JP6421160B2 - Method and system for generating medical reports - Google Patents

Description

  In medical imaging, the interpreters of the examination are free text reports regarding the presence or absence of the diagnostic imaging findings that they interpret and their interpretation of the diagnostic significance of such findings. Is generated. A great deal of effort is expended by individual test interpreters who produce reports containing commonly repeated phrases, structures, and concepts. The brain work and time involved in generating such reports is a major factor limiting productivity.

  To date, this problem has been addressed in a limited manner through the use of pre-made phrases and reporting templates. Pre-made phrases are lists of commonly used expressions, and report writers can insert relevant phrases into the report text, stitch such default expressions together with additional free-form The list of representations can be utilized when generating a report to form a final report. A template is a framework that is created in advance for a complete report on the findings commonly found, with a portion of the information left blank, and the report writer fills in the blank when preparing the final report.

  Generating a report through the use of a default expression involves the brain work involved in selecting an appropriate expression for reporting, summarizing the expression, and eliminating or correcting conflicting expressions. Since the writer's work remains entirely, it remains painstaking. In fact, the report writer will have to spend additional time to ensure that pre-made phrases inserted into the report do not contain any conflicting features.

  Another limitation of the pre-written phrase system is that only commonly used phrases that would be related to the “critical mass” of the user of the pre-written phrase system are the writer of the report. Is available to you. Such a “copy-and-paste” pre-written expression system does not adequately address the user's need to write reports about unusual situations that will be seen repeatedly, and the unusual situations are pre-set. With a frequency that is not enough to include the appropriate phrase in the standard phrase glossary. Placing such rarely used too many phrases in a standard distributed set of phrases would make the phrase set large and unwieldy and difficult to manipulate .

  Once the template is used, the report writer needs to put considerable effort in modifying the template to fit the unique circumstances of the reported case. A particular limitation of template-based systems used in existing report generation techniques is that such systems are highly structured and strictly defined. Such a system consists of strictly formed blocks of text with blank areas for the writer to fill by speech recognition, typing, or other means. This technique does not provide sufficient flexibility to be used in reporting cases that deviate from a limited set of predetermined patterns.

  Thus, there is a need for report writers to use “smart” report generation methods for generating reports in a more time efficient manner. There is also a need to allow report writers to have mechanisms that reduce the likelihood of errors or oversights in reporting.

  Embodiments of the present invention are a consistent and time efficient way for radiologists and other medical imaging interpreters to report both normal and abnormal medical findings using frequently repeated phrases. A good way can be provided. Inventors uniquely realize that phrases commonly used in radiology reports do not exist in isolation, but rather in predictable, interrelated patterns and create new report generation tools So use this phrase. Still further, by recognizing that there are predictable and interrelated patterns, the inventors' system deviates from the traditional view that report generation could not be automated to this level. . The inventor will not be able to determine the decisions made by the previous user regarding the insertion or deletion of phrases from the report, as well as the phrases extracted from the report text generated by the previous user, for future report writers. It was further noticed that the data can be used to generate proposals on the report content.

  The report generation method is a relationship between medical phrases commonly used to assist in manipulating the available phrases, initially such relationships that are clearly defined by the user (s) By using additional phrases that are subsequently extracted by the system through data gathered through previous user experiences on useful phrases, the user can use these useful methods in a more time-efficient manner. , Allowing reports to be generated using interrelated medical phrases.

  Such a system would also address the need for mechanisms that reduce the likelihood of reporting errors, such as failure to report vital information in a given medical context.

  The present invention departs from existing report generation tools that do not take advantage of the existence of relationships between phrases, and the existing report generation tool also captures data collected during the use of the system by previous report writers. Not utilized and therefore cannot be automated to the level of the present invention.

  The present invention seeks to facilitate the generation of medical imaging report by utilizing repeated theme patterns and networked ideas to increase the speed of report generation. These patterns exist in the logic of medical reports and can be used to speed up and simplify the process of generating medical reports. These patterns may be, for example, a predetermined commonly represented medical idea that is associated with or suggests other commonly represented medical ideas, or excludes other commonly represented medical ideas. Alternatively, handling a given subject in such a report may include the fact that it often suggests a recurring list of additional subjects that need to be handled. Such inclusion patterns, also known as inclusion rules, can be applied to phrases selected by the report writer to provide suggestions regarding additional phrases that the writer may wish to insert into the report while in progress. Applied. Similarly, exclusion patterns, also known as exclusion rules, are applied to selected phrases and related phrases to eliminate mutually conflicting phrases to ensure consistent reporting findings. The

  The first phrase used in the system is clearly defined by the user, and the relationship between phrases (included or excluded rules, or phrases related to each other through the “potential alternative” rules described below) is First, clearly defined by the user. For example, a user may define several phrases, and then when a given phrase is inserted into a report, certain other phrases are potentially considered for insertion into the report. Indicates that it should be flagged as useful. Alternatively, the user may specify that a given phrase is inconsistent and that when one phrase is inserted into a report, another given phrase (if present) should be removed. Good. These relationships involve “proposed additional” phrases and “excluded” phrases.

  Additional types of relationships that may be specified include “potential alternative” phrases. If a given phrase is excluded from the report due to a conflict with another phrase, if the potential substitute phrase remains consistent with the other phrases in the report, The “of” phrase may be automatically inserted at the position of the excluded phrase.

  The inventor has entered such a reporting system as it is used over time by multiple users using the system to create a complete medical image report. Information about the insertion and deletion of specific phrases performed by the previous user in the course of writing the report as well as the text of the previous report is a valuable set of information about potentially useful phrases and future users Have found that they provide their relationship with each other that may be used to suggest to future users when writing reports. This is due to the fact that the diagnostic imaging associated with the report is made with respect to specific body parts and specific diagnostic imaging methods within the scope of predictable pathological or normal findings. Such reports tend to be repetitive and contain a large number of frequently reused elements. Therefore, the inventor's report generation system not only allows the user to define useful phrases and the relationships between those phrases, but also other relevance to the current writer's report in progress. We will also actively pursue the definition and suggestion of more useful phrases by investigating the patterns of system usage by the report writers.

