DE102021208123A1 - Computing device and method for validating information - Google Patents

Abstract

The invention provides a computing device and method for validating information extracted from an injection contrast report image file. The computing device (100) is arranged to implement: an input interface (110) arranged to receive an injection contrast report image file (1), an optical character recognition module (120) arranged to extract textual information from at least one extracting part of the received injection contrast report image file (1) (S200), and a validation check module (130) adapted to perform a validation check procedure on the extracted text information, wherein in the validation check procedure at least one validation condition is assessed (S310, S320), wherein the validation check module (130) is further configured to output: - a positive result for the validation check procedure VCP if all of the at least one validation conditions are met, and - a negative result for the validation check procedure VCP if at least one of the at least one validation condition is not met .

Description

field of invention

The present invention relates to a computing device and method for validating information generated from injection contrast report image files. Preferably, the information generated from the injection contrast report image file is in the form of a contrast structured report. In these cases, the computing device can also be referred to as a computing device for validating a contrast-structured report and the method can be referred to as a method for validating a contrast-structured report. The invention also proposes several measures and options for proceeding after a validation success or failure. Furthermore, the invention provides a computer program product, a data storage medium and a data stream.

Background of the Invention

Medical imaging modalities (or imaging devices) such as magnetic resonance imaging (MRI) devices or computed tomography (CT) devices often use contrast agents to improve image quality or to visualize portions of interest of the scanned object (usually a medical patient). The contrast agent is injected into the subject using an injector device, which is usually coupled to the imaging device to ensure correct timing.

To accurately interpret the results of imaging using contrast agents, accurate information about the timing and amounts of contrast agent injected into the subject is required. To this end, the injector devices generate injection reports. These injection reports typically include some information about the injector device used, some information about the subject (patient), and at least some information about the amount of contrast agent injected. The information on the amount of contrast agent is important, for example, for monitoring a patient's total exposure to contrast agents over time and can be an integral part of the patient's medical record. It is therefore important that the information contained in the injection reports is accurate.

In addition, personal information about the patient or information that can be traced back to the patient (commonly referred to as "personal health information" or PHI for short) is subject to data protection regulations in most jurisdictions (e.g. the European General Data Protection Regulation of the European Union or the US Health Insurance Portability and Accountability Act HIPAA) and therefore must be kept confidential. PHI may contain information about health status, provision of medical services, or payment for medical services that can be associated with a specific patient. Protected Health Information may also be referred to as "Protected Health Information" or "Confidential Health Information".

To some extent, the need to protect PHI contrasts with the need for larger and more accurate datasets for the training and improvement of artificial intelligence entities in the medical field, such as artificial neural networks or machine learning algorithms to aid or provide diagnostics or the like. Accordingly, records containing PHI are often required to undergo de-identification, wherein individually identifiable PHI is removed from the record to protect patient privacy.

Similarly, the need to have a large amount of medical data available to a large number of different stakeholders (e.g., hospitals, doctors, clinics, research institutes, and the like) for which cloud storage is most appropriate contrasts with the need to keep personal health information PHI within a group just large enough to treat each patient optimally.

Another challenge in this area is that there are a large number of vendors of injector devices and no single standardized format for the injection reports has been established. Moreover, in many cases, injection reports are generated as image files (referred to herein as "injection contrast report image files" or ICRIFs). In this context, image files should be understood in such a way that the files, for example in the form of so-called secondary acquisitions tion DICOM images (and/or as pixel images) as opposed to, for example, structured reports in the form of textual data in database entries or the like.

By convention, such ICRIFs are often generated using white or colored text characters on a dark or black background, and are therefore sometimes referred to as "black frames". The lack of an established standard means that also the position, size and style of presentation of the different pieces of information within each ICRIF are inconsistent between different injector devices and/or between different vendors. An excerpt from an example of such an injection contrast report image file ICRIF is in 3 shown, although for formal reasons the ICRIF is shown in black-on-white instead of the usual white-on-black.

This hampers efforts to develop technical solutions such as automatically removing PHI from injection reports, automatically validating the content of ICRIFs, and/or the like. However, for modern big data applications, for the development of medical databases and large health records, it would be highly desirable if ICRIFs from any type of vendor and any type of injector device could be processed automatically.

Structured reporting is a documentation technique in the medical field that captures and/or displays specific data items in a specific format. In radiology, most reports are usually written as free text narratives, for example using a dictaphone, handwritten or by typing. Compared to such unstructured, free-text reports, structured reports provide consistency and clarity, prompt for necessary data elements, and lend themselves to scalable data collection, interoperability, and sharing, citing, for example, the scholarly publication of Schwartz LH, Panicek DM, Berk AR, Li Y, Hricak H, Improving communication of diagnostic radiology findings through structured reporting, (2011) Radiology, 260 (1): 174-81. doi:10.1148/radiol.11101913. It is therefore evident that it is very beneficial to be able to obtain structured reports of any type of medical imaging scan.

Summary of the Invention

An object of the present invention is to provide a computing device and a method by which a reliable validation check (or plausibility check) can be carried out for information generated from injection contrast report image files ICRIF, in particular for a validation check of contrast-structured reports CSR.

The problems described above are solved by the subject matter of the aspects of the present invention according to the subject matter of the independent claims.

Due to the sometimes long names of certain elements, steps, functions or modules, abbreviations are used. Where inclusion of the full notation is believed to facilitate understanding, the full notation is used along with the abbreviation, such as in "generic computing device GCD." In another context, when it is believed that including the full designation may impede understanding, for example in more complex sentences, only the abbreviation (in this example GCD) is used. For consistency, reference numbers accompany abbreviations where common.

