CN115938529A - Quality evaluation method and device for image report, electronic device and medium - Google Patents
Quality evaluation method and device for image report, electronic device and medium Download PDFInfo
- Publication number
- CN115938529A CN115938529A CN202211599497.4A CN202211599497A CN115938529A CN 115938529 A CN115938529 A CN 115938529A CN 202211599497 A CN202211599497 A CN 202211599497A CN 115938529 A CN115938529 A CN 115938529A
- Authority
- CN
- China
- Prior art keywords
- keyword
- image
- attribute information
- report
- image report
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention discloses a quality evaluation method and device for an image report, electronic equipment and a medium. The quality evaluation method of the image report comprises the following steps: acquiring a medical image and an image report for the medical image; extracting the image report to obtain attribute information corresponding to the keywords; wherein the keywords are used for characterizing abnormal features; and evaluating the quality of the image report according to the attribute information corresponding to the keyword and the matching result of the medical image. According to the method and the device, the keyword for representing the abnormal features and the corresponding attribute information are obtained by extracting the image report, the attribute information of the keyword is matched with the medical image, and finally the quality of the image report is evaluated according to the matching result.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a method and an apparatus for evaluating quality of an image report, an electronic device, and a medium.
Background
In the daily process of a hospital, an imaging doctor needs to write a report for each image examination of a patient to describe normal conditions and abnormal conditions in the image, and for some suspicious lesions, the doctor needs to describe the position, shape, size and other attributes of the lesion according to report specifications and give diagnosis opinions at the same time.
However, it is also possible for the imaging physician to report problems, such as non-normative location of the lesion, measurement problems, reporting internal logic problems, missed diagnosis, and the like. At present, the common way for controlling the report quality in hospitals is to manually perform secondary audit on the report, for example, a doctor with higher annual capital and richer seniority issues the report after auditing the report. The manual secondary report auditing method not only needs to consume more manpower, but also cannot ensure the accuracy of the report.
Disclosure of Invention
The invention provides a quality evaluation method and device for an image report, electronic equipment and a medium, and aims to overcome the defects that in the prior art, manual secondary audit is performed on the report, manpower is consumed, report accuracy cannot be guaranteed and the like.
The invention solves the technical problems through the following technical scheme:
a first aspect of the present invention provides a method for evaluating quality of an image report, including:
acquiring a medical image and an image report for the medical image;
extracting the image report to obtain attribute information corresponding to the keywords; wherein the keywords are used for characterizing abnormal features;
and evaluating the quality of the image report according to the attribute information corresponding to the keyword and the matching result of the medical image.
Optionally, the attribute information corresponding to the keyword includes location information, and the evaluating the quality of the image report according to the matching result between the attribute information corresponding to the keyword and the medical image includes:
determining a target organization corresponding to the keyword according to the position information corresponding to the keyword;
performing segmentation processing on the tissue in the medical image to obtain a first segmentation image;
and evaluating the quality of the image report according to the matching result of the target tissue and the corresponding tissue of the first segmentation image.
Optionally, the attribute information corresponding to the keyword includes size information, and the evaluating the quality of the image report according to the matching result between the attribute information corresponding to the keyword and the medical image includes:
segmenting the medical image according to the target area corresponding to the keyword to obtain a second segmented image;
and evaluating the quality of the image report according to the matching result of the size information corresponding to the keyword and the size information of the target area in the second segmentation image.
Optionally, the extracting the image report to obtain attribute information corresponding to the keyword includes:
inputting the image report into a keyword extraction model to obtain at least one keyword; the keyword extraction model is used for extracting keywords in the image report;
marking the position of a target keyword in the image report, and inputting the marked image report into an attribute information extraction model to obtain attribute information corresponding to the target keyword; the target keyword is any one of the at least one keyword, and the attribute information extraction model is used for extracting attribute information of the target keyword in the image report after the marking processing.
Optionally, the extracting the image report to obtain attribute information corresponding to the keyword includes:
inputting the image report into an attribute information extraction model to obtain at least one attribute information; the attribute information extraction model is used for extracting attribute information in the image report;
marking the position of the attribute information in the image report, and inputting the marked image report into a keyword extraction model to obtain a keyword corresponding to the attribute information; the keyword extraction model is used for extracting keywords corresponding to the attribute information in the image report.
Optionally, the extracting the image report to obtain attribute information corresponding to the keyword includes:
inputting the image report into a feature extraction model to respectively obtain at least one keyword and at least one attribute information;
and matching the keywords with the attribute information to obtain the attribute information corresponding to the keywords.
Optionally, after the image report is extracted to obtain attribute information corresponding to the keyword, the quality assessment method further includes:
and evaluating the quality of the image report according to whether the attribute information corresponding to the keyword accords with the standard expression.