  The system reports previously written potentially useful text, for example, by the presence of a rare word in the previously generated text that matches a rare word in the current writer's report in progress. You may identify from. If the writer selects the text in question and inserts the text into the report as it progresses, the system is informed that the text or “phrase” is indeed potentially useful. The system will then not only take various features relating to the phrase, but also other actions taken by the user in the same report in which the phrase was inserted, for example (will become a “suggested additional” phrase). Stores information such as the insertion of other phrases and the removal of other phrases (which would get an “excluded” relationship with the first inserted phrase). The “phrase” generated by such an automatic method is used repeatedly, and the usage pattern for such a phrase is to insert the phrase in question together with the other phrases in question. The system will make such phrases and other phrases that would be used to improve the system's ability to provide users with suggestions that are useful when writing reports. Develop a database of information about relationships and relationships.

  In one implementation of the invention, the user defines a set of phrases that are useful for the particular anatomical site of interest, and the user limits the complete set of expected phrases, It is possible to propose only the most relevant phrases for image inspection. Based on the selection of exam type data based on the anatomical site, the associated phrase or template is presented to the report writer for review into the text field.

  The proposed set of phrases may also be influenced by other medical imaging studies related to “metadata”, which is almost universally used in digital images and communications in medicine ( Stored with the actual image in most medical image data formats such as DICOM format. Such metadata is information about the type of diagnostic imaging performed, as well as information about which anatomical site was imaged, as well as other potentially relevant such as patient age and gender. Information to be included.

  Another embodiment of the present invention is a system for generating a medical imaging test report for inputting medical imaging data and test type user selection, phrase selection and other graphics or text. Generating one or more input devices configured to: one or more processors configured to process medical image test data, test type selection, and phrase selection; and generating a medical image test report One or more memories coupled to one or more processors configured to store a medical report file including one or more templates, test type data, phrase data, and a plurality of text fields Medical imaging to influence the system and which phrases are presented to the user as most potentially useful One or more processors configured to process data, and one configured to store examination type data and associated phrase data selected by the user and coupled to the one or more processors Or a plurality of memory systems, one or more processors configured to process one or more inclusion rules associated with the phrase, and associated with the phrase to eliminate inconsistent phrases from presentation to the user The present invention relates to a system comprising one or more processors configured to process an exclusion rule and an output device configured to display a generated medical report and connected to the processor.

FIG. 6 is a diagram of a computer system according to an exemplary embodiment of the present invention. 3 is a flowchart of a method according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. FIG. 6 illustrates a computer interface at various stages for generating a medical image data report according to an exemplary embodiment of the present invention. 4 is a screen shot of a system interface according to an exemplary embodiment of the present invention. 4 is a screen shot of a system interface according to an exemplary embodiment of the present invention.

  The following is a description of a method and system for creating and generating a medical imaging report according to an exemplary embodiment of the present invention.

  The methods and systems described herein allow a user to quickly organize and generate a medical imaging report. The elements of the system are the types of tests that correspond to the anatomical site being examined, and may limit the phrases presented for review, and the types of tests that may affect those phrases, A reporting framework consisting of a set of text fields corresponding to commonly used components, in which the reports are generally grouped in their text fields (eg, “findings” and “feelings” columns), A reporting framework in which the actual reporting text will be placed in the field, a set of “phrases”, or a textual representation that would be useful for insertion into the report, and such phrases (eg, given Option to select a "template" as well as relationships defined between them (such as certain phrases that suggest inclusion or exclusion of other phrases) No.

  The term “template” is used in conventional work to indicate a relatively complete report that has been generated in advance, and the relatively complete report is a blank that is filled in by the report writer to generate a complete report. Contains fields. The present invention provides a means of initiating the process of modifying a report with a relatively complete report, and therefore the term “template” is used again in the current context to describe similar ideas. . However, in the case of the present invention, such a “template” is simply a strict text frame that must be manually modified in detail to fit the unique case details for the report to which the template is applied. Rather than include work, rather, in the present invention, a “template” consists of a set of phrases that are used sequentially to generate a flow of report text. The decisive difference between the template of the implementation of the present invention and the “template” used in the conventional work is that each phrase in the report retains its unique identity as a phrase and only in the report. It is to maintain the relationship with other phrases both in the phrase database of the present invention. This is because the template is customized into the final report for a particular case, so the processing of the report text can be done, for example, by inserting additional phrases into the reporting framework that was initially occupied by text by the template. , Meaning being influenced by the relationship between the template phrase and other phrases. For example, the newly inserted phrase may conflict with the phrase used in the template. That is, when such a phrase is inserted, phrases that contradict that phrase may be automatically removed from the report, or may be flagged for review. Similarly, phrase patterns proposed for consideration for inclusion in the report are included in the report, whether due to inclusion in the original template or due to later insertion by the user. The entire set of phrases will be similarly affected.

  Phrases can include not only redundant or long forms for insertion into the standard “findings” column of image reports, but also short forms for inclusion within the standard “feelings” column of image reports, The “Findings” column contains a detailed description of the image findings and a full discussion of their significance, and the “Findings” column gives a concise summary of the major conclusions of the diagnostic interpreter.

  The system knowledge database contains not only phrases but also information about rules or relationships between phrases (see exemplary example 1 below). The system selects phrases from the knowledge database for suggestions to the user based on factors including patterns that define the relationships between phrases, and the user enters such phrases to enter into the report in progress May be selected. Upon selecting a phrase to be inserted into the reporting structure, an additional phrase is then selected by the processor from the knowledge database as suggested to the user. This reduces the possibility that the user will miss commonly considered information related to the detailed report or context of other information discussed by the user within the report. Exclusion rules based on conflicts between phrases in the knowledge database apply to selected phrases to exclude and / or identify phrases that conflict with other included phrases and report conflicting findings. Reduce the possibility of including errors. The processor keeps track of usage patterns (decisions made by the user in the process of compiling the report regarding decisions to include and exclude phrases) and uploads such patterns to the knowledge database.

