|    |
MagView Transcription Extractor™
A Transcription Interface for the Collection of ACR BI-RADS™ Data
Introduction
There are many benefits associated with collecting and storing ACR BI-RADS mammography findings and outcome data. Collecting this information offers an insight into the quality and efficiency of your mammography practice. Once collected, this information can also serve as a platform for many of the operational needs of your group. Patient notification and recall letters may be generated, patient tracking may be performed and compliance with the requirements of the FDA's MQSA can be achieved.
Collecting this information accurately and efficiently is the key to successful ACR BI-RADS implementation. Several software systems allow the collection of this information, but require the radiologist to spend additional time in specifying each BI-RADS code for the study. If the radiologist is also dictating the report, this information is essentially collected redundantly; once on a check-off sheet or in a computer screen, and once as the report is being dictated. Furthermore, manual collection methods permit inconsistencies based on differences in radiologists' interpretation of BI-RADS codes, and introduce the possibility of data collection and data entry errors.
The MagView Transcription Extractor interface allows you to collect ACR BI-RADS codes directly from the dictated and transcribed report, without the need for secondary data collection procedures. This product offers a unique option for the radiologist who continues to dictate mammography reports, and does not wish to use check-off forms or computer programs to collect and store the necessary study data.
Data Extraction
The collection of BI-RADS and other data is achieved through the extraction of words, phrases and symbols from the transcribed reports. For example, if reference is made to a particular finding or finding modifier within the body of a transcribed report, the MagView extractor will identify the presence of this word or phrase and store the appropriate code. Once identified as objective information, this data populates the MagView database, and may be used as if it had been directly entered.
The MagView Extractor includes several sophisticated techniques that ensure that data is collected with the highest possible level of accuracy. These include the following:
- An exhaustive cross reference database of words and phrases allows accurate code recognition. A single code may be identified from many different possible words or phrases. For example, existence of a primary mass finding may be identified by the presence of one of many different words or phrases, including "mass", "density", "node", or "nodule". The Extractor also correctly identifies the word based on its singular or plural form, and identifies surrounding words which would more specifically identify the type of mass. This means that the recognition of words is independent of the dictation style of the radiologist, and any variance that may exist from the BI-RADS lexicon terminology.
- The recognition of words and phrases is sensitive to the context in which they are presented. This means that the same word or phrase may indicate a different code depending on surrounding words and phrases. For example, the word "round" in a report could lead to an interpretation of a specific calcification finding or mass finding, depending on surrounding words and phrases. This ensures that the appropriate code is recognized and stored.
- Code validation rules are applied to the data once it has been extracted from the report. In cases where conflicting codes are identified in the extracted data, the conflict is resolved based on stored logic. In cases where key information has not been identified, codes representing generic data is stored. This ensures that all quality assurance and reporting features of MagView will not be affected by conflicting or missing data. If reports are transcribed directly into MagView, code validation messages are available to the transcriptionist to help identify and resolve such issues.
- The code extraction algorithm and cross reference database is user definable, and self adapting. If additional cross-reference words and phrases are desired, these may be added by the user to the cross reference database. If alternative logic for the relationship of words and phrases to codes is required, this also may be modified by the user. Furthermore, without any configuration by the user, the Extractor will automatically adjust the sensitivity of word recognition as reports are extracted.
- Data pointers are used to ensure the accuracy of the most important of the extracted data. These pointers may be included in the transcribed report to specifically identify the BI-RADS assessment and recommendations. For example, the text "BI-RADS ASSESSMENT:" may appear in all mammography reports followed by a code or phrase identifying the BI-RADS assessment. This would allow the MagView Extractor to reliably recognize the correct assessment and recommendations. Because this type of notice is commonly used in reports for other reasons, it usually requires no additional steps on the part of the radiologist or transcriptionist to implement. Other data pointers would be used for patient identifying information and other basic data collected by MagView.
Conclusion
The MagView Extractor offers a unique opportunity to implement the ACR BI-RADS Lexicon and take advantage of the many benefits of an automated mammography information system, without requiring any change or disruption in the way the radiologist works.
This computer software incorporates the Breast Imaging and Data Systems (BI-RADS) of the American College of Radiology, 1992. The American College of Radiology is not otherwise responsible for the contents or operation of this software, and disclaims any and all warranties, express or implied, in connection therewith.
 
|
     |