Speech Recognition and External Reporting Systems

Speech recognition (SR) converts spoken language into edit-able text without manual transcription. Instead of using a keyboard to type text into an editor package, the user talks and the words appear in the text area. The converted text has the same properties as text entered manually. Optional embedded SR support is provided, in which the SR tools are integrated into the viewer's basic reporting function. For users running an integrated reporting application on the workstation, an optional interface is included to use as an alternative to its embedded solution.

Nuance Dragon Naturally Speaking

The Embedded Dragon Naturally Speaking package is provided as an optional plug-in module. It requires a plug-in module license and a licensed copy of Dragon Medical Practice Edition 4 installed on each workstation.

Workstation requirements are as follows:

The embedded SR package is intended to be used with a Philips SpeechMike. Other microphones are supported, but the integrated controls are only available on the SpeechMike.

To configure the viewer to use embedded SR:

  1. Open the Tools/Options
  2. Select the Dictation/Speech Recognition/Speech Recognition Tools panel
  3. Select Nuance Dragon from dropdown
  4. Click OK

 

When using SR, it is possible to retain a copy of the WAV file created when dictating the report which is unnecessary if the user is responsible for editing the resulting text. However, if the results of the SR processing need editing by a third party, such as by a transcriptionist, save the original recording for validation. Check the box labeled Auto record WAV when dictating in the Settings panel under the Dictation/General Settings panel to save the WAV file.

eRAD PACS has integrated a set of voice commands specific to the features available in the viewer. These commands must be downloaded from the eRAD PACS server and uploaded into the Dragon vocabulary.

The available viewer specific commands are given in the following table:

Voice Command Description Comments
approve Approve a report and set state to Final  
bold font Set Selected text to bold font Applies to selected text only
close study Close the current study  
close views Close the image views  
compact report panel Display the compact report panel  
customize settings Open Customize Settings window  
end dictation End the dictation and disable the microphone  
end study End the current viewer session  
exit from viewer Close the study and the viewer  
fit image Set image zoom to fit-to-window Requires a selected image frame
focus to addendum | impression | note | observation Set cursor focus to the report’s Addendum, Impression, Note or Observation field Available with the full report panel is open
full report Display the full report panel  
full screen Display selected image in Full Screen mode Requires a selected image frame
help contents Pop up the Help window  
hide report panel Close the compact or full report panel  
invert image Invert the selected image’s greyscale setting Requires a selected image frame
italic font Set selected text to italics font Applies to selected text only
layout manager Display the layout manager  
next frame Scroll to the next image Requires a selected image frame
OK Select the OK button Available when the close prompt notice is displayed
open next study Open the next study on the worklist  
open previous study Open the previous study on the worklist  
overlays Show/hide overlays  
play cine Start recursive cine mode Requires a selected image frame
previous frame Scroll to the previous image Requires a selected image frame
record Start recording dictation  
record stop Stop recording dictation  
report templates Set focus to the report templates toolbar Available if the full report panel is open and it has the report templates toolbar displayed
reset image size | orientation | position Reset the selected image’s size, orientation, or position Requires a selected image frame
reset window level Reset the selected image’s window/level setting Requires a selected image frame
send report Send the report to the server  
set state read | dictated | preliminary | final

Set the study state to Read, Dictated, Preliminary or Final

 
show hash marks Display hash marks on all images  
show localizers Display localizer lines on all images  
show report panel Display the default report panel  
stop cine Stop recursive cine mode Requires a selected image frame
underline font Set the selected text to underline font Applies to selected text only

 

To download the commands:

  1. Open the Tools/Options
  2. Select the Dictation/Speech Recognition/Speech Recognition Tools panel
  3. Select Nuance Dragon from dropdown
  4. Select Download voice commands
  5. At the prompt, select OK
  6. Save under desired folder or on Desktop
  7. After the file transfer completes, open Nuance Dragon
  8. Find the folder where command file is saved and select Open
  9. Click Yes to validate the command file

Integrated Reporting Systems

Many speech recognition and reporting systems provide an interface to pass control between a PACS viewer and the client application. The interface consists of an XML file written to the local file system by the viewer and picked up by the speech recognition system. The XML file contains study details, user IDs and other information.  

 External reporting systems configured into the system by default include the following:

The list of external reporting systems is configurable. If the system is not displayed, the system administrator must add.

To configure the viewer to use an external reporting system as the default reporting solution, the eRAD PACS XML Interface Speech Recognition plug-in module must be licensed, the user must have permissions to load the plug-in module, and the report application’s client plus interface module, if necessary, must be installed and running on the workstation.

Set the dictation mode as follows:

  1. Open the Tools/Options
  2. Select the External Reporting panel
  3. Select the  built-in or pre-configured reporting systems from dropdown
  4. Expand the Process Upon menu and select Open or Record
  5. Enter the output file in the text field
  6. Click OK

The interface allows for creating an XML file when the viewer session opens in addition to when starting to record a report. Multiple Open and Record events can be assigned to integrated report system engines. Additionally, both events allow creating a second XML file when an end event occurs. End events happen only if the corresponding on event occurred.

Event On Event Off Event
Open Viewer session initiated or restored from a bookmarked state Study is closed or bookmarked, or the viewer session is terminated
Record User presses the Record button for the first time after the viewer session started or is restored from a bookmarked state The study is closed or bookmarked or the viewer session is terminated. Occurs only if the On Event occurred

 

The XML file created for a trigger event is defined in the configuration files specified in the Speech Recognition Engine Configuration Manual. The record event occurs by clicking the Record button in the EP viewer’s Audio toolbar after each trigger event in which study details are written to the configured XML file. The reporting system detects the file, reads its contents, removes the XML file and then takes over the reporting process. Consult the reporting system’s user manual for information on its operation.

Mouse and keyboard control generally remain with the eRAD PACS viewer, unless reporting system’s user interface is displayed and given the mouse and keyboard focus. The report created by the external reporting system does not appear in the viewer’s report panel immediately. The system sends the report to eRAD PACS server directly or through the RIS. The information written to the XML file passed from eRAD PACS to the external reporting system is configurable. Refer to the eRAD PACS Speech Recognition Configuration Manual for details.