WO2015044810A2

WO2015044810A2 - A system for assisting the transcription of lesion measurements

Info

Publication number: WO2015044810A2
Application number: PCT/IB2014/064270
Authority: WO
Inventors: Merlijn Sevenster; Joost Frederik Peters
Original assignee: Koninklijke Philips N.V.
Priority date: 2013-09-27
Filing date: 2014-09-05
Publication date: 2015-04-02
Also published as: WO2015044810A3

Abstract

A system for transcribing lesion measurements including a radiology report database which stores one or more radiology reports and a transcription device. The transcription device including a display which displays an interface and one or more processors programmed to receive a radiology report, identify one or more lesion measurement within the radiology report, display the radiology reports on the interface, and highlight the one or more lesion measurements on the display radiology report.

Description

A SYSTEM FOR ASSISTING THE TRANSCRIPTION OF LESION MEASUREMENTS

The present application relates generally to a system and method for assisting the transcription of lesion measurements. It finds particular application in conjunction with transcribing lesion measurements from radiology reports into databases or spreadsheets and will be described with particular reference thereto. However, it is to be understood that it also finds application in other usage scenarios and is not necessarily limited to the aforementioned application.

On a routine basis, clinical research associates (CRAs), also known as cancer registrars, and nurses are tasked with maintaining longitudinal records of cancer measurements of patients participating in clinical trials. The CRAs and nurses transcribe cancer measurements from a radiology report, which is typically written in a free text format, into a spreadsheet or proprietary database. The transcription process involves reading a prior report and identifying target and non-target lesions; reading a current report and identifying the target and non-target lesions; marking the target and non-target lesions, using the same tags used in the prior report; and copying measurement information from the current report into a spreadsheet or proprietary database. The transcription process is currently plagued by human error. If an error is made, patient care may be negatively affected, which may ultimately lead to suboptimal care delivery. The transcription process is also inefficient, which makes it time consuming and tiresome.

The present application provides new and improved methods and system which overcome the above-referenced problems and others.

In accordance with one aspect, a method for transcribing lesion measurement is provided. The method including receiving a radiology report, identifying one or more lesion measurement within the radiology report with a measurement detection engine, displaying the radiology reports on an interface, and highlighting the one or more lesion measurements on the display radiology report.

In accordance with another aspect, a system for transcribing lesion measurement is provided. The system including a radiology report database which stores one or more radiology reports and a transcription device. The transcription device including a display which displays an interface and one or more processors programmed to receive a radiology report, identify one or more lesion measurement within the radiology report, display the radiology reports on the interface, and highlight the one or more lesion measurements on the display radiology report.

In accordance with another aspect, a system for transcribing lesion measurement is provided. The system including a radiology report database which stores one or more radiology reports, a transcription device which includes a display which displays an interface and one or more processors which receive a radiology report, a measurement detection engine which identifies one or more lesion measurement within the radiology report, a display module which displays the radiology reports on the interface, and an application programming interface (API) which highlights the one or more lesion measurements on the display radiology report.

One advantage resides in the automated transcription of lesion measurements from radiology reports into databases or spreadsheets.

Another advantage resides in the error free transcription of lesion measurements from radiology reports.

Another advantage resides in providing a more efficient mechanism to transcribe lesion measurement from radiology reports.

Another advantage resides in increased workflow efficiency.

Still further advantages of the present invention will be appreciated to those of ordinary skill in the art upon reading and understanding the following detailed description.

The invention may take form in various components and arrangements of components, and in various steps and arrangement of steps. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention.

FIGURE 1 illustrates a block diagram of an exemplary embodiment of a lesion measurement transcription system according to aspects of the present application.

FIGURE 2 illustrates an interface of a lesion transcription device according to aspects of the present application.

FIGURE 3 illustrates a flowchart diagram of a method for transcribing lesion measurements according to aspects of the present application.