• eine Eingabeschnittstelle, die dafür ausgelegt ist, eine Injektionskontrastbericht-Bilddatei ICRIF zu empfangen,
• ein optisches Zeichenerkennungsmodul OCRM, das dafür ausgelegt ist, Textinformationen aus zumindest einem Teil der empfangenen Injektionskontrastbericht-Bilddatei ICRIF zu extrahieren, und
• ein Validierungsprüfmodul VCM, das dafür ausgelegt ist, eine Validierungsprüfprozedur VCP an den extrahierten Textinformationen ETI auszuführen, wobei in der Validierungsprüfprozedur VCP wenigstens eine Validierungsbedingung beurteilt wird (d. h. festgestellt wird, ob die wenigstens eine Validierungsbedingung erfüllt ist oder nicht),
• wobei das Validierungsprüfmodul VCM ferner dafür ausgelegt ist, Folgendes auszugeben:
  • - ein positives Ergebnis für die Validierungsprüfprozedur VCP, falls (und vorzugsweise wenn) alle der wenigstens einen Validierungsbedingungen erfüllt sind, und
  • - ein negatives Ergebnis der Validierungsprüfprozedur VCP, falls (und vorzugsweise wenn) wenigstens eine der wenigstens einen Validierungsbedingungen nicht erfüllt ist.

According to a first aspect of the present invention, there is provided a computing device for validating information generated from ICRIF injection contrast report image files, the computing device being adapted to implement:

• an input interface adapted to receive an injection contrast report image file ICRIF,
• an optical character recognition module OCRM designed to extract text information from at least part of the received injection contrast report image file ICRIF, and
• a validation check module VCM, which is designed to execute a validation check procedure VCP on the extracted text information ETI, wherein at least one validation condition is assessed in the validation check procedure VCP (ie it is determined whether the at least one validation condition is met or not),
• wherein the validation check module VCM is further adapted to output:
  • - a positive result for the validation check procedure VCP if (and preferably if) all of the at least one validation conditions are met, and
  • - a negative result of the validation check procedure VCP if (and preferably if) at least one of the at least one validation conditions is not met.

The extraction of the textual information from the injection contrast report image file ICRIF is preferably part of the generation of a contrast structured report CSR. Accordingly, the validation check can be carried out directly on the CSR itself. This has the additional advantage that the information contained in the injection contrast report image file ICRIF has already been entered into machine-readable data fields of the CSR, so that the validation check procedure can be carried out particularly easily. When a CSR is generated and the extracted textual information within the CSR is evaluated, the computing device can also be referred to as a computing device for validating (or for validating) a contrast structured report generated from injection contrast report image files ICRIF. A positive result of the validation check procedure VCP then indicates that the CSR is valid and a negative result of the validation check procedure VCP indicates that the CSR is invalid.

• - CSR, die direkt anhand Injektionskontrastbericht-Bilddateien ICRIF erzeugt werden, wenn sie beispielsweise von einer Injektorvorrichtung oder von einem Bildarchivierungs- und Kommunikationssystem PACS empfangen werden,
• - CSR, die anhand ICRIF erzeugt werden, bei denen die persönlichen Gesundheitsinformationen PHI entfernt wurden (PHIR-ICRIF)
• - oder von einem beliebigen dazwischen liegenden CSR-Typ.

In this way, contrast-structured reports can be CSR validated before further processing, without the need for time-consuming manual reviews by a team of medical specialists. The invention can be used for any type of contrast structured report CSR including for example the following:

• - CSR generated directly from injection contrast report image files ICRIF when received, for example, from an injector device or from an image archiving and communication system PACS,
• - CSR generated using ICRIF with PHI removed (PHIR-ICRIF)
• - or any CSR type in between.

For a better understanding of the present invention, some functions are described in terms of being performed by "modules" or "sub-modules". It should be noted that any or all of these modules may be implemented by hardware and/or software, which may be part of the apparatus or device described as comprising or executing the module or sub-module. Although different functions are attributed to different modules or sub-modules, it is to be understood that several or even all of the modules may be part of one or more software suites or applications that are executed by a single computing device (or multiple computing devices), where the program code responsible for each function cannot necessarily be programmed separately.

The computing device can be implemented as any device or any means for computing, in particular for running software, an app or an algorithm. For example, the computing device may comprise at least one processing unit such as at least one central processing unit CPU and/or at least one graphics processing unit GPU and/or at least one field programmable gate array FPGA and/or at least one application specific integrated circuit ASIC and/or a combination of the above mentioned. The computing device may further include memory operatively coupled to the at least one processing unit and/or non-volatile memory operatively coupled to the at least one processing unit and/or memory. The computing device may be partially and/or fully implemented in a local device and/or partially and/or fully implemented in a remote system, for example through a cloud computing platform.

In systems based on cloud computing technology, a large number of devices are connected to a cloud computing system via the Internet. The devices may reside on a remote facility connected to the cloud computing system. For example, the devices may include or consist of devices, sensors, actuators, robots, and/or machines in one or more industrial plants. The devices can be medical devices and devices in a healthcare unit. The devices may be home appliances or office appliances in a home/commercial setting.

The cloud computing system may enable remote configuration, monitoring, control, and maintenance of connected devices (commonly known as "assets"). Furthermore, the cloud computing system can store large amounts of data periodically collected from the devices, analyze the large amounts of data, and provide insights (e.g., key performance indicators, outliers) and alerts to operators, field technicians, or owners of the devices via a graphical user interface (e.g., of web applications). The insights and alerts can enable the devices to be controlled and maintained, resulting in efficient and fail-safe operation of the devices. The cloud computing system may also allow parameters related to the devices to be modified and issue control commands based on the insights and alerts via the graphical user interface.

The cloud computing system may include multiple servers or processors (also known as "cloud infrastructure") that are geographically distributed and interconnected through a network. A dedicated platform (hereinafter referred to as “Cloud Computing Platform”) is installed as a service (hereinafter referred to as “Cloud Service”) on the servers/processors to provide the aforementioned functionality. The cloud computing platform may include multiple software programs running on one or more servers or processors of the cloud computing system to enable delivery of the requested service to the devices and their users.

• Empfangen einer Injektionskontrastbericht-Bilddatei ICRIF,
• Extrahieren von Textinformationen aus zumindest einem Teil der empfangenen Injektionskontrastbericht-Bilddatei ICRIF und
• Ausführen einer Validierungsprüfprozedur an den extrahierten Textinformationen, wobei in der Validierungsprüfprozedur wenigstens eine Validierungsbedingung beurteilt wird (d. h. festgestellt wird, ob die wenigstens eine Validierungsbedingung erfüllt ist oder nicht), und

According to a second aspect of the present invention, a computer-implemented method for validating information generated from injection contrast report image files ICRIF is provided. The method comprises at least the following steps:

• receiving an injection contrast report image file ICRIF,
• extracting textual information from at least part of the received injection contrast report image file ICRIF and
• performing a validation check procedure on the extracted text information, wherein in the validation check procedure at least one validation condition is assessed (ie it is determined whether the at least one validation condition is met or not), and

• - eines positiven Ergebnisses für die Validierungsprüfprozedur, falls (und vorzugsweise wenn) alle der wenigstens einen Validierungsbedingungen erfüllt sind, und
• - eines negativen Ergebnisses der Validierungsprüfprozedur, falls (und vorzugsweise wenn) wenigstens eine der wenigstens einen Validierungsbedingungen nicht erfüllt ist.

