12.2 Quality Control

Quality Assurance vs Quality Control

Quality assurance (QA) is the broad management program that ensures optimal images at the lowest reasonable dose; quality control (QC) is the concrete set of measurements performed on equipment to verify it meets tolerances. QC has three tiers: non-invasive (technologist checks), intermediate (physicist checks), and invasive (service engineering). The ARRT exam expects you to know which QC test detects which problem, the approximate tolerance, and how often it is done. Memorizing a handful of numeric tolerances pays off directly.

Generator and Beam Tests

kVp accuracy verifies that the beam quality the generator produces matches the selected value. Measured kVp must be within about ±5% (roughly ±4-5 kVp) of the set value; an inaccurate kVp shifts contrast and dose. Timer accuracy checks that exposure time is correct — critical for single-phase units where a spinning-top test historically counted dots. Exposure reproducibility checks that repeating the same technique gives the same output: repeated identical exposures must vary by no more than 5% in mR/mAs (coefficient of variation ≤ 0.05). Exposure linearity checks that output tracks proportionally as you change mA/time stations: the mR/mAs between adjacent stations must agree within 10% (linearity coefficient ≤ 0.10).

Half-value layer (HVL) measures beam filtration/quality — the thickness of aluminum that halves the beam intensity. A minimum HVL (for example, ≥ 2.3 mm Al at 80 kVp, with 2.5 mm Al total filtration required on units operating above 70 kVp) confirms low-energy photons that only add skin dose are removed. Too low an HVL means inadequate filtration and excess patient dose.

AEC, Focal Spot, and Alignment

Automatic exposure control (AEC) consistency verifies that the AEC terminates at the correct receptor exposure across changes in kVp, part thickness, and field selection, keeping receptor exposure and the exposure index stable. AEC reproducibility should also fall within the ±5% band. Focal-spot size and resolution are checked with a line-pair resolution test tool, star pattern, slit camera, or pinhole; results are reported in line pairs per millimeter (lp/mm). Measured focal-spot size is allowed to run somewhat larger than the nominal size (NEMA tolerances permit up to roughly 50% oversize for small spots), and blurring from focal-spot bloom lowers spatial resolution. Collimator (beam-restriction) and beam-alignment testing uses a collimator/beam-alignment test tool: the light field and radiation field must coincide within 2% of the SID, and the central ray must be perpendicular within about 1 degree (the two steel balls of the beam-alignment cylinder superimpose when the CR is truly perpendicular). Positive beam limitation (PBL) is held to the same 2%-of-SID standard.

Image-receptor uniformity (flat-field / gain calibration for DR, or plate sensitivity checks for CR) confirms a uniform exposure produces a uniform image with no shading, banding, or dead-pixel lines. Monitor and display QC is covered in Section 12.3.

QC Test → Tolerance → Frequency

Repeat-Analysis Programs

A repeat (reject) analysis program tracks every discarded image and its reason. It is the QC feedback loop that turns error data into training and equipment fixes. Programs usually target a repeat rate under about 5-8%, and the single most common cause is positioning error (rotation, centering, clipped anatomy), followed by exposure/technique errors and motion. Because digital systems hide overexposure, a repeat program must also flag dose creep using exposure-index/deviation-index trends. A worked example: if a department exposes 5,000 images in a month and repeats 350, the repeat rate is 350 ÷ 5,000 = 7% — within range but trending high, prompting a positioning refresher.

QC Program Structure and a Linearity Example

QC duties are tiered: radiographers run daily/weekly checks (plate erasure, artifact review, marker and annotation audits), a medical physicist performs annual measurements (kVp, HVL, linearity, AEC), and service engineers handle invasive repairs. A single-phase timer can still be verified with a spinning-top test, in which each dot represents one useful pulse. A worked linearity example: if 10 mAs yields 30 mR (3.0 mR/mAs) and the adjacent 20 mAs station yields 66 mR (3.3 mR/mAs), the difference is (3.3 − 3.0) ÷ 3.15 ≈ 9.5%, just within the 10% limit — the generator passes. If 20 mAs instead read 75 mR (3.75 mR/mAs), the ≈22% difference would fail linearity and flag a generator or timer fault. Scheduling these numeric checks is what keeps receptor exposure — and the exposure index — stable across the equipment fleet.

Common exam trap: reproducibility and linearity are easy to confuse. Reproducibility = same technique, same output (≤5%); linearity = adjacent mA stations track proportionally (≤10%). If a question stresses repeating identical settings, the answer is 5%; if it stresses changing mAs stations, the answer is 10%.