The present application provides a transcription system 10 that automatically transcribes lesion measurements from radiology reports, which are typically written in a free text format, into a spreadsheet or proprietary database. Specifically, the transcription system 10 analyzes radiology reports and highlights pertinent lesion information. The transcription system 10 identifies and extracts dimensional lesion measurements from an analyzed radiology reports and parses the extracted lesion measurements by unit and number. On a user interface, the one or more radiology reports are displayed and the extracted lesion measurements highlighted. Utilizing the user interface, a user can move selected highlighted lesion measurements into a spreadsheet or proprietary database. The transcription system 10 also enables the selected lesion measurements to be transmitted to downstream consumers, such as another software application or proprietary database. Further, the transcription system 10 can classify each lesion measurement according to its relevance to the clinical trial and its temporal status. Additionally, the transcription system 10 can correlate the lesion measurements from two or more radiology reports.

With reference to FIGURE 1, the transcription system 10 includes a measurement transcription device 12, a radiology report database 14, a lesion measurement database 16, and the like. Suitably, the components of the system 10 are interconnected through a communication network 18, such as the Internet, a local area network, a wide area network, a wireless network, or the like. The measurement transcription device 12 analyzes one or more radiology reports and highlights pertinent lesion information contained within the one or more radiology reports. Specifically, the measurement transcription device 12 includes a measurement detection engine 20 that identifies and extracts dimensional lesion measurements from the analyzed one or more radiology reports and a measurement parsing engine 22 that parses the extracted lesion measurements by unit and number. The measurement transcription device 12 displays the one or more radiology reports and highlights the extracted lesion measurements with a display module 24. The transcription system 10 also includes an application programming interface (API) 26 that enables selection of transcribed lesion measurements to be transmitted to downstream consumers, such as another software application or proprietary database. In one embodiment, the transcription system 10 includes a context-sensitive highlighting engine 28 that classifies each lesion measurement according to its relevance to the clinical trial and its temporal status. In another embodiment, the transcription system 10 includes a temporal correlation engine 30 that correlates the lesion measurements from two or more radiology reports.

The one or more radiology reports are typically stored within a radiology report database 14 which is accessed by the measurement transcription device 12 through the communication network 18. The radiology reports are an interpretation of one or more medical images which include a description of the findings and impressions or diagnosis. Specifically, the radiology reports including information relating to the identification of the lesions, image data such as the slice of contain the lesion of interest, measurements of lesions, dates of the measurements, location and orientation of the lesions, temporal status, patient identification, clinical trial identification, current and past medical images, and the like. Typically, the radiology reports are in a free text format. However, it should be appreciated that the radiology reports can be formatted in a structured manner. The radiology reports can be generated manually and/or automatically. As to the former, a radiologist drafts the radiology reports with one or more input devices. As to the latter, the radiology reports are generated without user intervention. For example, parts or all of the radiology reports are automatically generated through the use of third party software.

As stated above, the measurement transcription device 12 automatically detects lesions measurements in free text in the radiology reports, highlights the measurements appropriately, and provides an interactive user interface. In order to accomplish this, the measurement transcription device 12 includes a measurement detection engine 20 that identifies and extracts dimensional lesion measurements from the analyzed one or more radiology reports. For example, consider the sentence: "There is a hypodense lesion in liver segment 8 measuring 1 x 2 cm, previously measuring 3 x 4 cm." The measurement detection engine 12 recognizes the strings "1 x 2 cm" and "3 x 4 cm" as measurements. The measurement detection engine 20 includes a pattern recognition engine to execute the task of detecting measurements, for instance, based on rules or regular expressions. For example, the regular expression: \d x \d (cm|mm) would match strings of the form "1 x 2 cm" and "3 x 4 cm" but not "1 x 2 x 3 cm," or "1 x 2 and 3 x 4 cm," which are addressed by additional rules or regular expressions. Alternatively, statistical methods can be utilized to detect measurement descriptors. The extracted lesion measurements and measurement descriptors are then stored at measurement transcription data.

The measurement transcription device 12 also includes a measurement parsing engine 22 that parses the extracted lesion measurements by unit and number. Specifically, the measurement parsing engine 22 parses a string describing a measurement according to an underlying measurement scheme. For instance, the parsing engine distinguishes three components in the string "1 x 2 cm": the first measurement "1", the second measurement "2" and the unit "cm". The measurement parsing engine 22 also assigns slots to each rule or regular expression. The strings filling the slots are the components that fit in the scheme. For example, in the sample regular expression, the first occurrence of \d is associated with the first dimension of the measurement; the second occurrence of \d is associated with the second dimension of the measurement; and the disjunction (cm|mm) is associated with the unit of the measurement. The measurement parsing engine 22 can also utilize a simple rule-based method to normalize the information inserted in the slots. For example, the measurement parsing engine 22 can utilize rules which normalize the occurrences of "cm", "Cm" and "CM" to the same entity representing centimeter. The extracted and parsed lesion measurements and measurement descriptors are then stored at measurement transcription data.

