Purpose: 
Accurately measuring interpretation time of head CTs is an important quality metric in acute stroke care where rapid administration of thrombolytics may be necessary and delays in CT interpretation may prolong the time to drug delivery. While the times of preliminary or final report generation are readily available, they do not capture the time of actual result communication. Typically, communication is documented in the report, but is not queryable.

An automated method to extract the date and time of communication stated in radiology reports was developed. The success rate and impact on measured time of interpretation was calculated. In addition, utility for detecting errors in communication statements was explored. 
Materials and Methods: 
1,656 reports from head CTs acquired during a "code stroke" in a teaching hospital over the course of 2 years were used. A python script was developed to extract the communication statement from the reports using various communication signal phrases. Preprocessing to optimize date/time detection was performed and the subsequent string was passed to dateutil, an open source python package which can parse most date/time formats. The extracted communication date/time was manually compared to the report to evaluate accuracy and determine error types. 
Results: 
Communication correctly detected in 1244 (75.1%). No communication correctly detected in 327 (19.7%). Error present in 85 (5.1%). Of cases with errors, 4 were related to the inability of the algorithm to detect nonstandard communication statements and 81 were because of erroneous or incomplete communication statements in the report, including wrong date (35), time only (21), typo (13), AM/PM mismatch (10) and blank template (2).

Average time from study completion to communication was 9.3 min, to prelim report was 20.5 min, and to final report was 87 min. Of reports with a communication statement, the communication time was earlier than final or prelim time for 1213 studies (97.5%). 
Conclusions: 
This method successfully extracts communication date/time from CT reports. Compared to preliminary or final report time stamps, this more accurately represents the true time of information delivery to the stroke team. Interpretation time based on communication time is typically quicker than times based on prelim or final time stamps. 

Additionally, this method could be used to automatically detect errors in communication statements and monitor compliance with department policies regarding communication in acute stroke.

Automated Extraction of Date and Time from Communication Statements in Free Text Radiology Reports  for Use in Code Stroke Head CT Quality Control

[Virtual Conference Page](https://event.vconferenceonline.com/microsite/html/login.aspx?id=1551&rlp=1)

[ASNR Website] (https://www.asnr.org//annualmeeting)

[Educational Exhibits](https://www.morressier.com/event/asnr2020exhibit/5e5d0be000701ceba4a26941)

Automated Extraction of Date and Time from Communication Statements in Free Text Radiology Reports for Use in Code Stroke Head CT Quality Control

Automated Extraction of Date and Time from Communication Statements in Free Text Radiology Reports for Use in Code Stroke Head CT Quality Control