Optionally, the image report includes image description content and diagnosis content, and the extracting process performed on the image report to obtain attribute information corresponding to the keyword includes:
extracting the image description content to obtain attribute information corresponding to a first keyword;
extracting the diagnosis content to obtain attribute information corresponding to a second keyword;
if the first keyword is matched with the second keyword, the quality evaluation method further comprises the following steps: and evaluating the quality of the image report according to the matching result of the attribute information corresponding to the first keyword and the attribute information corresponding to the second keyword.
Optionally, the attribute information corresponding to the keyword includes location information, and after the image report is extracted to obtain the attribute information corresponding to the keyword, the quality assessment method further includes:
determining a target organization corresponding to the keyword according to the position information corresponding to the keyword;
evaluating the quality of the image report according to the matching result of the target tissue and the reference tissue corresponding to the keyword; wherein the reference tissue comprises tissue corresponding to gender information in the image report.
Optionally, the visual report is a structured report or an unstructured report.
A second aspect of the present invention provides an apparatus for evaluating quality of an image report, including:
an acquisition module for acquiring a medical image and an image report for the medical image;
the extraction module is used for extracting the image report to obtain attribute information corresponding to the keywords; the keywords are used for representing abnormal features;
and the evaluation module is used for evaluating the quality of the image report according to the matching result of the attribute information corresponding to the keyword and the medical image.
Optionally, the attribute information corresponding to the keyword includes location information, and the evaluation module includes:
the first determining unit is used for determining a target organization corresponding to the keyword according to the position information corresponding to the keyword;
the first segmentation unit is used for carrying out segmentation processing on the tissues in the medical image to obtain a first segmentation image;
and the first evaluation unit is used for evaluating the quality of the image report according to the matching result of the target tissue and the corresponding tissue of the first segmentation image.
Optionally, the attribute information corresponding to the keyword includes size information, and the evaluation module includes:
the second segmentation unit is used for carrying out segmentation processing on the medical image according to the target area corresponding to the keyword to obtain a second segmentation image;
and the second evaluation unit is used for evaluating the quality of the image report according to the matching result of the size information corresponding to the keyword and the size information of the target area in the second segmentation image.
Optionally, the extraction module is specifically configured to input the image report into a keyword extraction model to obtain at least one keyword; the keyword extraction model is used for extracting keywords in the image report; marking the position of the target keyword in the image report, and inputting the marked image report into an attribute information extraction model to obtain attribute information corresponding to the target keyword; the target keyword is any one of the at least one keyword, and the attribute information extraction model is used for extracting attribute information of the target keyword in the image report after the marking processing.
Optionally, the extracting module is specifically configured to input the image report into an attribute information extracting model to obtain at least one attribute information; the attribute information extraction model is used for extracting attribute information in the image report; marking the position of the attribute information in the image report, and inputting the marked image report into a keyword extraction model to obtain a keyword corresponding to the attribute information; the keyword extraction model is used for extracting keywords corresponding to the attribute information in the image report.
Optionally, the extraction module is specifically configured to input the image report into a feature extraction model, to obtain at least one keyword and at least one attribute information, and perform matching processing on the keyword and the attribute information to obtain attribute information corresponding to the keyword.
Optionally, the evaluation module is further configured to evaluate the quality of the image report according to whether the attribute information corresponding to the keyword meets a specification expression.
Optionally, the image report includes image description content and diagnosis content, and the extraction module includes:
the first extraction unit is used for extracting the image description content to obtain attribute information corresponding to a first keyword;
the second extraction unit is used for extracting the diagnosis content to obtain attribute information corresponding to a second keyword;
the evaluation module is further configured to evaluate the quality of the image report according to a matching result of the attribute information corresponding to the first keyword and the attribute information corresponding to the second keyword under the condition that the first keyword is matched with the second keyword.
Optionally, the attribute information corresponding to the keyword includes location information, and the evaluation module is further configured to determine a target tissue corresponding to the keyword according to the location information corresponding to the keyword, and evaluate the quality of the image report according to a matching result between the target tissue corresponding to the keyword and a reference tissue; wherein the reference tissue comprises tissue corresponding to gender information in the image report.
Optionally, the visual report is a structured report or an unstructured report.
A third aspect of the present invention provides an electronic device, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the method for quality assessment of image reports according to the first aspect.
A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for quality assessment of video reports according to the first aspect.
On the basis of the common general knowledge in the field, the optional conditions can be combined randomly to obtain the preferred embodiments of the invention.
The positive progress effects of the invention are as follows: the image report is extracted to obtain keywords for representing abnormal features and corresponding attribute information of the keywords, the attribute information of the keywords is matched with the medical image, and finally the quality of the image report is evaluated according to a matching result.
Drawings
Fig. 1 is a flowchart of a quality evaluation method for an image report according to embodiment 1 of the present invention.
Fig. 2 is a schematic diagram of an image report according to embodiment 1 of the present invention.
Fig. 3 is a flowchart of another method for evaluating quality of an image report according to embodiment 1 of the present invention.
