Periodic Quality Assurance: ACR Phantom Testing & Geometry Verification

Periodic testing extends the daily constancy check

Daily water measurements are useful surveillance, but they do not measure all aspects of CT performance. Periodic testing evaluates CT-number accuracy across materials, low-contrast detectability, spatial resolution, slice thickness, uniformity, geometry and dose. Acceptance testing establishes performance when a scanner is installed; testing after major repairs addresses the affected systems. A qualified medical physicist conducts the annual evaluation and helps define technologist tests and corrective actions.

The ACR requires an annual physicist evaluation, with up to 14 months allowed between surveys. Accreditation criteria are professional program requirements; they are not automatically federal radiation laws. X-ray machine regulation depends on the applicable state and other jurisdictional requirements. Keep the distinction between a manufacturer's specification, a facility action limit and an accreditation scoring criterion clear.

The accreditation phantom's four modules

The ACR CT phantom contains different modules for different tasks. The first supports material CT numbers and slice-thickness evaluation. The second assesses low-contrast performance. The third assesses uniformity and related geometry. The fourth contains high-contrast resolution patterns. Select the correct slice and reconstruction for each task; a measurement through the wrong insert cannot be rescued by a precise ROI.

For the ACR accreditation CT-number assessment using the specified adult abdomen conditions at 120 or 130 kVp, the published material intervals are:

These values describe the named phantom materials and testing conditions. They do not establish immutable HU values for every patient's bone or every voltage. Bone and other non-water materials have energy-dependent attenuation. Verify that the test kVp, reconstruction and phantom are the required ones before comparing a result with this table.

Low-contrast performance is a signal-and-noise task

Low-contrast objects can disappear despite adequate high-contrast spatial resolution. The current ACR accreditation method calculates contrast-to-noise ratio (CNR) using the difference between the low-contrast rod mean and background mean divided by background SD. Use the absolute difference, the prescribed rod and correctly placed ROIs. The published requirements are greater than 1.0 for adult head and adult abdomen, greater than 0.7 for pediatric head, and greater than 0.4 for pediatric abdomen.

For an illustrative adult abdomen measurement, suppose the rod mean is 8 HU, the background mean is 3 HU, and background SD is 4 HU. CNR = |8 − 3|/4 = 1.25, which exceeds 1.0. If the same contrast difference has an SD of 6 HU, CNR becomes 0.83 and fails that criterion. This does not create a universal clinical lesion-detection threshold: phantom size, task, reconstruction and observer behavior differ from clinical imaging.

Uniformity and spatial resolution

For the accreditation uniformity assessment, compare peripheral ROIs with the center ROI in the prescribed adult abdomen image. Differences within 5 HU satisfy the uniformity criterion; differences between 5 and 7 HU represent a minor deficiency and values above 7 HU a major deficiency. Inspect for rings and other artifacts as well. A set of favorable ROI means cannot prove the absence of a narrow artifact elsewhere in the image.

High-contrast patterns or a measured modulation transfer function assess spatial resolution. A sharper kernel may improve edge response while increasing noise. A small display field of view reduces pixel size, but it cannot overcome the detector aperture, focal spot, reconstruction method or acquired slice sensitivity profile. Do not invent a universal requirement that every adult body examination must resolve a particular line-pair group. Follow the specific test's scoring method and compare scanner performance with its baseline and intended clinical task.

Geometry, thickness and mechanical checks

Test reconstructed slice thickness using the prescribed ramp or other test object and analysis. Verify table travel, positioning lights, reconstructed distances and display performance using approved methods. A couch can move the requested distance accurately yet be poorly aligned with the imaging plane; these are related but distinct tests. Phantom centering, table load and setup should be recorded when they influence the result.

For a simple distance example, a known 100 mm phantom spacing measures 102 mm. The relative error is (102 − 100)/100 × 100 = 2%. That is a measured error, not a universal acceptance limit. The physicist compares it with the applicable specification and determines whether calibration or service is needed.

Dose and protocol review

Measure CT dose indices with the appropriate phantom and instrumentation. Identify whether a displayed value uses a 16 cm or 32 cm reference phantom before comparing it with a measurement. Review representative clinical protocols for scan range, phase count and image quality as well as dose. Reducing displayed dose while producing nondiagnostic images is not optimization.

For accreditation phantom submissions, use the required fixed technique representing the clinical protocol, with tube-current modulation disabled as directed; reconstruction methods used clinically may be included according to the instructions. Document equipment, settings, results, deficiencies, corrective actions and follow-up. Periodic QA should connect measurement to a specific decision about performance rather than produce a collection of unexplained numbers.

References: ACR phantom testing instructions, ACR CT QC requirements, ACR CT dosimetry.