CT QC Copilot: Clinical Implementation of an Automated Quality-Control Decision-Support System

We describe the clinical implementation of the CT QC Copilot, an automated decision-support system for computed tomography quality control. The Copilot embodies a "system computes, physicist decides" model: it automates the end-to-end QC workflow -- from DICOM ingestion through metric computation, threshold evaluation, statistical drift detection, and root-cause diagnosis -- while preserving the qualified medical physicist's (QMP) role as the final decision-maker. The Copilot reduced per-scanner QC analysis time from an estimated 67 +/- 12 minutes (manual) to 4.2 +/- 0.8 minutes (automated computation plus physicist review), a 94% reduction that makes AAPM TG-233 trending-based QC practical for large clinical operations. Western Electric rule-based drift detection identified gradual degradation 3-6 months before threshold exceedance in 4 of 30 simulated scanners, demonstrating the early-warning value of statistical process control. All nine ACR-aligned metrics showed agreement with manufacturer console values within 1.2 HU for CT number metrics and within 0.10 mm for geometric metrics.