The display module 24 generates and provides the measurement transcription device interface. After the measurement detection engine 20 and measurement parsing engine 22 identifies and extracts dimensional lesion measurements from analyzed radiology reports and parses the extracted lesion measurements by unit and number, the display module 24 displays the one or more radiology reports in which strings recognized as measurements are highlighted. The interface has an interactive component, which enables the user to select highlighted fields and designate the selected fields as target or non-target lesion. The interface also enables the user to designate the unique identifier of the lesion that is consistent with a pre-existing identification scheme. It should also be appreciated that the display module can display current and prior radiology reports. Specifically, the display module 26 also highlights or presented visually as buttons the measurement descriptors, both in the current as in the prior report. An appropriate visualization technique ensures that it is clear from the interface which lesion is being transcribed. If multiple measurement descriptors are utilized, the interface provides measurement descriptor differentiation (e.g. different colors).

The measurement transcription device 12 also includes an API 26 that enables transmission of the measurement transcription data to downstream consumers, such as spreadsheets, proprietary databases, or third-party software. The measurement transcription data is structured using a suitable markup language (e.g. XML) and contains parsed measurement information, along with an array of metadata possibly including target/non- target lesion status and unique identifier. The measurement transcription data from the measurement parsing engine 22 is combined by the meta-information obtained in the transcription process and exposed as a structured data element, by means of, for instance a web service or through clipboard functionality. The API 26 also enables the user to select highlighted fields from the interface and transmit the selected fields to selected downstream consumers.

In another embodiment, the measurement transcription device 12 includes a context-sensitive highlighting engine 28 that classifies each measurement according to its relevance to the clinical trial and its temporal status. For example, the context-sensitive highlighting engine 28 detects, in the sample sentence above, the phrase "1 x 2 cm" referring to a measurement made in the current study, and that "3 x 4 cm" referring to a measurements made in the prior study. The context-sensitive highlighting engine 28 further recognizes that certain measurements do not pertain to lesions, e.g.: "The surgical scar is 5 cm from the nipple." The output of the measurement classifier can be used to change visual presentation (e.g. color) of the measurement in the user interface. In another embodiment, the context- sensitive highlighting engine 28 utilizes a statistical methodology to recognize if a measurement descriptor denotes a relative position ("5 cm from the nipple"), a measurement on prior exam ("measures 3 x 4 cm on prior exam"), or a measurement on current exam ("measuring 1 x 2 cm on today's study"). To accomplish this, the context-sensitive highlighting engine 28 utilizes a maximum entropy classifier. The classifier has access to a series of binary features. Each feature is associated with one or more regular expressions. The regular expressions define patterns that are commonly found for reporting a relative position/prior measurement/current measurement. For instance, one of the features contains expressions for detecting current measurements. For example, the expressions including (currently *)?measur(e|es|ing) * (approximately *)? \d x \d (cm|mm). For each sentence and each measurement descriptor occurring in the sentence (if any), the binary values of each feature are computed and the resulting vector is fed to the maximum entropy classifier, which returns one of the three classes: relative position/prior/current.

In another embodiment, the measurement transcription device 12 includes a temporal correlation engine 30 that correlates the measurements from two or more current and prior radiology reports. The temporal correlation engine 30 utilizes this correlation to suggest current lesion measurements for prior measurements that can then be approved by the user. Specifically, the measurement extraction engine 20 and the measurement parsing engine 22, obtain measurements from the current report. The measurements from the prior report are then received from radiology report database 14. The measurement transcription device 12 correlates the array of current measurement descriptors with the measurements in the previous report. It is contemplated that the correlation refers to any method that includes the following steps: a pairwise comparison of all prior and current measurement descriptors, and the parsed data that has been associated with it, along with the linguistic contents in which they appear. The correlation information is then presented in the form of a feature vector to a classifier that decides if the measurements are related or not on the interface. The one or more lesions and corresponding measurement transcription data are typically stored within a lesion database 16 which is accessed by the measurement transcription device 12 through the communication network 18. The measurement transcription data including identification of the lesions, image data such as the slice of contain the lesion of interest, measurements of lesions, dates of the measurements, location and orientation of the lesions, and so on. It should be contemplated that the lesion database 16 is a proprietary database. It should also be appreciated that the lesion database 16 includes related spreadsheets or be associated with third-party software. Further, the measurement transcription data can be generated manually and/or automatically. As to the former, a display 20 of the lesion management device 12 in conjunction with one or more input devices 22 enable a user to enter measurement transcription data presented on the display 20. As to the latter, measurement transcription data is automatically generated and routed without user intervention by the measurement transcription device 12. For example, the measurement transcription device 12 automatically transcribes the radiology reports and enters the measurement transcription data.

While the measurement transcription device 12, measurement detection engine 20, measurement parsing engine 22, display module 24, application programming interface (API) 26, context-sensitive highlighting engine 28, and temporal correlation engine 30 were shown as independent components, it is to be appreciated that each of the components can be part of an measurement transcription device 12. At least some of the components of the lesion management system 10 each include at least one processor 46 executing computer executable instructions from at least one memory 52 thereof. Components include the transcription device 12, measurement detection engine 20, measurement parsing engine 22, display module 24, application programming interface (API) 26, context-sensitive highlighting engine 28, and temporal correlation engine 30. The computer executable instructions embody the functionality of the components and include the applications of the transcription device 12, measurement detection engine 20, measurement parsing engine 22, display module 24, application programming interface (API) 26, context- sensitive highlighting engine 28, and temporal correlation engine 30. Further, at least some of the components each include a communication unit 58 and/or at least one system bus 64 . A communications unit provides a corresponding processor with an interface to at least one communication network, such as the communication network 18. A system bus allows the exchange of data between sub-components of the components. Sub-components include processors, memories, sensors, display devices, communication units, and so on. As used herein, a memory includes one or more of a non-transient computer readable medium; a magnetic disk or other magnetic storage medium; an optical disk or other optical storage medium; a random access memory (RAM), read-only memory (ROM), or other electronic memory device or chip or set of operatively interconnected chips; an Internet/Intranet server from which the stored instructions may be retrieved via the Internet/Intranet or a local area network; or so forth. Further, as used herein, a processor includes one or more of a microprocessor, a microcontroller, a graphic processing unit (GPU), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), and the like; a user input device includes one or more of a mouse, a keyboard, a touch screen display, one or more buttons, one or more switches, one or more toggles, and the like; a database includes one or more memories; and a display device includes one or more of a LCD display, an LED display, a plasma display, a projection display, a touch screen display, and the like, including 3D-capable versions of these.

FIGURE 2 illustrates an interface 100 of a lesion transcription device. The interface 100 represents the situation in which a prior radiology report has been processed by the transcription system. The interface 100 includes a followed lesions sector 102, a prior radiology report summary 104, and a current radiology report summary 106. The followed lesion sector 102 includes one lesion labeled "soft tissue focus." The measurement for 01/01/2001 has already specified; typically only the largest dimension is presented. In the interface, the user selects a lesion. In this case, the first and only lesion has been selected, which is visualized by a red border of the button 108. After selection of the lesion, the interface 100 highlights the corresponding measurement 110 in the prior report (dated 01/01/2001) and the measurement candidates 112 in the current report (dated 31/23/2001) on the right. The classifier has detected that the second measurement "4.2 x 2.4 cm" is a prior measurement descriptor 114, and that "2.9 x 1.2 cm" is a current measurement descriptor 116. This is visualized in the interface by a brown versus red coloring of the measurement descriptors. When the user clicks either measurement, the largest dimension of the selected measurement is inserted in the open slot of the selected soft tissue focus field (dated 31/12/2001) in the followed lesion sector 102.