Fig. 4 is a block diagram of a quality evaluation apparatus for video reports according to embodiment 1 of the present invention.
Fig. 5 is a schematic structural diagram of an electronic device according to embodiment 2 of the present invention.
Detailed Description
The invention is further illustrated by the following examples, which are not intended to limit the invention thereto.
Example 1
Fig. 1 is a flowchart illustrating a quality evaluation method for an image report according to this embodiment, where the quality evaluation method for an image report may be executed by a quality evaluation device for an image report, the quality evaluation device for an image report may be implemented by software and/or hardware, and the quality evaluation device for an image report may be a part or all of an electronic device. The electronic device in this embodiment may be a Personal Computer (PC), such as a desktop, an all-in-one machine, a notebook Computer, a tablet Computer, and the like, and may also be a terminal device such as a mobile phone, a wearable device, and a Personal Digital Assistant (PDA). The following describes the quality evaluation method of the image report provided in this embodiment with an electronic device as an execution subject.
As shown in fig. 1, the method for evaluating the quality of an image report according to this embodiment may include the following steps S11 to S13:
step S11, acquiring a medical image and an image report aiming at the medical image.
The medical image is an internal tissue image obtained in a non-invasive manner with respect to a target object, which may be a human body or some part of a human body. The medical image may be a Computed Tomography (CT) image, an ultrasound image, a nuclear magnetic resonance image, an X-ray image, or the like.
The image report for the medical image generally refers to an image report given by an image doctor according to the medical image, and may also be an image report after the image report given by the image doctor is reviewed by a secondary reviewer, for example, a higher-annual doctor. In implementations, the image report typically includes patient information, such as the patient's name, sex, age, etc., as well as image description and diagnosis. The image description content may also be referred to as an image-viewing or examination-viewing content, and refers to a content describing the medical image. The diagnosis content may also be referred to as a diagnosis opinion or a check conclusion, and refers to a diagnosis conclusion obtained from the medical image. Fig. 2 is a schematic diagram illustrating an image report. In the example shown in fig. 2, the image report includes patient information, image findings, and diagnostic comments.
In a specific implementation, the image report may further include an examination mode such as a CT examination, an ultrasound examination, a magnetic resonance examination, an X-ray examination, and the like, and may further include a partial screenshot of the medical image, and the like.
The video Report may be a Structured Report (SR) or an unstructured Report. The structured report refers to a report generated according to a preset rule or a preset template.
And S12, extracting the image report to obtain attribute information corresponding to the keywords. The keywords are used for characterizing abnormal features, such as density-increasing shadow, texture-free bright area, cyst, nodule, and the like.
In a specific implementation, the attribute information corresponding to the keyword may include shape information, position information, size information, and the like, and may further include binary information, such as presence or absence, and also, for example, see or not see. In a specific example, the "visible speckle-shaped non-texture highlight region at the rear section of the superior pulmonary lobe of the double lungs" in the image report is extracted, the obtained keyword is the non-texture highlight region, the corresponding position information is the rear section of the superior pulmonary lobe of the double lungs, and the corresponding shape information is the speckle shape. In another specific example, the extraction processing is performed on the "left lung superior lobe tip posterior segment does not see the density enhancement image" in the image report, the obtained keyword is the density enhancement image, the corresponding position information is the left lung superior lobe tip posterior segment, and the corresponding binary information is not seen.
And S13, evaluating the quality of the image report according to the matching result of the attribute information corresponding to the keyword and the medical image.
In the specific implementation of step S13, the attribute information corresponding to the keyword is matched with the medical image, and the quality of the image report is evaluated according to the matching result, and may specifically be evaluated as being qualified or unqualified. And when the quality of the image report is evaluated to be unqualified, specific reasons of the unqualified image report can be output, and reminding information can be output to remind corresponding personnel to change the image report. Specifically, after the quality evaluation result is output, the doctor may determine whether the image report needs to be further modified according to the output quality evaluation result, and if it is determined that the image report does not need to be modified, the doctor may submit the image report, and a secondary auditor, for example, a higher-annual-capital doctor, audits the image report, and issues the image report after the audit is correct. If the judgment needs to be modified, the doctor can modify the quality evaluation result according to the unqualified specific reason in the quality evaluation result. After the image report is modified, the quality evaluation method provided by the embodiment may be further used to perform quality evaluation on the modified image report again.
In the embodiment, the image report is extracted to obtain the keywords for representing the abnormal features and the corresponding attribute information thereof, the attribute information of the keywords is matched with the medical image, and finally the quality of the image report is evaluated according to the matching result.
In an optional implementation manner of step S12, the keywords in the video report are extracted first, and then the attribute information corresponding to the keywords is extracted. Specifically, the method may include the following steps S121a to S121b:
step S121a, inputting the image report into a keyword extraction model to obtain at least one keyword.
The keyword extraction model is used to extract keywords in the image report, and may be implemented specifically by using a technique of named entity recognition, for example, the keyword extraction model may be implemented by combining an LSTM (Long short-term-memory) model and a CRF (conditional random field) model, and the keyword extraction model may also be implemented based on a BERT (Bidirectional Encoder Representation from transformations, a pre-trained language Representation model) pre-training model.
And step S121b, marking the position of the target keyword in the image report, and inputting the marked image report into an attribute information extraction model to obtain attribute information corresponding to the target keyword.
The target keyword is any keyword in the at least one keyword, the attribute information extraction model is used for extracting attribute information of the target keyword in the image report after the marking processing, and the attribute information extraction model can be realized in a sequence marking task mode by adopting a BERT pre-training model.
In a specific implementation, the labeling process may add a label symbol, such as "#" or "-", to the position where the target keyword is located. In a specific example, the target keyword is a density increasing image, the marked image is reported as a density increasing image # of the upper lobe tip of the left lung with a strip shape and a spot shape, the boundary is clear, the display range of a mediastinal window is reduced, the image is soft tissue density, a bronchus image containing air can be seen in the soft tissue density, and the enhanced scanning is not abnormally strengthened; inputting the rest abnormal density shadow in the lung into an attribute information extraction model, and obtaining the position information corresponding to the density increasing shadow as follows: the corresponding shape information of the posterior segment of the upper lobe tip of the left lung is as follows: strip-shaped and spot-shaped.
In this embodiment, the image report is first input into the keyword extraction model to extract keywords, and then the image report with the position of the keywords marked is input into the attribute information extraction model to extract attribute information, so as to obtain attribute information corresponding to the keywords.
In another optional implementation manner of step S12, the attribute information in the video report is extracted first, and then the keyword corresponding to the attribute information is extracted. Specifically, the method may include the following steps S122a to S122b:
step S122a, inputting the image report into an attribute information extraction model to obtain at least one attribute information. The attribute information extraction model is used for extracting attribute information in the image report.
And step S122b, marking the position of the attribute information in the image report, and inputting the marked image report into a keyword extraction model to obtain a keyword corresponding to the attribute information. The keyword extraction model is used for extracting keywords corresponding to the attribute information in the image report.
In this embodiment, the attribute information is extracted by inputting the image report into the attribute information extraction model, and then the image report with the position of the attribute information marked is input into the keyword extraction model to extract the keywords, so as to obtain the keywords corresponding to the attribute information.
In another optional implementation manner of step S12, the keywords and the attribute information are extracted, and then the keywords and the attribute information are matched. Specifically, the method may include the following steps S123a to S123b:
step S123a, inputting the image report into a feature extraction model to obtain at least one keyword and at least one attribute information, respectively. In specific implementation, the image report can be input into the same feature extraction model to respectively obtain a keyword and attribute information; the image report can also be respectively input into different feature extraction models to respectively extract the keywords and the attribute information.
And S123b, matching the keywords and the attribute information to obtain the attribute information corresponding to the keywords.
In this embodiment, the image report is input into the feature extraction model to obtain the keywords and the attribute information, and then the keywords and the attribute information are matched to obtain the attribute information corresponding to the keywords. According to the embodiment, the keyword and the attribute information are extracted by using the unified feature extraction model, different models are not required to be trained to extract the keyword and the attribute information respectively, the extraction efficiency of the keyword and the attribute information can be improved, and the quality evaluation efficiency of the image report can be improved.
In an alternative embodiment, the step S13 includes the following steps S131a to S131c:
step S131a, determining a target organization corresponding to the keyword according to the position information corresponding to the keyword.
Step S131b, carrying out segmentation processing on the tissue in the medical image to obtain a first segmentation image.
Step S131c, evaluating the quality of the image report according to the matching result of the target tissue and the tissue corresponding to the first segmentation image.
The tissue may be different parts of the target object, such as chest, lung, brain, liver, head, etc.
In a specific example, the keyword is a density-enhanced image, and the corresponding position information is a posterior segment of a superior lobe of the left lung, so that the target tissue corresponding to the density-enhanced image can be determined to be the lung. The tissue in the medical image is segmented, and if the tissue corresponding to the obtained first segmented image is the liver, the tissue is not matched with the target tissue, and at this time, the quality of the image report can be evaluated as unqualified. If the tissue corresponding to the first segmented image is lung, it matches the target tissue, and the quality of the image report can be evaluated as acceptable.
In this embodiment, the target tissue is a tissue described in the image report, the tissue corresponding to the first segmented image is a tissue present in the medical image, and if the tissue described in the image report does not match the tissue present in the medical image, it indicates that the image report does not correspond to the medical image, that is, the quality of the image report is not good. The quality of the image report is further evaluated from the perspective of whether the tissue described in the image report is matched with the tissue presented in the medical image, so that the comprehensiveness of the quality evaluation is improved, and the accuracy of the quality evaluation is further improved.
