Overview

The Dose Index Registry (DIR) has introduced an artificial intelligence (AI) algorithm to reduce the effort for site personnel to map their exams to a standard lexicon (naming convention) and increase mapping accuracy. Recognizing that the mapping process is the most labor-intensive aspect of participating in the DIR, the AI mapping tool will support site personnel in accurately mapping their exams -- a critically important activity that enables the registry to aggregate data correctly for each site and across the registry to ensure meaningful feedback reports.


Hear from the DIR chair, A. Kyle Jones about the AI Mapping Tool.


Phase One

Sites participating in the DIR Fluoroscopy (DIR Fluoro) can now use the AI tool for exam mapping for the first phase of the AI mapping initiative. DIR Fluoro uses ACR Common as the universal exam lexicon and the AI tool will suggest an exam mapping to an ACR Common exam ID based on prior mappings completed across all DIR. Personnel who perform exam mapping will be able to accept the AI tool’s suggested mapping or choose their own. Over time the mapping algorithm will update and ‘learn’ as more exams are mapped. Learn more about using the AI tool here.


Future Phases

If the AI tool performs well, the intention is for the AI algorithm to automatically map exams for sites and provide facility staff with the option to accept or decline the automated mapping. The second phase of implementing the AI-assisted mapping tool will be rolled out when the pilot Digital Radiography (DIR DR) and Nuclear Medicine (DIR NM) modules become broadly available. Similar to DIR Fluoro, these modules are integrating the use of Common IDs for exam mapping.  


The third phase of the AI mapping initiative will extend the use of the AI tool to the DIR CT module. DIR CT currently uses Radlex Playbook IDs (RPIDs) for mapping site-specific exam names to a universal exam lexicon. However, because the ACR Common IDs allow more flexibility in registry functionality, our first step in implementing the AI tool for DIR CT will be to convert the CT exam mapping from RPIDs to ACR Common IDs. This process is already underway. Once the conversion from RPID to ACR Common is complete, a CT AI algorithm will be developed and implemented in a similar process as is now available for DIR Fluoro.


As this initiative moves forward, the NRDR team and DIR committee volunteers will provide educational resources to keep you up to date about the changes and how they affect your DIR participation.