Quality Control and Data Management
Daily and longitudinal QC
Scan the manufacturer's phantom daily, before patients, and plot results on a control chart. Investigate values beyond ±2 SD or a sustained trend. Maintain logs of calibration, service, and software upgrades.
Precision and LSC
Precision is the reproducibility of repeated measurements, expressed as coefficient of variation (CV%). ISCD recommends an in vivo precision assessment - commonly 30 patients, or 15 patients with duplicate scans and repositioning. Calculate the root-mean-square SD, then LSC = 2.77 × CV at 95% confidence. Only a change larger than the LSC is considered real.
Worked example: CV = 1.2%, so LSC = 2.77 × 1.2% = 3.3%. A 2% drop in spine BMD is within measurement error and should not be called bone loss.
Data management
Use the same scanner, software version, and reference database for follow-up. Cross-calibrate when equipment changes. Store DICOM images and reports according to HIPAA and facility retention rules, and back up databases. A complete report includes demographics, indication, sites measured, BMD, T- and Z-scores, comparison with prior studies, and an impression.
Traps
Using phantom (in vitro) CV to calculate LSC, comparing scans from different machines without cross-calibration, and ignoring a drifting control chart.
Rule of thumb: if the change is smaller than the site-specific LSC, report it as stable.