In an alternative embodiment, the step S13 includes the following steps S132a to S132b:
step S132a, the medical image is segmented according to the target area corresponding to the keyword, and a second segmentation image is obtained.
Step S132b, evaluating the quality of the image report according to the matching result of the size information corresponding to the keyword and the size information of the target area in the second segmentation image. Wherein, the size information of the target region in the second segmentation image can be obtained by measurement.
In a specific example, the keyword is a high-density image, the corresponding target region is a high-density image region, the size information corresponding to the keyword is 1.6cm × 1.7cm × 2.2cm, the medical image is segmented according to the high-density image region, if the difference between the size information of the high-density image region in the obtained second segmented image and the size information corresponding to the keyword is greater than a preset threshold, it is determined that the two are not matched, and at this time, the quality of the image report may be evaluated as being unqualified. And if the difference value between the size information of the high-density image area in the obtained second segmentation image and the size information corresponding to the keyword is smaller than or equal to a preset threshold value, the two are matched, and the quality of the image report can be evaluated to be qualified. The preset threshold value can be set according to actual conditions.
In this embodiment, the size information corresponding to the keyword is size information of a target area described in the image report, the size information of the target area in the second segmented image is size information of a target area presented in the medical image, and if the size information of the target area described in the image report does not match the size information of the target area presented in the medical image, it is determined that the image report does not correspond to the medical image, that is, the quality of the image report is not good. The quality of the image report is further evaluated from the perspective of whether the target area described in the image report is matched with the target area presented in the medical image, so that the comprehensiveness of the quality evaluation is improved, and the accuracy of the quality evaluation is further improved.
In an alternative embodiment, as shown in fig. 3, after the step S12, the quality evaluation method further includes the following step S2: and evaluating the quality of the image report according to whether the attribute information corresponding to the keyword accords with the standard expression.
In a specific implementation, various canonical expressions of the attribute information may be stored in a database, and whether the attribute information corresponding to the keyword matches with the canonical expression in the database is determined by determining whether the attribute information matches with the canonical expression.
In one specific example, the canonical representation of two lung five lobes includes: the upper left lung lobe, the lower left lung lobe, the upper right lung lobe, the middle right lung lobe and the lower right lung lobe. And if the attribute information corresponding to the keyword comprises the left lung middle lobe, the specification expression of two lungs and five lobes is not met.
In this embodiment, the attribute information corresponding to the keyword is the attribute information described in the video report, and if the attribute information described in the video report does not meet the specification expression, it indicates that the quality of the video report is not qualified.
In an optional embodiment, the step S12 specifically includes: extracting the image description content to obtain attribute information corresponding to a first keyword; and extracting the diagnosis content to obtain attribute information corresponding to the second keyword. In this embodiment, if the first keyword matches the second keyword, after the step S12, as shown in fig. 3, the quality assessment method further includes the following step S3: and evaluating the quality of the image report according to the matching result of the attribute information corresponding to the first keyword and the attribute information corresponding to the second keyword.
The first keyword and the image description content corresponding to the second keyword can be matched to match the first keyword and the second keyword. In a specific implementation, it may be determined whether the first keyword exists in the image description content corresponding to the second keyword, and if yes, the first keyword is considered to be matched with the image description content corresponding to the second keyword, that is, the first keyword is matched with the second keyword; if the first keyword does not exist, the first keyword is considered to be not matched with the image description content corresponding to the second keyword, namely the first keyword is not matched with the second keyword. The similarity between the first keyword and the image description content corresponding to the second keyword can be calculated, and when a plurality of image description contents corresponding to the second keyword exist, the similarity between the first keyword and each image description content is calculated respectively. If the similarity greater than a preset threshold value, for example, 90%, the first keyword and the second keyword are considered to be matched; if the similarity degree larger than a preset threshold value, for example, 90%, does not exist, the first keyword and the second keyword are considered not to match. The similarity may be calculated by using a pre-training model, or may be calculated by calculating a distance between word embedding vectors, and the method for calculating the similarity is not limited in this embodiment.
In one specific example, the first keyword extracted from the image description content is a density-enhanced image, the second keyword extracted from the diagnosis content is inflammation, and the density-enhanced image is a representation of inflammation, so that the first keyword and the second keyword can be determined to be matched. Further, the attribute information of the density-increasing image and the attribute information of the inflammation are matched, and the quality of the image report is evaluated according to the matching result of the density-increasing image and the inflammation.
It should be noted that, in the process of matching the attribute information corresponding to the first keyword with the attribute information corresponding to the second keyword, it is necessary to match the attribute information of the same type, for example, match the shape information corresponding to the first keyword with the shape information corresponding to the second keyword, or match the position information corresponding to the first keyword with the position information corresponding to the second keyword.