  FIG. 1 shows a report generation system 100 according to an exemplary embodiment of the present invention. This figure illustrates the basic configuration of the system 100 in which a central server or cluster of servers manages multiple coexisting users of the report editing system via the network 160, and includes templates, phrases, phrases and Access a database with data on template selection patterns and dynamically propose report content for multiple users. The database may also include phrases extracted from previously written reports over time and relationships observed between such phrases.

  The system 100 generally comprises a central computer 104, such as a server or server collection, and one or more (N) local computers 120 connected through a network 160. The local computer 120 may be, for example, a workstation, an image observation station, or an individual computer. The central computer 104 processes and stores the medical image inspection report file. Thus, the central computer 104 includes at least one processor 108 and at least one knowledge database processor 110. The designated processor may be separated or may be a single processor having multiple functions. The knowledge database processor 110 processes medical image data from the input 102 and data regarding phrases and relationships between phrases and generates phrases that are presented to the user via output 114 for selection to be inserted into the report. To do.

  One or more input devices 102 input data and information to the central computer 104. Input devices for computer-based, textual, or audio formats may include, for example, a keyboard, a touch screen, or a dictaphone. The input device 102 captures data such as medical image data and various concepts and phrase data that are selected and modified.

  Central computer 104 is also functionally connected to one or more output devices 114. The output device 114 may include a display, projector, printer, speaker, diskette, or other medium, for example, to store and display report files generated in various formats. The output device 114 allows the user to see the completed report or to see the work in progress during the creative process for evaluation. The output device may also provide reports of various file formats stored on separate storage devices such as DVDs, USB storage devices, or memory cards.

  Although the central computer 104 is shown as having a storage component 106, one or more additional storage devices 112 may also be operatively connected to the central computer 104. Such storage may include, for example, random access memory (RAM) and read only memory (ROM) devices, CD-ROM, flash memory, and various other storage disks. Other memory components may also be incorporated into the system to perform computer functions.

  The local computer 120 can be used by a user to generate, review, and modify phrases and concepts for reports contained within a report file on the central computer 104. Accordingly, the local computer 120 is shown as having a processor 122 and storage 124. In certain embodiments of the invention, the application of patterns to various phrases and report generation are performed and stored within the central computer 104, and only phrases and concepts are selected and modified at the local computer 120. The In another embodiment, the selection of phrases and concepts and the application of patterns to various phrases and concepts are performed on the local computer 120. Like the central computer 104, the local computer 120 will have one or more input devices 103 and one or more output devices 128, and additional storage 129 of the exemplary type described above. May also be included.

  Both the central computer 104 and the local computer 120 may include a graphical user interface for manipulating and displaying suggested phrases selected in the report file, as shown in FIG.

  Each computer described herein receives input data, processes data, communicates with other computers or processors in the network, and allows the user to view a variety of data It may be possible to operate alone.

  Embodiments of the present invention also use other computer systems, such as a network of processors and one or more storage units, or using a computer having a single processor and one or more storage units. May be used.

  In the exemplary embodiment as shown in FIG. 1, the central computer 104 and the local computer 120 may be embodied in a single computer system or, for example, in parallel, serial, or hub and You may connect through the network 160 by various structures, such as a structure of a spoke system. The computer network may include, for example, an internetwork such as a local area network (LAN), a corporate network, and the Internet. In this exemplary embodiment, the computers are connected via a LAN, such as an intranet within an organization. The network allows multiple users to work simultaneously on multiple sets of medical image data in an organization. The network may also be set up via the Internet to allow multiple parties to enter, view, or generate a medical imaging report.

  Both the central computer 104 and the local computer 120 can be a central processing unit (CPU), other centralized or main processor, a graphics processing unit (GPU), a set of computer-executable instructions, a software application for processing or medical use. There may be one or more software applications for processing data associated with the generation of an image inspection report, including any combination of the above processors.

  Input image data is generated by imaging techniques such as ultrasound, computed tomography (CT), computer-aided tomography (CAT), nuclear medicine, positron emission tomography (PET), and magnetic resonance imaging (MRI). Note that data representing the rendered image may be included, but is not limited to these data. Images can also be acquired from a modality such as an MRI scanner to a workstation through a network such as, for example, the Internet, a local area network (LAN), a wide area network (WAN), or other networks.

  It should further be noted that data entry into the system can be by any suitable method including, but not limited to, keyboard, touch screen, or voice.

  FIG. 2 is a flowchart for generating a medical image examination report according to an exemplary embodiment of the present invention. The report file is first created on the central computer or locally (S10) and then stored on the central computer. The medical image data can be read into the report file or can be a separate file. The report file has an adaptive structure that includes an examination type associated with the anatomical site to be examined and one or more fields or templates for accommodating imaging examination interpretations and diagnostic findings. The contents of the report file will be described in more detail below. Since there may be more than one region of interest corresponding to a given anatomical site, one or more examination types representing such regions of interest are the subject of the report It can be presented to the user via the output device for selection (S20). The user then determines which test type to report. The selected exam type is then entered into the report file. The test type is similar to the subject of the report. For each exam type, one or more diagnostic findings typically associated with a given exam type are generated by applying rule-based concepts such as inclusion rules in the knowledge database in the central server. . Further, more phrases are suggested to the user at the output device as related phrases or templates (S30). The rules and patterns in the knowledge database will be described in more detail below. When the user's selection is entered via the input device, the selected phrase is further processed at the server by applying exclusion rules from the knowledge database to exclude inconsistent or conflicting phrases as exclusion phrases. It is processed. The substitution rule may replace the excluded phrase with an alternative phrase (S40). If one selected phrase representing one medical finding or concept conflicts with another selected phrase or concept, the user will be warned of such conflict at the output device, or , The system can delete conflicting phrases. In certain embodiments of the invention, such a phrase is specified text in the reporting framework for the user to see potential conflicts with the remaining selected and / or suggested phrases. It may be presented in the field. This provides the user with safety measures that reduce errors in reporting. A preview of the report at the current stage of the report is continuously available during the process of creating and modifying the report (S50). That step may be repeated if it is desired to revise the report or add additional findings and observations to the report. Once all selections and modifications are complete, the report is completed and generated via the output device (S60). The pattern in which a user selects or excludes a phrase may be repeated and may be generated as a new rule uploaded to a knowledge database for future use by multiple users (S70), or A knowledge database specific to the user can be updated. A more detailed description of the report generation aspect according to an exemplary embodiment of the invention follows.

  There are several main systems, organs, and related concepts for virtually any type of test, and these body systems, organs, parts, and concepts are almost always “normal” forms and predetermined Medical imaging report is particularly adapted to this type of reporting system, given that it has a limited number of defined types of “abnormal” findings. Furthermore, abnormal findings do not exist in isolation, so that each normal and each abnormal finding has an equal opportunity to exist within a given report. Rather, certain abnormal findings exclude the presence of certain normal findings. Also, a given abnormal finding suggests that other, related, abnormal findings may also exist.

  Various embodiments of the present invention take advantage of the fact that medical imaging reports often contain a predetermined set of interrelated concepts, far from extensive open-ended reporting. The interrelated concept is a system that recognizes the existence of such a relationship, so that the user puts the report together for the user when inserting new ideas into the generated report. In order to generate a system that uses the existence of such relationships, the possibility of inclusion of other concepts in the report may be proposed or excluded. The notion of relevance that exists between commonly used phrases allows users to define new phrases and how such phrases relate to other phrases in the system. By doing so, the system's “knowledge base” for phrases used in report generation is customized. This allows the system to utilize and leverage those relationships to minimize the effort required to generate new reports for new cases.

  The knowledge database includes a library of commonly used medical phrases and data about how the phrases relate to each other. In addition, a template can be created that consists of a set of phrases associated with a commonly reported situation, such as a normal examination of a given type of standard. Phrases as well as phrases may also contain blank fields, and the user has been selected from a list of possibilities, such as a list of adjectives, such as a list of adjectives with a degree of mild, mild, moderate, severe, etc. You can put text in the blank field.

  Rapid generation of well-structured, complete radiology reports means that commonly used phrases in radiology reports do not exist in isolation, but rather in predictable, interrelated patterns It can be facilitated by recognizing and utilizing the fact of doing. Within that interrelated pattern, the use of one phrase may suggest the need to insert one or more related phrases, or the use of one phrase It may be suggested that it may be excluded.

  Each suggested related phrase can be associated with a priority value that is defined and entered by the user, or it can be determined by the system based on the pattern of phrase placement in the report by previous users. Such values are assigned according to the importance of the phrase, for example, a higher value indicates more important information that should be presented first in the report before less important information. For example, a phrase that refers to a life-threatening condition will be associated with a higher priority value, and this phrase will be added to the report following the addition of a less important finding Even if it is the last item, it will be automatically inserted towards the beginning of the report text while it is in progress. The user would be able to manually override this priority value based placement if desired. Conventional systems require the user to manually compile reports. Thus, the priority value feature of the present invention can save considerable time and effort for the report writer.

  In another aspect of the exemplary embodiment of the present invention, a user can define “variables” such as dates, measurements, etc. in expressions that the user can insert into sub-text fields. That is, the user says, “There is a [#] cm mass found in the right colon, which probably represents colon cancer. This [#] cm mass extends to the abdominal wall and possibly its abdominal wall. Would be able to present an expression that states " If the user updates a single case of that measurement, that case will be propagated to both locations in the text editing field on the user's screen when the template is inserted. Similarly, if the system extracts a sentence from the user's current text editing field for insertion into the database as an expected text suggestion, the system will be free for other instances of future use of the text representation by the user. Variables that can be set to the specific values required could be used to replace date, measurements, and other situation specific data.

  Embodiments of the present invention can maintain a database of multiple or all phrases previously used in reports by all users of the system or inserted into a standard template. Exemplary embodiments of the system can incorporate the concept of “standard templates” by allowing users to define a set of templates for their own use, which templates are also Will become part of a general database used or improved by other users. Furthermore, text in reports generated through the use of this system can be saved back to the database to continually improve the suggestions made to the user when writing the report. In an exemplary embodiment of the invention, all text in all reports is saved back to the database. The knowledge or report generation database is a means of proposing report corrections to users who continue to work on ongoing reports based on the frequency of use of a given user's predetermined phrases and based on patterns of use Used by the system, in the pattern of its use, the insertion of a given phrase suggests a high probability that another phrase will be used, or the insertion of a given phrase into the body of the text , Often accompanied by deletion of other phrases.

  The system will self-improve over time. A user who has found an expression or template generated by another user that is suitable for the user's needs but needs modification or improvement, is free to make such corrections or improvements to the template, The template is then again generally available to all users or selected users. If a modified phrase or template is generally selected later for use by other users more frequently than the original version before modification, the new modified version of the phrase or template will be It will be overridden by factors, which will result in the modified phrase or template being generally presented by future users of the system.

  When using all the text in all reports generated by the previous user (or the specified text in the specified report), the system generalizes the text in the previous actual report and diverges Delete names and identification numbers and generalize specific measurements, left-right difference display, severity display, etc. For example, a measurement value such as “5 cm” may be changed to “[#] cm” in which a specific measurement number is replaced with a blank. Similarly, phrases such as “moderate” can be replaced by a set of choices, and the user can insert “mild”, “moderate”, or “severe” from that choice at the reporting location. it can.

  A user can invoke a report phrase suggestion using a search box as shown in FIG. 23, and the user types some words in that search box that are relevant to what the user wants to report. That would cause the database to be searched for text or fragments of default phrases that previously contained the word and were previously generated by the user. For example, in FIG. 23, a previously generated phrase is presented to the user by searching for the word “injury”. A text suggestion invoked by this search could be stored according to the expected utility. Its expected usefulness is the frequency with which other users have selected the text proposal in question, and other users have selected other text proposals written by the same writer as the given text proposal. The frequency and frequency that the current user of the system has selected other text representations written by the writer of the text proposal in question and also appear in the text proposal in question and work by the current user of the system The number of words in the current text edit field that can be continued, this last factor being increased by the rarity of the word in question (also present in the user's current report text) Words that are rarely used and appear in the expected text suggestion also increase the probability of the displayed text suggestion, making the text suggestion more obvious. To the to the result that) display, and a medical image inspection metadata such as the patient's age and sex, could be based on factors such as other such factors.

