The ACR’s recent introduction of the “cmbX” designation in the Dose Index Registry (DIR) is a meaningful step forward. By separating combination studies, the DIR now allows for more accurate mapping to combination RPIDs—or, when necessary, assignment to RPID88 to remove them from the “unmapped” category.

But as many facilities are now discovering, this improvement comes with a new operational challenge:

cmbX mapping is extremely time-consuming.

At some facilities, the number of cmbX studies requiring review can range from a few dozen to several hundred—or more. Each one must be evaluated individually to determine the most appropriate RadLex Playbook mapping. Doing this manually is tedious, resource-intensive, and difficult to sustain.

A Practical Test: AI vs. Manual Mapping

To better understand the impact of automation, I tested an AI-assisted approach across two facilities and compared the results to a third site performing manual mapping.

Here’s what we found:

• Site 1 (AI-assisted)
  • 130 unmapped cmbX studies
  • 36 successfully mapped
  • 28% mapped
• Site 2 (AI-assisted)
  • 687 unmapped cmbX studies
  • 97 successfully mapped
  • 14% mapped
• Site 3 (Manual only)
  • 273 unmapped cmbX studies
  • 11 successfully mapped
  • 4% mapped

The Impact

Across these sites, the use of an AI-assisted workflow improved mapping performance by:

350% to 687% compared to manual mapping

This is not a marginal improvement—it’s a fundamental shift in what’s achievable.

Why This Matters

Unmapped studies in the DIR are more than just an administrative nuisance. They directly impact:

• Dose benchmarking accuracy
• Inter-facility comparisons
• Internal quality improvement efforts

Leaving large volumes of cmbX studies unmapped limits the value of your DIR participation.

At the same time, expecting staff to manually work through hundreds of studies is simply not realistic in most environments.

Where AI Fits

AI is not replacing expertise—it’s amplifying it.

By combining:

• The RadLex Playbook
• A structured set of mapping rules
• And an AI tool capable of interpreting study descriptions

…it becomes possible to rapidly identify likely RPID matches and dramatically reduce the time required per study.

The result is a workflow that is:

• Faster
• Scalable
• And far more sustainable

Bottom Line

The cmbX change in the DIR is a positive development—but it requires a new approach.

Manual mapping alone won’t keep up.

AI-assisted mapping offers a practical, immediately available solution that can significantly improve mapping completeness and help facilities realize the full value of their dose data.

---

If you’re working through cmbX mapping and looking for a more efficient approach, feel free to reach out. I’ve been helping facilities navigate DIR workflows and would be glad to share what’s working.