In this embodiment, the first keyword is an abnormal feature described in the image description content, the second keyword is an abnormal feature described in the diagnosis content, and when the two keywords are matched with each other, the quality of the image report is evaluated according to the matching result of the attribute information of the two keywords, and if the attribute information of the two keywords is not matched with each other, it is indicated that there is a problem of logic error in the image report. The quality of the image report is further evaluated from the perspective of whether the image description content and the diagnosis content are matched, so that the comprehensiveness of quality evaluation is improved, and the accuracy of quality evaluation is further improved.
In an alternative embodiment, as shown in fig. 3, after the step S12, the quality evaluation method further includes the following step S4: and determining a target organization corresponding to the keyword according to the position information corresponding to the keyword, and evaluating the quality of the image report according to the matching result of the target organization corresponding to the keyword and a reference organization. Wherein the reference tissue comprises tissue corresponding to gender information in the image report.
In one specific example, the gender information in the image report is female, and the reference tissue includes ovary, uterus, fallopian tube, etc. The keywords are density-increasing shadows, the corresponding position information is bilateral prostate, and the target tissue corresponding to the keywords can be determined to be prostate. The target tissue does not match the reference tissue, indicating that the tissue and gender information described in the image report do not correspond, i.e., the quality of the image report is not acceptable.
In this embodiment, the target tissue is a tissue described in the image report, the reference tissue includes a tissue corresponding to the sex information in the image report, and if the target tissue does not match the reference tissue, it indicates that there is a problem in the image report that there is a description error or a sex information error. The quality of the image report is further evaluated from the perspective of whether the tissue described in the image report is matched with the tissue corresponding to the gender information, so that the comprehensiveness of the quality evaluation is improved, and the accuracy of the quality evaluation is further improved.
The present embodiment further provides a quality evaluation apparatus 40 for image report, as shown in fig. 4, which includes an obtaining module 41, an extracting module 42, and an evaluating module 43. The acquisition module 41 is used for acquiring a medical image and an image report for the medical image. The extraction module 42 is configured to extract the image report to obtain attribute information corresponding to the keyword; wherein the keywords are used for characterizing abnormal features. The evaluation module 43 is configured to evaluate the quality of the image report according to the matching result between the attribute information corresponding to the keyword and the medical image.
In an optional implementation manner, the attribute information corresponding to the keyword includes location information, and the evaluation module includes: the first determining unit is used for determining a target organization corresponding to the keyword according to the position information corresponding to the keyword; the first segmentation unit is used for carrying out segmentation processing on the tissues in the medical image to obtain a first segmentation image; and the first evaluation unit is used for evaluating the quality of the image report according to the matching result of the target tissue and the tissue corresponding to the first segmentation image.
In an optional implementation manner, the attribute information corresponding to the keyword includes size information, and the evaluation module includes: the second segmentation unit is used for carrying out segmentation processing on the medical image according to the target area corresponding to the keyword to obtain a second segmentation image; and the second evaluation unit is used for evaluating the quality of the image report according to the matching result of the size information corresponding to the keyword and the size information of the target area in the second segmentation image.
In an optional implementation manner, the extraction module is specifically configured to input the image report into a keyword extraction model to obtain at least one keyword; the keyword extraction model is used for extracting keywords in the image report; marking the position of the target keyword in the image report, and inputting the marked image report into an attribute information extraction model to obtain attribute information corresponding to the target keyword; the target keyword is any keyword in the at least one keyword, and the attribute information extraction model is used for extracting attribute information of the target keyword in the marked image report.
In an optional implementation manner, the extraction module is specifically configured to input the image report into an attribute information extraction model to obtain at least one attribute information; the attribute information extraction model is used for extracting attribute information in the image report; marking the position of the attribute information in the image report, and inputting the marked image report into a keyword extraction model to obtain a keyword corresponding to the attribute information; the keyword extraction model is used for extracting keywords corresponding to the attribute information in the image report.
In an optional implementation manner, the extraction module is specifically configured to input the image report into a feature extraction model, respectively obtain at least one keyword and at least one attribute information, and perform matching processing on the keyword and the attribute information to obtain attribute information corresponding to the keyword.
In an optional implementation manner, the evaluation module is further configured to evaluate the quality of the image report according to whether the attribute information corresponding to the keyword meets a specification expression.
In an alternative embodiment, the image report includes image description content and diagnosis content, and the extraction module includes: the first extraction unit is used for extracting the image description content to obtain attribute information corresponding to a first keyword; the second extraction unit is used for extracting the diagnosis content to obtain attribute information corresponding to a second keyword; the evaluation module is further configured to evaluate the quality of the image report according to a matching result of the attribute information corresponding to the first keyword and the attribute information corresponding to the second keyword under the condition that the first keyword is matched with the second keyword.
In an optional implementation manner, the attribute information corresponding to the keyword includes location information, and the evaluation module is further configured to determine a target organization corresponding to the keyword according to the location information corresponding to the keyword, and evaluate the quality of the image report according to a matching result between the target organization corresponding to the keyword and a reference organization; wherein the reference tissue comprises tissue corresponding to gender information in the image report.
