Purpose
The purpose of this article is to provide an overview of the AI assisted Exam Name Mapping tool. The AI assisted Exam Name Mapping tool was designed to automatically propose suggested mappings for exams in the Dose Index Registry (DIR). The first iteration of this model was created for Fluoroscopy. Models for Digital Radiography, Nuclear Medicine and Computed Tomography will be developed in the future.
Features of the AI tool
•System automatically indicates exams which do not have Suggested Mapping value yet populated and with nightly process it populates Suggested Mapping value for them.
•AI algorithm provides Confidence Level (%) together with Suggested Mapping value
Note: AI algorithm can only suggest mapping for Fluoro modality exams at this point.
What the User Can Do
•Once Suggested Mapping value populated user can accept it or user can decline it and continue using their own mapping
To accept suggested mapping:
To decline suggested mapping:
•User can submit to ACR their feedback about Suggested Mapping accuracy and propose improvements
•User can get or update Suggested Mapping value for exam record
To get suggested mapping value:
To update a suggested mapping value:
Note: The AI algorithm will only return the same result for the same input procedure name, until AI algorithm is upgraded with the new version and logic of searching the best matching mapping is adjusted. Then returned Suggested Mapping value can be different comparing to what was provided before.
What the System Can Do
•Based on Confidence Level, if it is >50%, System can accept Suggested Mapping automatically
Alert: