Next Generation Report
Radiologist create value by generating reports. These reports are the primary communication of our work and take large amounts of time and attention. New technology and software can make our reporting more efficient and accurate and move our communications from simple lines of text to documents that include pictures and links to improve the value of Radiologist’s interpretations.
Powerscribe 1 implementation (Next generation dictation).
A series of upgrades are occurring which will be invisible to the radiologists but will improve the back end of the dictation system, including moving the processing of speech to the cloud. This is in preparation for a significant version upgrade which will hopefully be available in the second half of 2022. The next version of power scribe will have a completely different user interface and multiple new features including Autosuggestion of recommendations. Before this version goes live, we will be having workshops and creating educational materials to maximize use of the new features. .
Radiology is an image-based specialty, yet our primary mode of communication with patients and referring providers is through text. Now that radiology is completely digital, including significant images, tables and charts (ex: tumor tracking) is feasible and will likely improve physician and patient satisfaction. Building a multimedia report has been so far a protracted process because of software limitations we will be addressing with Modlink and PS1.
There is an abundance of templates for all subspecialty divisions within the department. This makes it difficult to structure any data within the reports, to ensure complete reporting as required for reimbursement, and to provide consistency to the referring physicians. Templates will be streamlined across the board with an emphasis on including disease specific common data elements where pertinent making reports richer in content and complying with coding and billing requirements.
Radiology language (“T2 hyperintensity”, “hyperechoic nodule”, “Kerley B-lines”, etc.) can be confusing and even anxiety-producing to patients. Using advanced NLP techniques, complex radiology reports can be simplified and summarized to the patient level. These reports can be automatically generated with final edit/approval by the interpreting radiologist. We will likely trial a start-up commercial product called Scanslated to further this goal.
Auto data population
Importing data that is available in other systems into radiology reports improves efficiency and accuracy. Several 3rd party apps along with improvements in our PACS and EMR interfaces are being evaluated with the motto that “anything that can be auto filled should be”. We are currently using software called Modlink to autoimport data from DEXA scans and CT dose data, and we will be expanding this effort to include Ultrasound. We have completed a project to autofill comparison dates from Visage and are working on importing measurements. We are continuing work on sending data between Epic and Powerscribe, and will be working with new AI vendors to autoimport the findings into PS as well.
As AI language processing capabilities advance, they can eventually add value for radiologists by directly assisting report creation. We will be testing an application that can suggest an impression based on your findings in real time as you dictate the report. This product from RadAi may speed report creation, improve accuracy, and reduce cognitive load.