  Over time, the system will develop phrases that can be applied even in unusual circumstances, and will begin to incorporate forms of subject knowledge for use by future users. become. Therefore, the system could be a knowledge database in addition to a simple phrase suggesting agency. Proposal presentations represent accurate information or concepts, although inaccurate information may enter the system because the user is influenced by the frequency of accepting or rejecting the given representation An attractive means would be to be prioritized over time when the system is used by a group of users, and the system will be able to determine from user decisions based on the system's aggregated knowledge. Enjoy the benefits. For example, in the case of diagnostic imaging report generation, a list of expected diagnoses for a given finding is “There is a nodule of about [#] cm in the [left or right] adrenal gland in the system. The field field measurement is [#] HU, given the low Hounsfield unit measurement, this probably represents an adenoma or myelolipoma.MRI with repeated phase imaging will be further evaluated. It will be "

  The means by which the system will suggest phrase insertion or deletion has a more detailed explanation.

  The system may choose a representation that is generally selected following the use of another representation, or a representation that is generally excluded from the report body following the use of another representation, or other expected inclusion or In order to intelligently propose an expected exclusion from the report text, for example, the passage of time with respect to a commonly selected expression based on, for example, the type of examination, the age of the patient, and other text existing in the report text You can take advantage of the patterns of use of expressions that you see along with.

  That is, if it is observed that a given expression, which will be referred to as expression # 2, is generally inserted into the text body after expression # 1 is inserted, the system lists the suggestions that contain expression # 2 May be presented to the user and the user may consider inserting the expression # 2 into the report.

  Furthermore, the system will note that if expression # 3 is present in the report, that expression # 3 is normally removed after expression # 1 is inserted. The system may suggest that the user remove representation # 3 from the system under these circumstances.

  Also, any information available to the system that will affect expression suitability can be exploited for the purpose of making an appropriate proposal. For example, other text that is currently in the user's report body can be searched for words, particularly rare words, and these can be compared to words in the expression available to prioritize their suggestions. For example, if a radiologist is writing a report about an unusual type of tumor, a suggestion is that other expressions and other radiologist users about the type of such tumor were stated by the system. Can be displayed.

  Three illustrative examples are given below to illustrate the steps involved in generating a medical imaging report. Illustrative example 1 illustrates the data structure (knowledge-based elements such as phrases and associated rules) used in the following two examples.

  Illustrative Example 1: The system has the following data structure: (1) Exam Type, (2) Phrase, and (3) Group. Each test type has the following characteristics: (i) the name of the test type, (ii) a group of phrases associated with the test type, and (iii) a phrase that is stitched together to produce a standard normal report. ID #. Each phrase has the following characteristics: (i) a phrase ID # that is not visible to the end user but useful for programming, (ii) a title or reference for referring to the phrase in the menu for the user to identify Name, (iii) a priority value that indicates the relative importance of the findings when present in the report, (iv) includes other phrases that should be excluded from the report when this phrase is inserted Exclusion list, (v) other phrases that the user will want to insert if this phrase is included in the report, and that the system will substitute for this phrase if this phrase is excluded Has a suggestion list that includes alternative phrases that will be tried. Each group is a plurality of phrases having elements, ie, the name of the group and the ID # of the phrase within the group.

The following structure definitions are used:
Definition of examination type CT abdominal pelvis Phrase group associated with CT abdominal pelvis:
Lung floor, abdomen, pelvis, bone and soft tissue Standard normal reports for CT abdominal pelvis:
Insertion phrase: 1, 2, 4, 8
Phrase group definition:
Lung bottom
Inclusive phrase: 1
abdomen
Inclusion phrases: 2, 3, 4, 5, 6, 7, 10, 11, 12
pelvis
Inclusive phrase: (not used in this example)
Bone and soft tissue
Inclusion phrases: 8, 9, 13
Definition of phrases 1) Phrases in the main report:
The lung bottom is clear. Pleural effusion or pulmonary cirrhosis is not clearly seen in the lung bottom.
Title for menu: Normal lung bottom Feeling equivalent:
The lung bottom has a normal appearance.
Priority value: 1
Exclude: None Suggestion: None Try to insert if excluded: None Include in group: Abdomen 2) Phrases of main report text:
There is a normal appearance of the liver, gallbladder, spleen, pancreas, kidney, adrenal gland, and intestine.
Title for menu: normal abdominal organs and intestines.
Equivalent Equivalent: Same as complete phrase Priority value: 1
Exclude: None Suggestion: None Try to insert if excluded: 3, then (... list of other possibilities)
3) Main report text phrases:
There is a normal appearance of the liver, spleen, pancreas, kidneys, adrenals, and intestines.
Title for menu: Normal abdominal organ / intestinal sensation equivalent except gall bladder: Same as main phrase 4) Main report phrase:
No free liquid or free gas.
Menu title: Same as main phrase Impression equivalent: Same as main phrase Priority value: 1
Exclude: None Suggestion: None Try to insert if excluded: 5, then 6 if 5 is impossible
5) Phrases of main reports:
No free liquid.
Menu title: Same as main phrase Impression equivalent: Same as main phrase Priority value: 1
Exclude: None Suggestion: None Try to insert if excluded: None 6) Principal report text phrase:
No free gas.
Menu title: Same as main phrase Impression equivalent: Same as main phrase Priority value: 1
Exclude: None Suggestion: None Try to insert if excluded: None 7) Principal report text phrase:
No serious lymphadenopathy.
Equivalent Equivalent: Same as main phrase Priority value: 1
Exclude: None Suggestion: None Try to insert if excluded: None 8) Principal report text phrase:
Bone and soft tissue appear within normal limits.
Title for menu: Normal bone and soft tissue Feeling equivalent: Same as main phrase Priority value: 1
Exclude: None Proposal: None Try to insert if excluded: 9, (... then list of other possibilities)
9) Main report text phrases:
No bone lesions appearing to be aggressive. Soft tissue appears within normal limits.
Menu title: No aggressive bone lesions. Normal soft tissue.
Equivalent feeling: Same as main phrase Priority value: 5
Exclude: None Suggestion: None Try to insert if excluded: None 10) Principal report text phrase:
Gallstones are found in the gallbladder that is mildly dilated. There is mild gallbladder wall enlargement and a slight liquid around the gallbladder. The findings suggest acute cholecystitis. Clinical correlation is recommended.
Title for menu: Acute cholecystitis Equivalent: Gallstones found in the gallbladder with mild gallbladder wall enlargement, gallbladder expansion, and perigal fluid, suggest acute cholecystitis.
Priority value: 70
Exclude: 2, 4, 5, (... and others)
Proposal: 11, 12
If excluded, try to insert: none 11) Main report body phrase:
Bile duct dilation is not seen. Common bile duct stones are not seen.
Title for the menu: no bile duct dilation or stones
No bile duct dilation or common bile duct stones.
Priority value: 5
Exclude: None Suggestion: None Try to insert if excluded: None 12) Principal report text phrase:
[#] Mm gallstones are also found in the common bile duct at the head of the pancreas with [mild / moderate / severe] intrahepatic and extrahepatic bile duct dilatation.
Title for menu: Intra-bile duct (CBD) gallstone with biliary dilatation Equivalent:
[#] Mm gallstones in the common bile duct in the head of the pancreas, with [mild / moderate / severe] intrahepatic and external biliary dilatation.
Priority value: 5
Exclude: None Suggestion: None Try to insert if excluded: None 13) Principal report text phrase:
There is a mild degenerative disease of the lumbar spine. Mild degenerative disease of the lumbar spine.
Title for the menu: Mild lumbar degenerative disease Feeling equivalent:
Mild lumbar degenerative disease.
Priority value: 30
Exclude: 8
Suggestion: None Try to insert if excluded: None

  The phrase selection based on the above definition is exemplified in Examples 2 and 3.

  Illustrative Example 2: The following is a workflow executed in accordance with an illustrative embodiment of the present invention, using the structural definitions defined in Example 1. First, as illustrated in FIG. 3, the user selects examination types, for example, CT of the abdomen and pelvis from a list of examination types. The report text is populated with a standard normal template for this type of test. That is, a standard normal report template is used that consists of a set of standard phrases that are inserted together in the report and used to describe a typical normal test. The normal reporting of this standard can be defined as such under this system. FIG. 4 illustrates the appearance of the report edit field after this standard normal template has been applied.

  When proceeding from the normal report of this standard, the user inserts pathological or other findings in stages. The phrases in this example are grouped together for easy selection by the user. First, the user is presented with a set of pop-up menus that reference various anatomical sites that are visible or arguable on the “abdominal and pelvic CT”. As illustrated in FIG. 5, there may be a pop-up menu or other GUI element that allows the user to select a condition for the lung bottom, abdomen, pelvis, or bone and soft tissue. If the user clicks on, for example, “abdomen”, the user selects the lower hierarchy, “gallbladder”, and the following standard phrase indicating signs of acute cholecystitis: “gallstones are mildly expanded It is found in the gallbladder. There is mild gallbladder wall enlargement and slight perigal fluid. Findings suggest acute cholecystitis. Clinical correlation is recommended. " The title of this phrase for easy reference in the menu is “Acute Cholecystitis”.

  This phrase will also be associated with the “feelings” equivalent, “gallstones found in the gallbladder with mild gallbladder wall enlargement, gallbladder enlargement, and perigal fluid, suggest acute cholecystitis.” .

  This phrase is also associated with an exclusion rule for some standard normal expressions. For example, the report cannot state that there is no free liquid because there is free liquid around the gallbladder. The report also cannot state that there is a normal appearance of the abdominal organs and intestines. An exclusion rule that excludes the phrase “no free liquid or gas” would be applied. There is also an exclusion rule that prohibits phrases that state that the abdominal organs and intestines are normal. If the above rule for free liquid is ruled out by a rule stating that a gallbladder / cholecystitis phrase cannot be associated with this phrase, the phrase “no free liquid or free gas” is possible for itself Suggests “no free gas”. This is due to the relationship between phrases, in which the phrase “no free gas” attempts to be inserted into the report text by the system when “no free liquid or free gas” is excluded. This is a standard potential alternative phrase. This alternative phrase is not excluded by any other rule, so the alternative phrase is inserted.

  If an unusual finding is inserted into the report, the system will be identified from the “most normal” representation so that the system will find the first appropriate surrogate representation that describes a lesser set of normal findings. Start and search through the list to find the first normal expression that is not excluded by other phrases present in the report, and finally insert the first “most normal” normal expression that is not excluded, Many sets of standard normal descriptions for abdominal organs may be created.

  From an algorithmic point of view, when a “normal” expression is excluded from the report, the action is associated with this exclusion and the phrase suggests that the program insert another normal expression at that position. This next expression is checked for exclusion, and if that expression is further excluded, it suggests another normal expression. This continues from one lack of normality to another phrase until a phrase that can eventually be inserted is found. This finally insertable phrase is inserted into the report.

  Each representation also has a “priority” number that determines the order of insertion into the report. Items with higher “priority” numbers are inserted first. Items with lower “priority” numbers are placed later. A higher “priority” number is associated with a more critical, life-threatening item or an item that requires prompt intervention. The gallbladder phrase for cholecystitis will have a high priority number, perhaps 70 outside the 100 possible range. The normal normal representation has a low priority number such as 1. The report then includes additional phrases that have been selected by the user and is illustrated in FIG.

  The system then suggests the generally associated phrase to the user via the knowledge database. These phrases may have been defined by the user at the time the user put them together, or could be determined automatically based on the previous usage / insertion pattern of the phrase. A well-thought-out system will contain a large number of phrases from each type of source.

  What is associated with the gallbladder / gallbladder phrase in this example is the following commonly used phrase (a) “No bile duct dilation seen. Common bile duct stones not seen.” mm gallstones are also found in the common bile duct in the head of the pancreas, with [mild / moderate / severe] intrahepatic and extrahepatic bile duct dilation ”.

  The user selects (b) when gallstones are visible in the common bile duct with the expanded bile duct. When this phrase is selected, the user types “5” indicating the gallstone measurement value into the blank space of [#]. The user also chooses between “mild”, “moderate”, and “severe” to indicate the degree of bile duct dilation. The sentimental equivalent for this phrase is “[#] gallstones in the common bile duct with biliary dilatation”. Thus, when a user inserts a phrase once in the report, the user enters only the measure and severity information degree. However, doing so simultaneously updates the information in the sentiment equivalent. The updated report is illustrated in FIG.

  The user generally does not want to review the list of associated expressions and select anything further. The user may then select a description stating “Mild degenerative disease of the lumbar spine” from the “Bone and Soft Tissue” pop-up menu. This expression has a priority value of “5”, and the impression equivalent of the expression “mild degenerative disease of the lumbar spine” has a priority value of “5”. Thus, these items are less important than the gall bladder description, but more important than the standard normal report. These statements exclude the description "Bone and soft tissue are within normal limits." Because the phrase is excluded, it tries to propose a replacement phrase “No bone lesions that appear aggressive. Soft tissue is within normal limits” that are not excluded by any phrase currently in the report. . Therefore, this substitution phrase is inserted. The updated report is shown in FIG.

  Alternatively, the user will also prefer to have a brief impression. The user can arbitrarily set a threshold value of the priority value for including the phrase element in the impression. For example, no more than three findings will be reported unless such a finding has a priority level of 50 or more. The user will then ensure that any findings needed in the impression would have to have a priority level of 50 or higher if it was inserted into the report. . Since the user has specified that no more than four impression items should be included unless the priority value is greater than 50, the final report excludes the fifth item and is illustrated in FIG. The

  Illustrative Example 3: This example illustrates the use of another exemplary embodiment of the present invention to edit a text field in a medical imaging report. In this example, a report on interpretation of a normal abdominal CT image examination is illustrated. First, the user opens a blank report screen as shown in FIG. The user then searches for a normal report template for reporting a normal abdominal CT imaging examination and types a search term such as “negative abdominal CT”. The system generates several suggested phrases at the server. These phrases have been previously defined and are associated with search terms. The proposed phrase is then displayed for review by the user, as shown in FIG. The user selected the first suggested phrase to be placed in the report edit field, as shown in FIG. In doing so, several updated proposals are created based on the previous usage pattern of this user's system, other users in the system, and words in the report text. In this example as illustrated in FIG. 13, the patient is a 70-year-old male and his age and gender information is also entered into the system, which information is then almost assured as osteopenia etc. Is used to suggest other phrases and expressions, which are generally selected when the user edits reports about patients of that age and gender group. The user may save the entire template or a portion of the template in the general knowledge database for future use when using similar findings.

  While having a basic reporting framework in place, the user wants to report on anomalies. In this example, a right renal mass with a medium density of about 5 cm in size blocks the right kidney (right hydronephrosis). The user decides to search for a phrase for a renal mass in the search box of the reporting interface, as shown in FIG. At the time of image interpretation, the user is not sure what the mass is, but the first representation in FIG. 14 is considered a useful framework to start with that representation, and therefore the representation is select. The user also enters a value in the column given a choice between “left” and “right” for the tumor size and parameters. Directional data (ie, left or right) and size data need only be entered at one location, as shown in FIG. 15, and are reflected throughout the text. This allows the user to save time from entering repeated information and reduces reporting errors. When adding a renal mass phrase, the system applies exclusion rules to the phrase to detect reciprocity and inconsistencies. Expressions in the first normal report inserted by the user inconsistent with the renal mass findings are, for example, italic red and flagged, alerting the user to potential reciprocity. The user can click on the red expression in italics to delete, edit, or select a commonly substituted expression, as shown in FIG. FIG. 17 shows a substitute phrase suggested by the system when the user clicks the red expression in the first italic font. The user may update the report as shown in FIG.

  Figure 19 shows how a user can also insert free-form text in a subtext field into a defined expression to report details not included in the ordinary expression from the database. Indicates. This allows the user to deviate from the standard defined representation in order to fully customize the report. In this example, the user has added “or hemorrhagic hemangiomyelolipoma” within the impression. Adding this text, the system identifies "angiomyelolipoma" as a rare term that would be valuable for suggesting additional phrases. An additional phrase containing the term angiomyelolipoma is then suggested to the user via the interface as shown in FIG. The user chooses to insert an additional phrase in the report as shown in FIG. 21 to complete the report.

  FIG. 22 shows a screenshot of a system interface according to an exemplary embodiment of the present invention.

  The present invention includes a method as described herein, a computer readable medium programmed to perform the method, and a computer system configured to perform the method. A computer system includes a machine readable medium having executable code, a device having one or more processors, a memory coupled to the processor, and a machine readable medium having machine readable program code. And input and output devices connected to the processor (s) for producing reports or other material. While the invention has been described in terms of exemplary embodiments, further advantages and modifications will occur to those skilled in the art. Accordingly, the invention of the superordinate aspect of the invention is not limited to the specific details shown and described herein.

  Modifications may be made without departing from the spirit and scope of the invention, for example modifications to the type of report and specific features of the software. Accordingly, the invention is not limited to the specific exemplary embodiments, but is within the full spirit and scope of the described embodiments and appended claims, and their equivalents. It is intended to be construed as being.

Claims (19)

A computer-based method for generating a medical imaging report, comprising:
(A) phrases associated with multiple types of imaging examinations to be interpreted in medical imaging examinations, (b) relationships between those phrases, and (c) a template comprising a set of such phrases, and / or (d A) a medical image report generation database on a computer readable medium containing one or more of the complete text of reports previously written within and / or out of scope of the system from which phrase suggestions can be derived Storing, and
Storing executable computer code in a computer processor connected to the computer readable medium;
Generating a medical imaging report for display on a computer display device using the stored report generation database by executing the computer code using the computer processor, the medical code The medical imaging report includes a structure containing one or more text fields, generating a medical imaging report and displaying to a user via a user output device by executing the computer code Generating a plurality of examination types for:
A template selected by the user based on the selected examination type and / or by executing the computer code using the computer processor when the user selects an examination type via a user input device Providing one or more suggested initial phrases of potential interest from the report generation database to the user via the user output device;
When the user selects one or more initial phrases via the user input device, the report generation based on the selected initial phrases by executing the computer code using the computer processor Providing one or more suggested related phrases of potential interest from a database via the user output device;
When the user selects one or more related phrases via the user input device and / or when the user enters other desired arbitrary text into the report, the computer code is used using the computer processor. Applying one or more inclusion rules based on usage patterns to the one or more selected first phrases and / or in a previously written radiology report database Presenting the one or more suggested phrases to the user via a user output device by applying pattern matching rules to identify potentially useful text from other reports stored in And steps to
In order to identify one or more excluded phrases to be excluded from the report by executing the computer code using the computer processor, the one or more initial phrases and / or the one or more Applying one or more exclusion rules based on usage patterns to a plurality of related phrases and / or the ongoing report via the user output device and resulting from these processes sent to the report output device Generating a report;
Applying a substitution rule to replace the excluded phrase with an alternative phrase by executing the computer code using the computer processor and proposing the alternative phrase to a user .   The method of claim 1, further comprising updating the exclusion rules and inclusion rules based on at least a portion of the generated report and inclusion and exclusion selection decisions made by the user.   3. The method further comprising the step of performing the method through a network, wherein updating the exclusion rule and the inclusion rule is based on at least a part of one or more reports generated by one or more users. The method described in 1.   Providing medical image data associated with one or more medical image examinations corresponding to one or more anatomical sites to be considered and interpreted in the medical imaging examination; and the computer processor Generating the one or more proposed examination types based on the medical image data by executing the computer code using a computer.   The method of claim 1, wherein at least a portion of the one or more text fields are occupied by one or more templates, each template including a set of phrases associated with the selected exam type. .   Assigning priority values to one or more of the phrases, and ordering the phrases in the report according to the priority values by executing the computer code using the computer processor The method of claim 1, further comprising:   The one or more text fields further include one or more subfields for user input text, wherein the user input text is optional in the text field of the related phrase in the report. The method of claim 1, wherein the method is selectively propagated.   Storing data representing the frequency of use of the modified phrase in the medical image reporting database; comparing the frequency of use of the modified phrase with the frequency of use of the corresponding unmodified phrase; Presenting any proposed phrase as a corrected phrase at the position of any corresponding uncorrected phrase having a lower usage frequency.   Storing data representing the frequency of use of the modified template in the medical image reporting database; comparing the frequency of use of the modified template with the frequency of use of the corresponding unmodified template; Presenting any proposed template as a modified template at the location of any corresponding unmodified template having a lower usage frequency. A medical image inspection report generation system,
One or more input devices for entering examination types, selecting phrases, and text;
One or more processors for processing the examination type selection, the phrase selection, the text, and one or more rule-based concepts;
One or more memory systems operatively coupled to the processor;
An output device operatively connected to the processor;
A system configured to perform the reporting steps of claim 1. A computer-based method for generating a medical imaging report, comprising:
(A) phrases associated with multiple types of imaging examinations to be interpreted in medical imaging examinations and inclusion and exclusion rules, and / or (b) a report template comprising a collection of such phrases, and / or (C) of data relating to patterns prior to the use of such phrases in cases prior to use of the system, and / or (d) a database of complete reporting texts that can be utilized to identify additional phrase suggestions Storing a medical image report generation database comprising at least one in a computer readable medium;
Executing computer code using a computer processor connected to the computer-readable medium;
Displaying a medical imaging report including a structure containing one or more text fields using the stored report generation database by executing the computer code using the computer processor. Generating above,
Presenting a plurality of examination types for display to a user via a user output device by executing the computer code using the computer processor;
When the user selects an examination type via a user input device, a report template having a plurality of phrases based on the selected examination type is output to the user by executing the computer code using the computer processor Providing to the user via a device;
Presenting one or more phrases associated with the template phrase via the user output device;
When the user selects one or more related phrases, one or more inclusion rules based on usage patterns are selected from the template phrase and the selected by executing the computer code using the computer processor. Applying to a related phrase and, when the user selects one or more related phrases, executing the computer code using the computer processor, thereby providing one or more based on a usage pattern for the phrase Applying an exclusion rule to the template phrase and the selected related phrase, thereby identifying one or more phrases to be excluded from the medical image report;
Applying a substitution rule to replace the excluded phrase with an alternative phrase by executing the computer code using the computer processor and proposing the alternative phrase to a user ;
Generating a medical image report including the selected phrase in the text field and the examination type via a report output device; at least a portion or all of the text in the generated report; Updating the database by storing the inclusion and exclusion selection decisions made by the user back into the database. One or more template phrases and one or more related phrases are associated with a priority value, and the method includes:
The medical report to the user via the user output device with the one or more prioritized related phrases and prioritized template phrases ordered according to their associated priority values. The method of claim 11, further comprising presenting within.   The method of claim 11, further comprising defining a set of templates associated with a particular user.   The method of claim 11, further comprising presenting a phrase based on a pattern prior to use of the phrase.   12. The method of claim 11, further comprising presenting the phrase based on the frequency of use of the phrase in similar reports and identifying a new phrase based on use in the previous report. Storing data representing the frequency of use of the modified phrase in the medical image reporting database;
Comparing the frequency of use of the modified phrase with the frequency of use of a corresponding unmodified phrase;
Presenting any suggested phrase as a corrected phrase at the position of any corresponding uncorrected phrase having a lower usage frequency;
The method according to claim 1, further comprising: Storing data representing the frequency of use of the modified template in the medical image reporting database;
Comparing the frequency of use of the modified template with the frequency of use of a corresponding unmodified template;
Presenting any proposed template as a modified template at the location of any corresponding unmodified template having a lower usage frequency;
The method according to claim 1, further comprising:   All or part of the text in the report generated by the user, including any new text that is uniquely generated by the user and entered into the system by the user. 12. The method according to claim 1 or 11, further comprising the step of saving in an image report database.   When any phrase or text is entered into the system by the user, not only from phrases previously defined in the system, but also from text contained in the database of previously written radiology reports, The method of claim 11, further comprising proposing additional expected phrases that the user may wish to incorporate into the report using a pattern matching method for identifying phrases.

Images