Spend:

• - a positive outcome for the validation check procedure if (and preferably if) all of the at least one validation conditions are met, and
• - a negative outcome of the validation check procedure if (and preferably if) at least one of the at least one validation conditions is not met.

Extracting the text information is preferably part of the step of generating a contrast structured report CSR from the injection contrast report image file ICRIF. When a CSR is generated and the extracted textual information within the CSR is evaluated, the method can also be referred to as a method for validating (or validating) a contrast structured report generated from injection contrast report image files ICRIF. A positive result of the validation check procedure VCP then indicates that the CSR is valid and a negative result of the validation check procedure VCP indicates that the CSR is invalid.

According to a third aspect of the present invention there is provided a computer program product comprising executable program code which when executed is adapted to carry out the method according to an embodiment of the method according to the second aspect of the present invention.

According to a fourth aspect of the present invention, there is provided a non-transitory computer-readable data storage medium comprising executable program code which, when executed, is adapted to perform the method according to an embodiment of the method according to the second aspect of the present invention.

According to a fifth aspect of the present invention there is provided a data stream comprising or adapted to generate executable program code adapted when executed to perform the method according to an embodiment of the method according to the second aspect of the present invention to execute.

Further advantageous embodiments, refinements or variants of embodiments of the subject of the aspects of the present invention are present in the dependent claims and are described below in particular with reference to the figures and their detailed description.

According to some advantageous embodiments, refinements or variants of embodiments of the computing device according to the first aspect of the present invention, in the validation check procedure, within the extracted text information, a plurality of addends and at least one corresponding sum value that is said to be equal to the sum of the plurality of addends are identified. A first validation condition of the at least one validation condition can consist in the sum of the multiple summands being equal to the at least one corresponding sum value within a first area of uncertainty.

In this way, errors in particular in the procedure for optical character recognition OCR executed by the optical character recognition module OCRM can be determined easily and automatically.

Optical Character Recognition OCR is the extraction of text information from a raster graphic (like a picture with pixels). "Text recognition" is also used as a synonym for "optical character recognition". Here individual pixels of the image can be grouped and then mapped to a letter, number or character (represented by a number in a text encoding such as ASCII or Unicode). The grouping can be based on intensity values of the pixels. Each pixel has at least one intensity value that specifies the brightness or color of the pixel in the black image. Optionally, context analysis methods (“intelligent character recognition”, “ICR” for short) can be used to correct errors at the word level or sentence level, for example by comparing them with a dictionary. The process of optical character recognition and context analysis is well known to those skilled in the art, and for a more detailed description see, for example, the article "A survey of modern optical character recognition techniques" by Eugene Borovikov (https://arxiv.org/abs /1412.4183v1). Furthermore, optical character recognition is provided by several program packages (some of which are freely available).

It should be noted that in an injection contrast report image file ICRIF there can be several groups (of summands and a corresponding sum value) and that each such group can be subject to a validation condition. The phrase "at least one corresponding sum" indicates that in some cases one or more subtotal values for a given number of summands can be provided so that not only can the total be evaluated for plausibility reasons, but also the one or more subtotal values can be evaluated can become. In this case, each group (of summands and a corresponding subtotal value) can also be subject to a validation condition.

The first region of uncertainty can be defined as a percentage of the sum value and/or the sum of the summands. The first uncertainty region may also be a pre-determined (i.e., predefined) fixed value based on prior experience regarding how much substantially correct values will drift or differ from each other due to rounding, measurement uncertainty, etc. The first uncertainty region may also be a combination where both types of tests are performed and the difference between the sum value and the sum of the multiple addends is determined to be within the first uncertainty region only if both test types pass.

According to some advantageous embodiments, refinements or variants of embodiments, in the validation check procedure at least one value tuple (or n-tuple) is identified in the extracted text information, wherein a first value of the value tuple is a fixed value and a second value of the value- tuple is an actual value. Preferably, the at least one n-tuple is at least a 2-tuple or "pair", each pair consisting of the first value (fixed value) and the second value (actual value).

Each n-tuple (preferably pair) in the injection contrast report image file ICRIF may represent a different injected substance. For example, a first n-tuple pair may refer to a contrast agent and a second n-tuple may refer to a saline solution. Obviously, it is advantageous to check (or validate) that for each substance injected, the specified value (i.e. the amount specified to be injected) is equal to the actual value (i.e. the amount actually injected).

A second validation condition of the at least one validation condition can consist in the fact that in each of the at least one value tuples the respective first value is equal to the respective second value within at least one respective second region of uncertainty. In this context, it is to be understood that the "respective" second region of uncertainty is intended to mean that an individual second region of uncertainty can be defined for each tuple, particularly when the tuples are associated with different substances that can usually be injected in significantly different amounts from one another. Alternatively, the same second region of uncertainty can be defined for all tuples.

Each second uncertainty region can be defined as a percentage of the first value and/or the second value. The second region of uncertainty may also be a pre-determined (i.e., predefined) fixed value based on prior experience regarding how much substantially correct values will drift or differ from each other due to rounding, measurement uncertainty, etc. The second uncertainty region can also be a combination where both types of tests are performed and the difference between the first and second values of each tuple is determined to be within the respective second uncertainty region only if both types of tests are successful.

According to some advantageous embodiments, refinements or variants of embodiments, the computing device is further configured to implement a permission handling module AMM configured to whitelist an injector device identifier IDI indicating an injector device that has generated the injection contrast report image file ICRIF. Add database of valid injector devices if the validation module gives the positive result of the validation check procedure VCP.

Alternatively or additionally, the permission handling module AMM may be configured to remove an injector device identifier IDI from a blacklist database of unauthorized injector devices if the validation module outputs the positive result of the validation check procedure VCP.

