and the quality inspection results are determined, including: Based on the medical examination report, determine the textual diagnostic content and identify the textual diagnostic content to determine a first diagnostic result; based on the medical examination report, determine the image diagnostic content and identify the image diagnostic content to determine a second diagnostic result; based on the image-text consistency standard, determine the image-text consistency detection result based on the matching result of the first diagnostic result and the second diagnostic result; and / or Extract the description of examination findings and the diagnostic conclusion information from the medical examination report; based on the logical consistency standard, determine the logical consistency test result according to the conflict judgment result between the diagnostic conclusion information and the description of examination findings. The quality inspection result is determined based on the image-text consistency detection result and / or the logical consistency detection result.

8. The method according to any one of claims 1-7, characterized in that, The method also includes: The target quality control standard set is updated based on the quality inspection results of the medical examination report.

9. A quality inspection device for multimodal medical examination reports, characterized in that, The device includes: The report modality type determination module is used to acquire image information of medical examination reports and determine the modality type of the medical examination reports based on the image information. The quality control standard set matching module is used to determine a target quality control standard set that matches the modality type from the quality control standard library; wherein, the quality control standard library includes multiple candidate quality control standard sets, different modality types correspond to different quality control standard sets, and each candidate quality control standard set includes at least an integrity standard determined according to the technical principle and inspection scope corresponding to the modality type; The quality inspection module is used to perform quality inspection on the medical examination report according to the target quality control standard set and determine the quality inspection results.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the quality inspection method for multimodal medical examination reports according to any one of claims 1-7.

Images