What each phantom in a nuclear medicine QA set actually measures, why flood and resolution phantoms come first, and how a new department can build a working set in three phases instead of buying everything at once.
A procurement committee is handed a phantom list by the medical physicist and asked to approve it. There are eight items on it, none of the names mean anything to the people signing, and the obvious reaction is to cut the list in half. That instinct is often right. The method is usually wrong, because the phantoms get cut by price rather than by what the department can safely operate without.
This is a guide to what each phantom in the Leeds Test Objects nuclear medicine range measures, and to which ones a department genuinely cannot start without.
The descriptions below are functional: what test the phantom supports and when you would run it. Several of these phantoms are supplied in more than one configuration, so confirm inserts, fill options and accessories at quotation stage rather than assuming from the model name.
| Phantom | What it supports | When you use it |
|---|---|---|
| Gamma Flood Phantom | Uniformity of detector response across the field of view, integral and differential | Daily, and after any detector or collimator service |
| Gamma Resolution Phantom | Spatial resolution, scored visually from a bar pattern | Weekly, and at acceptance |
| Gamma Distortion Phantom | Spatial linearity and geometric distortion from an orthogonal pattern | Monthly, and after a correction map update |
| SPECT IQ | Tomographic image quality: contrast, uniformity and resolution in reconstructed slices | Quarterly, at acceptance, and after any reconstruction software change |
| PET IQ Phantom | PET image quality, including recovery of hot and cold structures | At PET acceptance and periodically thereafter |
| PET CT MRI Cubes | Spatial registration and alignment between PET, CT and MRI datasets | At acceptance, after gantry or couch service, and periodically |
| VERTO Dynamic Line Phantom | Time dependent and dynamic acquisition behaviour using a line source arrangement | Commissioning and periodic checks of dynamic protocols |
| Quattro 4DCT Respiratory Motion Phantom | Respiratory motion, gating and 4DCT acquisition behaviour | Commissioning of gated and four dimensional protocols |
There are two ways to produce a uniform flood field. A sealed cobalt sheet source needs no preparation, but it is a sealed radioactive source: it needs an import approval, a registration entry, a leak test schedule and eventually a disposal route, and its activity falls steadily so acquisition times lengthen year on year. A refillable flood phantom uses the technetium eluate the department already has, which is why most Bangladeshi departments with a generator end up preferring it.
The catch is that a refillable phantom is only as uniform as the person who filled it. A trapped bubble reads as a cold spot, and a technologist who does not know that will report a detector fault that does not exist.
Departments often buy a bar pattern phantom and stop there, on the reasoning that if the camera resolves fine bars it must be healthy. It is a reasonable assumption and it is wrong often enough to matter. Resolution and linearity are different properties, and a positioning circuit fault can leave resolution acceptable while bending straight lines across part of the field.
The practical consequence appears in SPECT rather than in planar imaging. Planar work tolerates modest distortion because the reader is looking at relative uptake. Tomographic reconstruction assumes each projection maps predictably into the volume, so distortion propagates into the reconstructed slices as blurring and as apparent uptake in the wrong place. If your department does any SPECT at all, the distortion phantom belongs in the set.
A flood tests the detector. A bar pattern tests the detector and the collimator. Neither of them touches reconstruction, and reconstruction is where a modern nuclear medicine system does most of its work. A tomographic image quality phantom such as SPECT IQ exercises acquisition, orbit, reconstruction filter, attenuation correction, scatter correction and display in a single measurement, which is why it is the phantom to run after a software upgrade.
Software upgrades are the reason this matters more than departments expect. A vendor engineer updates the console, the default reconstruction parameters shift, and the images look slightly different for months before anybody says so out loud. A SPECT IQ acquisition taken before and after the upgrade turns that into a documented, arguable change rather than a feeling.
Acquire it with your standard clinical protocol, not a special QA protocol. The point is to test what you actually do to patients.
PET image quality is assessed with its own phantom, because the failure modes differ: contrast recovery for small structures, background variability, and the behaviour of attenuation correction derived from CT. A PET IQ phantom is an acceptance item first and a periodic item second.
Registration is the other PET specific concern, and it is the one that is easy to ignore. If the PET and CT datasets are misaligned, attenuation correction is applied to the wrong voxels and the result is an artefact that looks like real physiology. Registration cubes visible in PET, CT and MRI let you check alignment across modalities directly, which is the only way to catch a slow drift after couch or gantry service. In a hospital where the PET/CT and the MRI both feed a radiotherapy planning system, this check has consequences well beyond the nuclear medicine department.
Dynamic and motion phantoms are the ones to defer unless you are actually doing the work. A dynamic line phantom supports checks on time dependent acquisition behaviour, which matters for dynamic renal and cardiac protocols. A respiratory motion phantom supports commissioning of gated acquisitions and 4DCT.
There is a common Bangladeshi case where the motion phantom pays for itself early: a hospital where the CT simulator is shared between nuclear medicine and radiation oncology. Gated and four dimensional acquisition on that scanner has to be commissioned properly for the radiotherapy side regardless, so the phantom is a shared asset and the cost sits in a different budget line.
One more point that comes up in e-GP tenders: if you specify only a brand and model, a technically weaker substitution can be argued in on price. If you specify the measurement, the tolerance you need and the documentation required, the evaluation has something objective to work with.