According to some advantageous embodiments, refinements or variants of embodiments, the permission handling module AMM is adapted to initiate an error handling workflow in case the validation module returns the negative result of the validation check procedure VCP. The error handling workflow may include a number of known error handling techniques performed by human specialists and/or artificial intelligence entities. In this way, the computing device and the algorithms or modules it comprises and uses can be continuously updated and improved in order to be able to add as many injector device identifiers IDI as possible to the white list database.

Alternatively or additionally, the permission handling module AMM can be designed to add an injector device identifier IDI indicating an injector device that has generated the injection contrast report image file ICRIF to a blacklist database of unauthorized injector devices if the validation module outputs the negative result of the validation check procedure VCP.

• - „grau“ (gegenwärtig in einem Greylist-Dateispeicher GFS),
• - „in Arbeit“ (gegenwärtig geprüft, beispielsweise durch das Rechensystem, insbesondere durch das Validierungsprüfmodul VCM),
• - „schlecht“ (nicht erlaubt) oder
• - „ok“ (zulässig).

The whitelist database and/or the blacklist database can be implemented in a known manner. For example, only a single permission table can be maintained, which includes all the injector device identifiers IDI known to the computing device or to a system comprising the computing device. For example, in the Permissions table, each IDI present therein may be labeled (or tagged) as one of the following according to its current permission status:

• - "grey" (currently in a greylist filestore GFS),
• - "in progress" (currently checked, for example by the computing system, in particular by the validation check module VCM),
• - “bad” (not allowed) or
• - "ok" (allowed).

In other words, the whitelist database can consist of all entries in the permissions table that are labeled "ok" (i.e. legal) and the blacklist database can consist of all entries in the permissions table that are labeled "bad" (ie not allowed) are labeled. Accordingly, when adding an injector device identifier IDI to the whitelist database, the IDI can be labeled as "ok", while adding an IDI to the blacklist database may label the IDI as "bad". When removing an IDI from the whitelist database, the IDI may be relabeled from "ok" to "bad" or "work in progress" depending on the context of the removal, or the IDI may be deleted from the allow table altogether. When removing an IDI from the blacklist database, the IDI may be relabeled from "bad" to "ok" or to "in progress" depending on the context of the removal, or the IDI may be deleted from the allow table altogether.

• Feststellen, ob eine erste Validierungsbedingung der wenigstens einen Validierungsbedingung erfüllt ist, durch:
  • Identifizieren mehrerer Summanden und wenigstens eines entsprechenden Summenwerts, der gleich der Summe der mehreren Summanden sein soll, innerhalb der extrahierten Textinformationen und
  • Feststellen, ob die Summe der mehreren Summanden innerhalb eines ersten Unbestimmtheitsbereichs gleich dem wenigstens einen entsprechenden Summenwert ist.

Other methods of implementing the blacklist database and/or the whitelist database will be apparent to those skilled in the art. According to some advantageous embodiments, refinements or variants of embodiments, the method further comprises the following steps:

• Determining whether a first validation condition of the at least one validation condition is met by:
  • identifying within the extracted textual information a plurality of addends and at least one corresponding summation value intended to be equal to the sum of the plurality of addends; and
  • determining whether the sum of the plurality of addends is equal to the at least one corresponding sum value within a first uncertainty range.

• Feststellen, ob eine zweite Validierungsbedingung der wenigstens einen Validierungsbedingung erfüllt ist, durch:
  • Identifizieren wenigstens eines Werte-Tupels, wobei ein erster Wert des Werte-Tupels ein festgelegter Wert ist und ein zweiter Wert des Werte-Tupels ein tatsächlicher Wert ist, innerhalb der extrahierten Textinformationen und
  • Feststellen, ob in jedem der wenigstens einen Werte-Tupel der jeweilige erste Wert innerhalb eines zweiten jeweiligen Unbestimmtheitsbereichs gleich dem jeweiligen zweiten Wert ist.

According to some advantageous embodiments, refinements or variants of embodiments, the method further comprises the following steps:

• Determining whether a second validation condition of the at least one validation condition is met by:
  • identifying at least one value tuple, wherein a first value of the value tuple is a specified value and a second value of the value tuple is an actual value, within the extracted text information and
  • determining whether in each of the at least one value tuples the respective first value is equal to the respective second value within a second respective range of uncertainty.

• Hinzufügen einer Injektorvorrichtungskennung IDI, die eine Injektorvorrichtung angibt, welche die Injektionskontrastbericht-Bilddatei ICRIF erzeugt hat, zu einer Whitelist-Datenbank zulässiger Injektorvorrichtungen, falls das positive Ergebnis der Validierungsprüfprozedur VCP ausgegeben wird.

According to some advantageous embodiments, refinements or variants of embodiments, the method comprises:

• Adding an injector device identifier IDI indicating an injector device that has generated the injection contrast report image file ICRIF to a white list database of allowed injector devices if the positive result of the validation check procedure VCP is issued.

• Entfernen einer Injektorvorrichtungskennung IDI aus einer Blacklist-Datenbank nicht erlaubter Injektorvorrichtungen,
• falls das positive Ergebnis der Validierungsprüfprozedur VCP ausgegeben wird.

According to some advantageous embodiments, refinements or variants of embodiments, the method comprises:

• removing an injector device identifier IDI from a blacklist database of disallowed injector devices,
• if the positive result of the validation check procedure VCP is issued.

• Einleiten eines Fehlerbehandlungs-Arbeitsablaufs, falls das negative Ergebnis der Validierungsprüfprozedur VCP ausgegeben wird.

According to some advantageous embodiments, refinements or variants of embodiments, the method comprises:

• Initiating an error handling workflow if the negative result of the validation check procedure VCP is returned.

• Hinzufügen einer Injektorvorrichtungskennung IDI, die eine Injektorvorrichtung angibt, welche die Injektionskontrastbericht-Bilddatei ICRIF erzeugt hat, zu einer Blacklist-Datenbank nicht erlaubter Injektorvorrichtungen, falls das negative Ergebnis der Validierungsprüfprozedur VCP ausgegeben wird.

