Every activity a department administers passes through one instrument, and a slow drift in it is invisible on the images. The four quality control tests, what each one catches, and how to read a constancy log before an annual check catches you out.
Every activity a nuclear medicine department administers passes through one instrument, and that instrument is the only thing standing between the prescription and what actually goes into the patient. A gamma camera that misbehaves shows you a bad image. A dose calibrator that has drifted by seven per cent shows you a perfectly reasonable number, on every patient, until somebody measures it against a certified source.
That is the entire argument for a quality control programme on the chamber. It is not paperwork for the regulator. It is the only way the department finds out.
A dose calibrator is a sealed, pressurised, re-entrant ionisation chamber with a well down the middle. Activity placed in the well ionises the gas, the electrometer measures the resulting current, and the display converts that current into becquerels using a calibration factor. The factor is different for every radionuclide, because different emissions ionise the gas differently, and it is selected by the dial setting or the nuclide key.
Two consequences fall out of that description, and both cause real errors in real departments. The instrument does not identify what is in the well, so a technetium eluate assayed on an iodine setting returns a confident and completely wrong number. And the geometry of the source matters, because the chamber responds to where in the well the activity sits and to what is absorbing photons on the way out. A dose in a 1 ml syringe and the same dose in a 10 ml vial are not the same measurement.
A complete programme is four tests plus a background reading, and they are not interchangeable. Each detects a different failure, and a department that runs constancy daily and nothing else is protected against exactly one of them.
| Test | How often | What it detects | What a failure usually means |
|---|---|---|---|
| Background | Every day, before the first assay | Contamination in the well or liner, and a source left too close to the chamber | Something has been spilled in the liner, or the generator has been moved nearer the calibrator |
| Constancy | Every day of use, on the settings you use | Drift in the chamber, the electrometer or the pressure seal | A trend rather than an event. Read the log, not the single result |
| Accuracy | At acceptance, after repair, and annually | Error against a traceably certified reference source | The calibration factors or the chamber are out, and everything measured since the last good check is out with them |
| Linearity | At acceptance, after repair, and annually | Non-linear response across the activity range actually in use | Saturation at the top of the range, or noise and background at the bottom |
| Geometry | At acceptance, and whenever a new container or volume enters routine use | Dependence on volume, container material and position in the well | Doses assayed in one container and administered from another are reading differently |
Constancy is done with a long lived sealed reference source, usually caesium 137 or cobalt 57, measured on the same settings in the same position every morning. The point is not that today's reading is inside tolerance. The point is what the readings have been doing for the last six months.
Record the measured value, the decay corrected expected value and the ratio between them. Then plot the ratio. A chamber losing seal pressure does not fail one morning: it drifts by a fraction of a per cent each month for a year and then fails an annual accuracy test that arrives as a complete surprise. A ratio plotted on graph paper on the hot lab wall catches that in month three. A column of numbers in a register that nobody adds up does not.
Two practical points. Caesium 137 has a half-life of about thirty years, so it decays imperceptibly across a working life and makes a stable reference. Cobalt 57 has a half-life of about 272 days, so its activity falls to a level where the measurement becomes noisy within two or three years, and the source needs replacing on a budget line departments routinely forget. When constancy readings start scattering, check the age of the source before you suspect the chamber. Consumables of this kind are covered in the annual running cost of a nuclear medicine department.
Linearity asks whether the chamber reads correctly at the top and the bottom of the activity range, not only in the middle. The classic method uses decay. Assay a high activity technetium source, usually a fresh generator elution, repeatedly over two or three days until it has fallen to the lowest activity you would ever measure, and compare each reading against the value predicted by the half-life. The test takes a weekend of intermittent measurements and produces a curve that shows exactly where the instrument stops behaving.
Shield or sleeve sets do the same job in an afternoon by attenuating a single source in calibrated steps. They are quicker and they are perfectly acceptable, provided the set itself has been validated against a decay measurement at least once, because a sleeve set that has been dropped or has a damaged insert quietly writes its own error into every subsequent test.
The detail that matters more than the method is the range. Test the range you use. A department assaying diagnostic technetium doses has one range. A department that also assays radioiodine therapy activities has a top end several orders of magnitude higher, and a chamber that is linear across diagnostic activities can saturate well below a therapy activity. If therapy is anywhere in your plan, the linearity test has to reach up to it before the first therapy patient rather than after. The facility side of that is in thyroid uptake, imaging and radioiodine therapy.
Geometry is the response of the chamber to the shape, volume, material and position of whatever is in the well. It is the test most often skipped and the one that most often explains a discrepancy nobody can account for.
The practical form is straightforward. Take a fixed activity, measure it in the container and volume you treat as your reference, then transfer or dilute it into every other container and volume the department uses in routine work: the elution vial, the kit vial, a 1 ml syringe, a 5 ml syringe, a 20 ml syringe, a plastic vial and a glass one. Express each reading as a ratio to the reference. Those ratios are your geometry correction factors, and they belong pinned up beside the calibrator rather than filed in a folder.
Then apply the working rule, which is simpler than the factors: assay in the geometry you calibrated in, or correct for the difference deliberately. Most departments do the first without realising it, until the day somebody assays a therapy capsule in an unfamiliar pot and gets a number that is out by more than the tolerance allows.
When the numbers stop making sense, the pattern of the error tells you where to look before an engineer is called.
Two environmental fixes are worth building in from the start. Keep the generator, the delivery pot and the waste bin away from the chamber, because a rising background from a neighbouring source looks exactly like an instrument fault. And fit shielding rings around the well. Nuclear Shields make them in 20 mm and 50 mm of lead, dimensioned to fit specific chamber bodies including the Biodex, Capintec, Comecer and ISOMED 2010 patterns, so the make and model of your calibrator has to be stated when they are ordered. They cut the background reading and they cut the dose to whoever stands at that bench all morning, which in a busy hot lab is a measurable share of somebody's annual whole body dose. How that dose gets recorded is covered in personal dosimetry in a Bangladeshi nuclear medicine department.
Finally, decide at purchase where the instrument will be calibrated and by whom. A chamber that has to leave Bangladesh for its accuracy check is out of the department for weeks, so either the contract includes a loan instrument or the department needs a written plan for what it does in the meantime. Ask the question during tender evaluation. Nobody asks it afterwards until it has become a problem.