FIGURE 3 illustrates a flowchart diagram of a method for transcribing lesion measurements. In a step 200, one or more radiology reports are received from a radiology report database. In a step 202, lesion measurements from the one or more radiology reports are identified and extracted with a measurement detection engine. In a step 204, the extracted lesion measurements are parsed by measurement unit and number. In a step 206, the one or more radiology reports are displayed on an interface. In a step 208, the extracted lesions measurement are highlighted on the displayed one or more radiology reports. In a step 210, lesion measurements selected from the interface transmitted to downstream consumers. In a step 212, each lesion measurement is classified according to its relevance to the clinical trial and its temporal status with a context-sensitive highlighting engine. In a step 214, each lesion measurement is correlated from two or more radiology reports with a temporal correlation engine.

The invention has been described with reference to the preferred embodiments. Modifications and alterations may occur to others upon reading and understanding the preceding detailed description. It is intended that the invention be construed as including all such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims

CLAIMS:

1. A method for transcribing lesion measurements, the method comprising:

receiving a radiology report;

identifying one or more lesion measurement within the radiology report with a measurement detection engine;

displaying the radiology reports on an interface; and

highlighting the one or more lesion measurements on the display radiology report.

2. The method according to claim 1, further including:

selecting at least one of the highlighted one or more lesion measurements;

transmitting the selected highlighted one or more lesion measurements to at least one of a spreadsheet and a proprietary database.

3. The method according to either one of claims 1 and 2, further including:

parsing the one or more lesion measurements by a measurement unit and a number.

4. The method according to any one of claims 1-3, further including:

classifying the one or more lesion measurement according to their relevance to the clinical trial and temporal status.

5. The method according to any one of claims 1-4, further including:

correlating the one or more lesion measurements from two or more prior radiology reports.

6. The method according to any one of claims 1-5, wherein the interface displays the current radiology reports, previous radiology reports, and one or more lesions of interest.

7. A system comprising:

a display for displaying an interface; and

one or more processors programmed to perform the method of any one of claims 1-6.

8. A system for transcribing lesion measurements, the system comprising: a radiology report database which stores one or more radiology reports; and a transcription device which includes:

a display which displays an interface; and

one or more processors programmed to:

receive a radiology report;

identify one or more lesion measurement within the radiology report;

display the radiology reports on the interface; and highlight the one or more lesion measurements on the display radiology report.

9. The system according to claim 8, wherein the one or more processors are

programmed to:

select at least one of the highlighted one or more lesion measurements;

transmit the selected highlighted one or more lesion measurements to at least one of a spreadsheet and a proprietary database.

10. The system according to either one of claims 8 and 9, wherein the one or more processors are programmed to:

parse the one or more lesion measurements by a measurement unit and a number.

11. The system according to any one of claims 8-10, wherein the one or more processors are programmed to:

classify the one or more lesion measurement according to their relevance to the clinical trial and temporal status.

12. The system according to any one of claims 8-11, wherein the one or more processors are programmed to:

correlate the one or more lesion measurements from two or more prior radiology reports.

13. The system according to any one of claims 8-12, wherein the interface displays the current radiology reports, previous radiology reports, and one or more lesions of interest.

14. A system for transcribing lesion measurements, the system comprising: a radiology report database which stores one or more radiology reports;

a transcription device which includes:

a display which displays an interface; and

one or more processors which receive a radiology report;

a measurement detection engine which identifies one or more lesion measurement within the radiology report;

a display module which displays the radiology reports on the interface; and

an application programming interface (API) which highlights the one or more lesion measurements on the display radiology report.

15. The system according to claim 14, wherein the interface enables a user to select at least one of the highlighted one or more lesion measurements and the API transmits the selected highlighted one or more lesion measurements to at least one of a spreadsheet and a proprietary database.

16. The system according to either one of claims 14 and 15, wherein the transcription device further includes:

a measurement parsing engine which parses the one or more lesion measurements by a measurement unit and a number.

17. The system according to any one of claims 14-16, wherein the transcription device further includes:

a context sensitive highlighting engine which classifies the one or more lesion measurement according to their relevance to the clinical trial and temporal status.

18. The system according to any one of claims 14-17, wherein the transcription device further includes:

a temporal correlation engine which correlates the one or more lesion measurements from two or more prior radiology reports.