In an alternative embodiment, the visual report is a structured report or an unstructured report.
It should be noted that the quality evaluation device of the image report in this embodiment may be a separate chip, a chip module, or an electronic device, or may be a chip or a chip module integrated in an electronic device.
The modules/units included in the quality evaluation apparatus for video reports described in this embodiment may be software modules/units, or may also be hardware modules/units, or may also be part of software modules/units and part of hardware modules/units.
Example 2
Fig. 5 is a schematic structural diagram of an electronic device provided in this embodiment. The electronic device includes at least one processor and a memory communicatively coupled to the at least one processor. Wherein the memory stores a computer program executable by the at least one processor, the computer program being executable by the at least one processor to enable the at least one processor to perform the method of quality assessment of image reports of embodiment 1. The electronic device provided by this embodiment may be a personal computer, such as a desktop, an all-in-one machine, a notebook computer, a tablet computer, and the like, and may also be a mobile phone, a wearable device, a palmtop computer, and other terminal devices. The electronic device 3 shown in fig. 5 is only an example, and should not bring any limitation to the functions and the use range of the embodiment of the present invention.
The components of the electronic device 3 may include, but are not limited to: the at least one processor 4, the at least one memory 5, and a bus 6 connecting various system components including the memory 5 and the processor 4.
The bus 6 includes a data bus, an address bus, and a control bus.
The memory 5 may include volatile memory, such as Random Access Memory (RAM) 51 and/or cache memory 52, and may further include Read Only Memory (ROM) 53.
The memory 5 may also include a program/utility 55 having a set (at least one) of program modules 54, such program modules 54 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.
The processor 4 executes various functional applications and data processing, such as the above-described quality evaluation method of image reports, by running a computer program stored in the memory 5.
The electronic device 3 may also communicate with one or more external devices 7, such as a keyboard, pointing device, etc. Such communication may be via an input/output (I/O) interface 8. Also, the electronic device 3 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the internet) via the network adapter 9. As shown in fig. 5, the network adapter 9 communicates with other modules of the electronic device 3 via the bus 6. It should be appreciated that although not shown in FIG. 5, other hardware and/or software modules may be used in conjunction with the electronic device 3, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems, to name a few.
It should be noted that although in the above detailed description several units/modules or sub-units/modules of the electronic device are mentioned, such a division is merely exemplary and not mandatory. Indeed, the features and functions of two or more of the units/modules described above may be embodied in one unit/module according to embodiments of the invention. Conversely, the features and functions of one unit/module described above may be further divided into embodiments by a plurality of units/modules.
Example 3
The present embodiment provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the quality evaluation method of the video report of embodiment 1.
More specific examples, among others, that the readable storage medium may employ may include, but are not limited to: a portable disk, hard disk, random access memory, read only memory, erasable programmable read only memory, optical storage device, magnetic storage device, or any suitable combination of the foregoing.
In a possible implementation, the invention may also be implemented in the form of a program product comprising program code for causing an electronic device to perform a method for quality assessment of an image report implementing embodiment 1, when said program product is run on said electronic device.
Where program code for carrying out the invention is written in any combination of one or more programming languages, the program code may be executed entirely on the electronic device, partly on the electronic device, as a stand-alone software package, partly on the electronic device, partly on a remote device or entirely on the remote device.
While specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that this is by way of example only, and that the scope of the invention is defined by the appended claims. Various changes or modifications to these embodiments may be made by those skilled in the art without departing from the principle and spirit of this invention, and these changes and modifications are within the scope of this invention.
Claims (10)
1. A method for evaluating quality of an image report, comprising:
acquiring a medical image and an image report for the medical image;
extracting the image report to obtain attribute information corresponding to the keywords; wherein the keywords are used for characterizing abnormal features;
and evaluating the quality of the image report according to the matching result of the attribute information corresponding to the keyword and the medical image.
2. The quality evaluation method according to claim 1, wherein the attribute information corresponding to the keyword includes position information, and the evaluating the quality of the image report according to the matching result of the attribute information corresponding to the keyword and the medical image includes:
determining a target organization corresponding to the keyword according to the position information corresponding to the keyword;
performing segmentation processing on the tissue in the medical image to obtain a first segmentation image;
and evaluating the quality of the image report according to the matching result of the target tissue and the corresponding tissue of the first segmentation image.
3. The quality evaluation method according to claim 1, wherein the attribute information corresponding to the keyword includes size information, and the evaluating the quality of the image report according to the matching result of the attribute information corresponding to the keyword and the medical image includes:
segmenting the medical image according to the target area corresponding to the keyword to obtain a second segmentation image;
and evaluating the quality of the image report according to the matching result of the size information corresponding to the keyword and the size information of the target area in the second segmentation image.
