Low energy high resolution, general purpose, high sensitivity, pinhole and fan beam collimators do different jobs. How to choose the right one for each study, and how to store and handle a component that cannot be repaired locally.
A department in a divisional hospital takes delivery of a dual head gamma camera with one pair of low energy high resolution collimators. Bone, renal and thyroid work runs well for a year. Then an endocrinologist asks for post-therapy I-131 imaging, and the department discovers that the only collimators in the building will let 364 keV photons pass straight through the septa. The images come back with a haze that no reconstruction setting will remove, and the fix is an import order with a lead time measured in months.
Collimators are the least glamorous item on a gamma camera purchase order and the one that most often gets trimmed at the negotiation stage. They also determine the resolution and count rate of every study the department will ever do. Nothing downstream, no reconstruction algorithm and no detector technology, recovers information that the collimator threw away.
A gamma camera detector records where a photon landed but not the direction it came from. Without a collimator, every point in the patient illuminates the whole detector and the image is a uniform blur. A parallel hole collimator is a lead honeycomb placed in front of the detector: photons travelling roughly parallel to the holes pass through, and photons arriving at an angle are absorbed by the septa between them. The image is formed by rejection. Only a very small fraction of emitted photons make it to the crystal, which is why nuclear medicine acquisitions are measured in minutes rather than milliseconds.
Two properties compete. Spatial resolution improves with narrower holes and longer septa, because the accepted cone of directions is narrower. Sensitivity, the fraction of emitted photons recorded, improves with wider holes and shorter septa, for the same reason in reverse. The relationship is not linear: sensitivity varies roughly with the square of resolution, so a collimator that halves the accepted angle costs you far more than half your counts. That is the whole design space, and every named collimator type is a chosen point within it.
Two practical consequences follow. First, resolution degrades as the source moves away from the collimator face, so patient contact matters. A technologist who leaves a five centimetre gap because the couch is easier to position that way has thrown away more resolution than any purchasing decision could recover. Second, adding count time can partly compensate for a low sensitivity collimator, but nothing compensates for a low resolution one.
Almost all routine work uses Tc-99m at 140 keV, so most departments live inside the low energy family. The three common members are the same idea at different settings.
| Collimator | Bias | Typical use |
|---|---|---|
| Low energy high resolution (LEHR) | Resolution over counts | Bone scans, renal cortical imaging, most planar and SPECT work with Tc-99m where the patient can lie still long enough |
| Low energy general purpose (LEGP or LEAP) | Balanced | Mixed lists, dynamic studies where frames are short, larger or restless patients, departments that own only one pair |
| Low energy high sensitivity (LEHS) | Counts over resolution | First pass and flow studies, very short frames, paediatric work where movement is inevitable and speed matters more than fine detail |
If a department can afford one pair only, LEHR is usually the right single choice for a general list, with the understanding that fast dynamic work will be noisier than it needs to be. If the department expects a lot of flow and first pass studies, the general purpose collimator is the safer single purchase. Buying only high sensitivity collimators to make lists faster is a false economy that shows up as unreportable spine images.
Septal thickness has to scale with photon energy, because higher energy photons penetrate lead more readily. Push an isotope beyond what a collimator was designed for and photons travel through the septa rather than the holes. The result is a characteristic star or haze pattern, a loss of contrast, and counts appearing in places where there is no activity.
Parallel hole collimators map the patient one to one onto the detector. Where the organ is much smaller than the detector, that wastes most of the field of view, and a different geometry does better.
A pinhole collimator is a lead cone with a small aperture at the tip. It magnifies a small structure onto the full detector, which is why it remains the best planar tool for thyroid imaging, parathyroid localisation, paediatric hips and small joints. Most pinhole collimators accept interchangeable aperture inserts of different diameters, trading resolution against counts in the same way as a parallel hole collimator. Two things trip up new users: sensitivity falls off steeply as the organ moves away from the aperture, so distance has to be reproducible if you want comparable images; and the image is geometrically distorted towards the edges, so a nodule position near the edge of the field is not where it looks.
A fan beam collimator has holes that converge in one direction and stay parallel in the other, magnifying the brain onto a detector designed for a torso. Brain SPECT with a fan beam gains counts and resolution together, which is unusual and welcome. The trade is truncation: anything wider than the converged field is cut off, so it is a brain tool and not a general one. Converging and diverging cone geometries and slant hole designs exist for particular applications and are rarely worth specifying unless a named clinical service requires them.
| Study | Isotope | Collimator |
|---|---|---|
| Whole body bone scan and bone SPECT | Tc-99m MDP or HDP | LEHR |
| Renography and other dynamic renal work | Tc-99m DTPA, MAG3 or EC | LEHR or LEGP, depending on frame length |
| Thyroid uptake and nodule imaging | Tc-99m pertechnetate or I-123 | Pinhole for detail, LEHR for a survey view |
| Post-therapy and whole body iodine imaging | I-131 | High energy |
| Infection and inflammation imaging | Ga-67 or labelled leucocytes with In-111 | Medium energy |
| Myocardial perfusion SPECT | Tc-99m sestamibi or tetrofosmin | LEHR |
| Gated blood pool and first pass | Tc-99m labelled red cells or pertechnetate | LEGP for gated work, LEHS for first pass |
| Brain perfusion SPECT | Tc-99m HMPAO or ECD | Fan beam where available, otherwise LEHR |
A collimator is a slab of lead with a fragile face. The honeycomb is thin lead, the cover over it is a light foil or composite skin, and both crush easily. A dent is not cosmetic. It bends septa, and that produces a fixed non-uniformity which appears on every subsequent image and on every flood. There is no local repair. A damaged collimator is an import order and a hole in your service list.
Collimator specifications are where a low bid quietly becomes an expensive machine. A line reading collimators as required is an invitation to supply the minimum.