According to some advantageous embodiments, refinements or variants of embodiments, the method comprises:

• Adding an injector device identifier IDI indicating an injector device that has generated the injection contrast report image file ICRIF to a blacklist database of not allowed injector devices in case the negative result of the validation check procedure VCP is issued.

character list

The invention will be explained in more detail with reference to exemplary embodiments shown in the accompanying drawings.

• 1 ein schematisches Blockdiagramm einer Rechenvorrichtung gemäß einer Ausführungsform der vorliegenden Erfindung,
• 2 ein schematisches Flussdiagramm eines Verfahrens gemäß einer anderen Ausführungsform der vorliegenden Erfindung,
• 3 einen Auszug aus einem Beispiel einer typischen Injektionskontrastbericht-Bilddatei ICRIF in Schwarz-auf-Weiß anstelle des eigentlich üblichen Weiß-auf-Schwarz,
• 4 ein schematisches Blockdiagramm eines Computerprogrammprodukts gemäß einer wieder anderen Ausführungsform der vorliegenden Erfindung und
• 5 ein schematisches Blockdiagramm eines Datenspeichermediums gemäß einer noch anderen Ausführungsform der vorliegenden Erfindung.

The accompanying drawings are included to provide a further understanding of the present invention and are incorporated in and constitute a part of this specification. The drawings illustrate the embodiments of the present invention and together with the description serve to explain the principles of the invention. Other embodiments of the present invention and many of the advantages provided by the present invention will be readily appreciated as they become better understood from the following detailed description. Show it:

• 1 a schematic block diagram of a computing device according to an embodiment of the present invention,
• 2 a schematic flow diagram of a method according to another embodiment of the present invention,
• 3 an excerpt from an example of a typical injection contrast report image file ICRIF in black-on-white instead of the usual white-on-black,
• 4 a schematic block diagram of a computer program product according to yet another embodiment of the present invention and
• 5 12 is a schematic block diagram of a data storage medium according to yet another embodiment of the present invention.

Detailed description of the invention

1 10 shows a schematic block diagram of a computing device 100 according to an embodiment of the present invention.

2 shows a schematic flow diagram of a method according to another embodiment of the present invention.

The procedure off 2 may be performed using computing device 100, although not limited thereto. Accordingly, all options, variations, modifications, and refinements made to the computing device 100 may be made 1 were described, also affect the process 2 be applied and vice versa.

Accordingly, for the following description of computing device 100, reference is often made to the corresponding method and in particular to the flow chart 2 referenced.

The computing device 100 comprises an input interface 110 adapted to receive an injection contrast report image file ICRIF 1 . For example, the injection contrast report image file ICRIF 1 may be received from a local file server, from a cloud computing data store, and/or the like. The injection contrast report image file ICRIF 1 may be accompanied by an injector device identifier IDI 2 identifying an injector device or method used to generate the injection contrast report image file.

The injector device identifier IDI 2 may be part of a DICOM tag that may be associated with each injection contrast report image file ICRIF 1 . DICOM stands for "Digital Imaging and Communications in Medicine" and is a common standard for the communication and treatment of medical imaging information and associated data. It can be used for storing and transferring medical image files, enabling integration of medical imaging devices (such as scanners, printers, servers, workstations, network hardware, or PACS) from different vendors or manufacturers. Whenever "DICOM" is mentioned herein, it may specifically refer to the DICOM standard version DICOM PS3.1 2021b, although any earlier or later versions may also be used.

wobei „@“ das „@“-Zeichen als Teil einer Zeichenkette ist und die Symbole SIx, wobei „x“ von 1 bis 6 reicht, jeweils als „Injektorindex x“ bezeichnet werden und Folgendes angeben: SI1 einen Hersteller SI2 ein Herstellermodell SI3 einen Stationsnamen SI4 eine photometrische Interpretation SI5 eine Vorrichtungsseriennummer SI6 eine Softwareversion, wobei einige davon optional sein können.An injector device identifier IDI 2 can accordingly be generated according to a formula, preferably a DICOM-compatible formula, for example as follows: $$\text{IDI}=\text{'@'}+\text{SI1}+\text{'@'}+\text{SI2}+\text{'@'}+\text{SI3}+\text{'@'}+\text{SI4}+\text{'@'}+\text{SI5}+\text{'@'}+\text{SI6}+\text{'@'},$$

where "@" is the "@" sign as part of a string and the symbols SIx, where "x" ranges from 1 to 6, are each denoted as "injector index x" and indicate the following: SI1 a manufacturer SI2 a manufacturer model SI3 a station name SI4 a photometric interpretation SI5 a device serial number SI6 a software version, some of which may be optional.

Accordingly, the injector device identifier IDI 2 may be a character string beginning and ending with the character “@”.

Accordingly, as in 2 shown, in step S100 of the method an injection contrast report image file ICRIF (1) is received, preferably together with an injector device identifier IDI 2 .

The computing device 100 off 1 further comprises an optical character recognition module OCRM adapted to extract text information from at least part of the received injection contrast report image file ICRIF. Any method known in the art for optical character recognition OCR can be used, in particular methods that are set up to recognize light or white characters on a dark, especially black, background as in the usual formatting of the injection contrast report image files ICRIF 1. A more detailed description of OCR is provided, for example, by the article "A survey of modern optical character recognition techniques" by Eugene Borovikov (https://arxiv.org/abs/1412.4183v1). Furthermore, optical character recognition is provided by several program packages (some of which are freely available).

Accordingly, as again in 2 is shown, in step S200 text information is extracted from at least part of the received injection contrast report image file ICRIF 1, preferably using optical character recognition OCR.

As described above, step S200 may preferably be part of generating a contrast structured report CSR from the injection contrast report image file ICRIF1. The generation of the CSR means not only performing the OCR, but also assigning the extracted text information ETI to the machine-readable data fields of the CSR. The generation of the CSR can be performed by any method known in the art.

The computing device 100 further comprises a validation check module VCM 130, which is designed to execute a validation check procedure VCP on the extracted text information ETI, wherein at least one validation condition is determined in the validation check procedure VCP.

• - ein positives Ergebnis für die Validierungsprüfprozedur VCP, falls alle der wenigstens einen Validierungsbedingungen erfüllt sind und
• - ein negatives Ergebnis der Validierungsprüfprozedur VCP, falls wenigstens eine der wenigstens einen Validierungsbedingungen nicht erfüllt ist.