4. The quality evaluation method according to claim 1, wherein the extracting the image report to obtain the attribute information corresponding to the keyword comprises:
inputting the image report into a keyword extraction model to obtain at least one keyword; the keyword extraction model is used for extracting keywords in the image report;
marking the position of a target keyword in the image report, and inputting the marked image report into an attribute information extraction model to obtain attribute information corresponding to the target keyword; the target keyword is any keyword in the at least one keyword, and the attribute information extraction model is used for extracting attribute information of the target keyword in the marked image report.
5. The quality assessment method according to claim 1, wherein after extracting the image report to obtain the attribute information corresponding to the keyword, the quality assessment method further comprises:
and evaluating the quality of the image report according to whether the attribute information corresponding to the keyword accords with the standard expression.
6. The quality evaluation method according to claim 1, wherein the image report includes image description content and diagnosis content, and the extracting the image report to obtain attribute information corresponding to the keyword includes:
extracting the image description content to obtain attribute information corresponding to a first keyword;
extracting the diagnosis content to obtain attribute information corresponding to a second keyword;
if the first keyword is matched with the second keyword, the quality evaluation method further comprises the following steps: and evaluating the quality of the image report according to the matching result of the attribute information corresponding to the first keyword and the attribute information corresponding to the second keyword.
7. The quality assessment method according to any one of claims 1 to 6, wherein the attribute information corresponding to the keyword includes position information, and after extracting the image report to obtain the attribute information corresponding to the keyword, the quality assessment method further comprises:
determining a target organization corresponding to the keyword according to the position information corresponding to the keyword;
evaluating the quality of the image report according to the matching result of the target tissue and the reference tissue corresponding to the keyword; wherein the reference tissue comprises tissue corresponding to gender information in the image report.
8. An apparatus for evaluating quality of an image report, comprising:
an acquisition module for acquiring a medical image and an image report for the medical image;
the extraction module is used for extracting the image report to obtain attribute information corresponding to the keywords; wherein the keywords are used for characterizing abnormal features;
and the evaluation module is used for evaluating the quality of the image report according to the matching result of the attribute information corresponding to the keyword and the medical image.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method for quality assessment of an image report according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out a method for quality assessment of an image report according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211599497.4A CN115938529A (en) | 2022-12-12 | 2022-12-12 | Quality evaluation method and device for image report, electronic device and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211599497.4A CN115938529A (en) | 2022-12-12 | 2022-12-12 | Quality evaluation method and device for image report, electronic device and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115938529A true CN115938529A (en) | 2023-04-07 |
Family
ID=86551922
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211599497.4A Pending CN115938529A (en) | 2022-12-12 | 2022-12-12 | Quality evaluation method and device for image report, electronic device and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115938529A (en) |
-
2022
- 2022-12-12 CN CN202211599497.4A patent/CN115938529A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5952835B2 (en) | Imaging protocol updates and / or recommenders | |
| US8612261B1 (en) | Automated learning for medical data processing system | |
| RU2686627C1 (en) | Automatic development of a longitudinal indicator-oriented area for viewing patient's parameters | |
| CN112712879B (en) | Information extraction method, device, equipment and storage medium for medical image report | |
| US20220068449A1 (en) | Integrated diagnostics systems and methods | |
| JP2017509946A (en) | Context-dependent medical data entry system | |
| EP3654343A1 (en) | Application of deep learning for medical imaging evaluation | |
| US11615890B2 (en) | Method and system for the computer-assisted implementation of radiology recommendations | |
| CN111986182A (en) | Auxiliary diagnosis method, system, electronic device and storage medium | |
| CN110706815A (en) | Evaluation method and device of image report and electronic equipment | |
| CN111627512A (en) | Recommendation method and device for similar medical records, electronic equipment and storage medium | |
| Beddiar et al. | Automatic captioning for medical imaging (MIC): a rapid review of literature | |
| US20210313045A1 (en) | Automatic Bounding Region Annotation for Localization of Abnormalities | |
| Dalal et al. | Determining follow-up imaging study using radiology reports | |
| AU2022246663A1 (en) | Systems and methods for artificial intelligence-assisted image analysis | |
| US20220148166A1 (en) | Systems and methods for assessing a likelihood of cteph and identifying characteristics indicative thereof | |
| Zhang et al. | Comparison of chest radiograph captions based on natural language processing vs completed by radiologists | |
| CN117594183A (en) | Radiological report generation method based on inverse fact data enhancement | |
| Gaggion et al. | Chexmask: a large-scale dataset of anatomical segmentation masks for multi-center chest x-ray images | |
| Wang et al. | Toward standardized premarket evaluation of computer aided diagnosis/detection products: insights from FDA-approved products | |
| US20230035575A1 (en) | Analysis device, analysis method, and recording medium | |
| CN115938529A (en) | Quality evaluation method and device for image report, electronic device and medium | |
| US20230118299A1 (en) | Radiologist fingerprinting | |
| CN115206478A (en) | Medical report generation method and device, electronic equipment and readable storage medium | |
| CN111275558B (en) | Method and device for determining insurance data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |