Quality control in radiology is the test schedule: phantoms, daily / weekly / monthly / annual checks, reject analysis, and the ACR / FDA / Joint Commission rules that make those tests non-optional. It is not the QA program (pillars, policies, training). It is not peer review of the report. It is not device IQ/OQ/PQ.
If you meant the QA program (pillars / policies / training) → medical imaging quality assurance. If you meant peer review / report QA → quality assurance in radiology. If you meant device IQ/OQ/PQ → medical device validation process. If you meant device QMS → medical device quality management system.
If the work is overlaying a QC mark or running an imaging model inside a department QA/QC stack, PYCAD is the imaging piece (viewer / model), not a QA/QC vendor.
What QC is
Consistency and accuracy. Monday’s CT and Friday’s CT, same machine or not, have to be the same diagnostic object. AAPM’s 1977 diagnostic-radiology QC protocol is the ancestor of the modern schedule; digital detectors and MRI coils changed the tests, not the idea.
- Acceptance testing. New or repaired kit does not see a patient until the baseline is written down.
- Routine monitoring. Daily, weekly, monthly — the cadence that catches drift.
- Error correction. Out of limits → find it, fix it, re-test. A red number with no re-test is a log, not QC.
Who does the work
- Medical physicist. Designs the program, runs acceptance and the annual deep eval, sets the pass/fail numbers.
- Technologist. Daily warm-up and the phantom. They see Tuesday’s grain before anyone else.
- Radiologist. Artifact and “this study is not diagnostic” feedback. That loop is how a test that passed on the phantom still gets a service call.
- Service engineer. The repair. Physicist and tech name the fault; the engineer closes it.
A clinically qualified physicist is scarce in a lot of systems. That is a staffing constraint, not a reason to skip the annual. It is also not a device-QMS article — that is 7869.
The schedule, by modality
| Modality | Daily | Weekly | Monthly | Annual |
|---|---|---|---|---|
| CT | Air/water phantom, monitor, room | Noise, uniformity, CT number | Hardcopy / monitor calibration | Physicist: dose, resolution, slice thickness, lasers |
| DR / CR | Warm-up, detector calibration, plate hygiene | Uniformity, artifacts, monitor | Cassette / erase-cycle check | Physicist: exposure index, resolution, dose, collimation |
| Mammography | Monitor, phantom acquisition (processor QC if film remains) | Phantom score: fibers, specks, masses | Compression, visual checklist | Physicist: dose, phantom score, AEC, collimation |
| MRI | Start-up, table, cryogen level | Center frequency, SNR, uniformity | Geometry, slice thickness | Physicist: homogeneity, RF coils, image quality |
SNR and spatial resolution as concepts are a chapter on 517. This table is when you run them and who signs.
A typical light-field / x-ray-field misalignment trigger is about 1–2% of SID. That is a service call, not a note in a drawer. Annual beam-quality and alignment are how you prove dose and collimation still match the baseline.
Who writes the rules
| Body | What they own | Cadence you will actually feel |
|---|---|---|
| FDA | Equipment performance standards; MQSA for mammography | MQSA inspection is annual |
| ACR | Modality accreditation: phantom + clinical images, physicist survey | Renewal about every 3 years |
| Joint Commission | Hospital quality program, equipment eval, radiation safety, competency | Unannounced survey on a ~36–39 month cycle |
| State health | Registration, shielding, personnel dosimetry, local test frequencies | Annual to every few years, by state |
ISO 9001 is a documentation habit some sites borrow. It is not a medical-device QMS how-to. Device QMS is 699 7869. Device IQ/OQ/PQ is 699 5503.
Reject analysis and a program that is used
A QC program that exists only in the binder fails the first busy Tuesday. Three things make it real:
- Training that says why. The daily phantom is how you avoid a week of grainy chests, not a checkbox.
- A log people will fill in. Digital, short, reviewed. A binder nobody opens is not documentation.
- Reject / repeat rates. Why the study died (positioning, motion, exposure, equipment). A rate with no reason is a vanity metric. Uptime and radiologist “not diagnostic” flags sit next to it.
AI that watches SNR drift and books a tube change is predictive maintenance. Useful. Not a second 517 article, and not the radiology-report-QA market that 561 owns. Do not dress QC in that costume.
FAQ
Is QC the same as QA?
No. QA is the program. QC is the test. The split is written on medical imaging quality assurance.
Is this peer review?
No. Peer review / discrepancy / TAT is quality assurance in radiology.
Does PYCAD sell a QC suite?
No. The imaging piece, not a phantom suite. See the line above.