The validation check module VCM 130 is also designed to output:

• - a positive result for the validation check procedure VCP if all of the at least one validation conditions are met and
• - A negative result of the validation check procedure VCP if at least one of the at least one validation condition is not met.

The positive or the negative result can be output in the form of an output signal 71 from the validation check module VCM 130 . The output signal 71 may be output to another module within the computing device 100 or output to another computing device, such as a cloud computing platform implemented by a cloud computing system, via a wireless or wired interface.

In this way, in step S300 the process is off 2 a validation check procedure VCP is carried out on the extracted text information ETI, with at least one validation condition being assessed in the validation check procedure VCP S310, S320.

• - ein positives Ergebnis für die Validierungsprüfprozedur VCP ausgegeben, falls alle der wenigstens einen Validierungsbedingungen erfüllt sind, und
• - ein negatives Ergebnis der Validierungsprüfprozedur VCP ausgegeben, falls wenigstens eine der wenigstens einen Validierungsbedingungen nicht erfüllt ist.

In step S400 of the method 2 become

• - a positive result is issued for the validation check procedure VCP if all of the at least one validation conditions are met, and
• - A negative result of the validation test procedure VCP is issued if at least one of the at least one validation condition is not met.

Any number of validation conditions can be determined in the validation check procedure VCP, it being possible for the entire validation check procedure VCP to be regarded as having been carried out successfully only if it has been determined that all validation conditions have been individually met.

As previously described, if the extracted text information ETI was also used to generate a contrast-structured report CSR or if the ETI for the validation check procedure VCP are read from the CSR, the validation can also be viewed as a validation of the contrast-structured report CSR.

4 Figure 12 shows an excerpt from an example injection contrast report image file ICRIF 1, which includes various fields with various information displayed as text but encoded in a pixel data format.

On the right side of 4 There is a box titled "Injection Dates" which in turn is divided into three subsections, namely an upper section related to pressure, a middle subsection related to injected substances and their respective amounts, and a lower subsection related to injection time.

For the presently described embodiments, at least those in the middle subsection, i. H. textual information contained in the subsection related to the injected substances and their respective amounts is extracted.

The information therein can be transcribed into the following Table 1: Table 1: Subsection of Injection Contrast Report Image File ICRIF 1 related to injected substances and their amounts injected programmed total contrast agent 6.0ml 6.0ml total saline solution 30.0ml 30.0ml total quantity 36.0mL 36.0ml

The present invention uses the logical connections between some of the entries of this Table 1 to define validation conditions.

• VC1: die Summe aus dem „injizierten gesamten Kontrastmittel“ und der „injizierten gesamten Kochsalzlösung“ muss gleich der „injizierten Gesamtmenge“ sein (oder in Matrixeinträgen aus Tabelle 1: (2,2)+(3,2)=(4,2))
• VC2: die Summe aus dem „programmierten gesamten Kontrastmittel“ und der „programmierten gesamten Kochsalzlösung“ muss gleich der „programmierten Gesamtmenge“ sein (oder in Matrixeinträgen aus Tabelle 1: (2,3)+(3,3)=(4,3))
• VC3: das „injizierte gesamte Kontrastmittel“ muss gleich dem „programmierten gesamten Kontrastmittel“ sein (oder in Matrixeinträgen aus Tabelle 1: (2,2)=(2,3))
• VC4: die „injizierte gesamte Kochsalzlösung“ muss gleich der „programmierten gesamten Kochsalzlösung“ sein (oder in Matrixeinträgen aus Tabelle 1: (3,2)=(3,3))
• VC5: die „injizierte Gesamtmenge“ muss gleich der „programmierten Gesamtmenge“ sein (oder in Matrixeinträgen aus Tabelle 1: (4,2)=(4,3))

In the present example, these individual validation conditions are:

• VC1: the sum of “total contrast injected” and “total saline injected” must be equal to “total injected” (or in matrix entries from Table 1: (2,2)+(3,2)=(4,2 ))
• VC2: the sum of “programmed total contrast” and “programmed total saline” must be equal to “programmed total” (or in matrix entries from Table 1: (2,3)+(3,3)=(4,3 ))
• VC3: the "total contrast agent injected" must be equal to the "total contrast agent programmed" (or in matrix entries from Table 1: (2,2)=(2,3))
• VC4: the “total saline injected” must be equal to the “total saline programmed” (or in matrix entries from Table 1: (3,2)=(3,3))
• VC5: the "total injected" must be equal to the "total programmed" (or in matrix entries from Table 1: (4,2)=(4,3))

Validation is only considered successful if all five validation conditions VC1..VC5 are met (i.e. have been successfully assessed). H. the validation is TRUE if VC1 AND VC2 AND VC3 AND VC4 AND VC5 is TRUE.

Accordingly, in the example described here, in the validation check procedure, within the extracted text information ETI, a plurality of addends and at least one corresponding sum value, which is to be equal to the sum of the plurality of addends, are identified (step S311).

When the contrast structured report CSR is generated as part of the procedure, the identification of the summands, total values, etc. is usually particularly easy because they have already been assigned to data files of the CSR meaning different values.

In the present example, this step is performed twice: obviously should, as with reference to 3 and Table 1 notes that the first two numbers (from the top) in the "Injected" column (second from the left) must be equal to the third number (from the top) in the "Injected" column because according to Table 1 only "Contrast agent" and "saline" are injected and because "total amount" must be the sum of "total contrast" and "total saline". The same applies to the respective "programmed" column (first from the right).

Accordingly, in step S311 the addends "6.0 ml" and "30.0 ml" from the first and second rows of the second column are ("injected") from Table 2 together with the sum value "36.0 ml" from the third Row of the second column from Table 2, which should be equal to the sum of the summands "6.0 ml" and "30.0 ml", identified.

Furthermore, a different number of addends are identified with a corresponding total value, the addends "6.0 ml" and "30.0 ml" from the first and second rows of the third column ("programmed") of Table 2 along with the total value "36.0 ml" from the third row of the third column of Table 2, which should be equal to the sum of the summands "6.0 ml" and "30.0 ml".

Although the position of the subsection of the injection contrast report image file ICRIF 1 that relates to the injected substances and their respective amounts may not necessarily be known for each type and format of ICRIFs 1, the validation check module VCM 130 can be designed to use predefined keywords in the manner of "contrast agent", "saline", "total", "injected", "programmed" or the like, or even the specific wording used in each type and format of known ICRIFs. Accordingly, the precise information on where and how to extract the textual information may be pre-programmed into the computing device 100 . Such techniques can also be used to generate the contrast structured report CSR.

Alternatively, an artificial intelligence entity AIE may have been trained to extract the relevant textual information from a large number of injection contrast report image files ICRIF 1 and the trained AIE may then be provided within the validation check module VCM 130. The AIE can be trained with training data comprising labeled ICRIFs 1, the labels providing the location of the textual information to be extracted. In some variants the AIE may have been designed and trained to perform the OCR procedure and in other variants the AIE is designed and trained to determine the position within the ICRIFs which are then subjected to the OCR procedure. In particular, the AIE can be an artificial neural network ANN, preferably a convolutional neural network.

A contrast-structured report CSR is preferably generated by known means as part of the method and the extracted text information necessary for the assessment of the validation conditions can be taken from the corresponding data fields of the CSR.

After the extracted text information ETI has been provided in step S311, it is determined in step S312 whether the sum of the multiple addends equals the corresponding sum value. A range of uncertainty can be applied within which the terms can vary while still being considered equal, but for the purpose of this description of specific embodiments the issue of uncertainty is bypassed by assuming that all values in the injection contrast report image file ICRIF 1 are perfectly accurate, therefore all uncertainties can be set to zero. In the foregoing, however, several options and alternatives have been described in relation to regions of uncertainty, which those skilled in the art are free to apply (or modify) depending on the specific task at hand.

Now, in the present example, step S312 is performed first for validation condition VC1 by verifying that 6.0ml + 30.0ml is actually equal to 36.0ml, taking all values from the "Injected" column. Accordingly, as a result of step S312, the first validation condition VC1 is judged to be met (S310).

Because validation constraint VC2 is of the same type as validation constraint VC1, step S310 (comprising steps S311 and S312) is repeated once more for the "programmed" column. Again the result in this example is that the second validation condition VC2 is also met because again 6.0ml + 30.0ml is actually equal to 36.0ml.

The remaining validation conditions VC3, VC4 and VC5 are of a different type. In order to determine in a respective step S320 whether each of these is satisfied, at least one value tuple is first identified within the extracted text information ETI in a respective step S321, a first value of the value tuple being a fixed value and a second value of the value tuple is an actual value.

In the present example, three 2-tuples (or "pairs" for short) are found, one associated with each of the remaining validation conditions VC3, VC4, and VC5. Each pair includes a specified numerical value (referred to herein as "programmed") and a corresponding actual numerical value (referred to herein as "injected").

In a respective step S322 it is then determined whether in each of the three value pairs the respective first value within a second respective uncertainty range is equal to the respective second value, although we again neglect the question of uncertainties for this example and set the uncertainty range to zero. If this is the case for a pair, it is determined in the respective step S320 that the validation condition associated with this pair is met.

Because in the present case, in fact, all "Injected" values for "Total Contrast", "Total Saline" and "Total Amount" are each equal to their corresponding "Programed" value, it is found that also the third, fourth and fifth validation conditions VC3, VC4 and VC5 are met.

Accordingly, the overall result of step S300 (or the validation check module VCM 130) is that the validation was successful, so the output signal 71 is issued in step S400 to indicate a positive result of the validation.

The computing device 100 may further be configured to implement a permission handling module AMM 140 configured to perform further actions depending on the result of the validation check procedure VCP.

• In Schritt S510 kann eine Injektorvorrichtungskennung IDI 2, die eine Injektorvorrichtung angibt, welche die Injektionskontrastbericht-Bilddatei ICRIF 1 erzeugt hat, zu einer Whitelist-Datenbank zulässiger Injektorvorrichtungen hinzugefügt werden, falls das positive Ergebnis der Validierungsprüfprozedur für ICRIF 1 ausgegeben wird.
• In Schritt S520 kann eine Injektorvorrichtungskennung IDI 2 aus einer Blacklist-Datenbank nicht erlaubter Injektorvorrichtungen entfernt werden, falls das positive Ergebnis der Validierungsprüfprozedur VCP ausgegeben wird.
• In Schritt S530 kann ein Fehlerbehandlungs-Arbeitsablauf eingeleitet werden, falls das negative Ergebnis der Validierungsprüfprozedur VCP ausgegeben wird.
• In Schritt S540 wird eine Injektorvorrichtungskennung IDI 2, die eine Injektorvorrichtung angibt, welche die Injektionskontrastbericht-Bilddatei ICRIF 1 erzeugt hat, zu einer Blacklist-Datenbank nicht erlaubter Injektorvorrichtungen hinzugefügt, falls das negative Ergebnis der Validierungsprüfprozedur VCP ausgegeben wird.

As in 2 shown, some or all of the following actions can be performed, depending on the task at hand:

• In step S510, an injector device identifier IDI 2 indicating an injector device that generated the injection contrast report image file ICRIF 1 may be added to a white list database of allowed injector devices if the positive result of the validation check procedure for ICRIF 1 is returned.
• In step S520 an injector device identifier IDI 2 can be removed from a blacklist database of not allowed injector devices if the positive result of the validation check procedure VCP is output.
• In step S530, an error handling workflow can be initiated if the negative result of the validation check procedure VCP is returned.
• In step S540, an injector device identifier IDI 2 indicating an injector device that generated the injection contrast report image file ICRIF 1 is not allowed to a blacklist database Added injector devices in case the negative result of the validation check procedure VCP is issued.

Alternatively or additionally, the permission handler module AMM 140 may also be configured to perform other steps of the error handling workflow.

The error handling workflow may include a number of known error handling techniques performed by human specialists and/or artificial intelligence entities. For example, algorithm or software updates may be provided. In some cases, as described above, the troubleshooting workflow may include blacklisting an injector device identifier IDI, for example by adding the IDI to a record within a blacklist database of disallowed IDIs 2 . This condition may exist until a working OCR solution has been found for the injection contrast report image files ICRIF 1 associated with that particular injector device identifier IDI 2 .

Making the injector device identifier IDI 2 legal, or in other words making it legal by the whitelist database, can be done simply by adding the IDI 2 to a record of legal IDIs 2 in the whitelist database. Making the IDI 2 allowable is equivalent to making the previously unknown injector device part of the whitelist of allowable injector devices.

Although whitelist and blacklist are described herein primarily in terms of a whitelist database and a blacklist database, it should be understood that these can be implemented in any known manner. For example, only a single permissions table can be maintained that includes all known IDIs 2 , with each IDI therein classified as either "gray" (not currently evaluated), "work in progress" (currently in validation or evaluation), "poor" (not allowed) or "ok" (acceptable). In other words, the whitelist database can consist of all entries in the permissions table that are labeled "ok" (i.e., legal) and the blacklist database can consist of all entries in the permissions table that are labeled "bad" (i.e., not allowed) are labeled. Other methods of implementing the blacklist database and/or the whitelist database will be apparent to those skilled in the art.

4 Figure 2 shows a schematic block diagram of a computer program product 200 according to an embodiment of the third aspect of the present invention. The computer program product 200 comprises an executable program code 250 which, when executed, is designed for the method according to an embodiment of the method according to the second aspect of the present invention, in particular as with reference to FIG 1 and 2 described to perform. In particular, the program code 250 can be designed to execute or control the function or functionality of any of the steps S100..S540.

3 FIG. 3 shows a schematic block diagram of a non-transitory computer-readable data storage medium 300 according to an embodiment of the fourth aspect of the present invention. The data storage medium 300 comprises an executable program code 350 which, when executed, is designed to implement the method according to an embodiment of the method according to the second aspect of the present invention, in particular as referred to in FIG 1 and 2 described to perform. The program code 350 can be designed in particular to execute or control the function or functionality of any of the steps S100..S540.

Claims (14)

Computing device (100) for validating information generated from injection contrast report image files ICRIF (1), designed to implement: an input interface (110) designed to receive an injection contrast report image file ICRIF (1), an optical one Character recognition module OCRM (120) designed to extract text information from at least part of the received injection contrast report image file ICRIF (1) (S200), and a validation check module VCM (130) designed to perform a validation check procedure on the extracted text information to execute, wherein in the validation check procedure at least one validation condition is assessed (S310, S320), wherein the validation check module (130) is further adapted to output: - a positive result for the validation check procedure VCP if all of the at least one validation conditions are met, and - a negative result of the validation check procedure VCP if at least one of the at least one validation condition is not met. Computing device (100) according to claim 1 , wherein in the validation check procedure, within the extracted text information, a plurality of addends and at least one corresponding summation value, which is to be equal to the sum of the plurality of addends, are identified (S311) and wherein a first validation condition of the at least one validation condition consists in that the sum of the plurality of addends is equal to the at least one corresponding sum value within a first uncertainty range. Computing device (100) according to claim 1 or claim 2 , wherein in the validation check procedure at least one value tuple is identified within the extracted text information, wherein a first value of the value tuple is a specified value and a second value of the value tuple is an actual value, and wherein a second validation condition of the at least one Validation condition consists in that in each of the at least one value tuple the respective first value within a respective second area of uncertainty is equal to the respective second value. Computing device (100) according to one of Claims 1 until 3 which is further adapted to implement a permission handling module AMM (140) adapted to initiate an error handling workflow (S530) and/or an injector device identifier IDI (2) specifying an injector device generating the injection contrast report image file ICRIF has indicates to add to a blacklist database of not allowed injector devices (S540) if the validation module outputs the negative result of the validation check procedure VCP. Computing device (100) according to claim 4 wherein the permission handling module AMM (140) is further adapted to add (S510) an injector device identifier IDI (2) indicating an injector device that has generated the injection contrast report image file ICRIF (1) to a white list database of permitted injector devices and /or remove an injector device identifier IDI from a blacklist database of unauthorized injector devices (S520) if the validation module (130) outputs the positive result of the validation check procedure VCP (S400). Computer-implemented method for validation of information generated from injection contrast report image files ICRIF (1), comprising at least the following steps: receiving (S100) an injection contrast report image file ICRIF (1), extracting (S200) text information from at least part of the received injection contrast report image file ICRIF (1) and performing (S300) a validation check procedure on the extracted text information, wherein in the validation check procedure at least one validation condition is judged (S310, S320), and Output (S400): - a positive result for the validation check procedure VCP if all of the at least one validation conditions are met, and - A negative result of the validation check procedure VCP if at least one of the at least one validation condition is not met. procedure after claim 6 , further comprising: judging (S310) whether a first validation condition of the at least one validation condition is met by: identifying (S311) a plurality of addends and at least one corresponding summation value that is to be equal to the sum of the plurality of addends within the extracted text information and determining (S312) whether a sum of the plurality of addends is equal to the at least one corresponding sum value within a first uncertainty range. procedure after claim 6 or claim 7 , further comprising: assessing (S320) whether a second validation condition of the at least one validation condition is met by: identifying (S321) at least one value tuple, wherein a first value of the value tuple is a specified value and a second value of the value -tuple is an actual value within the extracted text information and determining (S322) whether in each of the at least one value tuple the respective first value is equal to the respective second value within a second respective range of uncertainty. Procedure according to one of Claims 6 until 8th , further comprising the step of: adding (S510) an injector device identifier IDI specifying an injector device which has generated the injection contrast report image file ICRIF to a white list database of permissible injector devices if the positive result of the validation check procedure VCP is issued (S400). Procedure according to one of Claims 6 until 9 , further comprising the step of: removing (S520) an injector device identifier IDI from a blacklist database of unauthorized injector devices if the positive result of the validation check procedure VCP is issued (S400). Procedure according to one of Claims 6 until 10 , further comprising the step of: initiating (S530) an error handling workflow if the negative result of the validation check procedure VCP is output (S400). Procedure according to one of Claims 6 until 11 , further comprising the step of: adding (S540) an injector device identifier IDI indicating an injector device that has generated the injection contrast report image file ICRIF to a blacklist database of prohibited injector devices if the negative result of the validation check procedure VCP is issued (S400) . Computer program product (200) comprising an executable program code (250) which, when executed, is designed to perform the method according to any one of Claims 6 until 12 to execute. A non-transitory computer-readable data storage medium (300) comprising executable program code (350) adapted, when executed, to perform the method of any one of Claims 6 until 12 to execute.

Images