| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Commentary |
Department of Nuclear Medicine, Box 170, Addenbrooke's Hospital, Cambridge CB2 2QQ, UK
 |
Introduction |
|---|
 Top Introduction References |
|---|
In principle, all that is needed to allow a dual headed gamma camera to perform GCPET imaging is the addition of coincidence circuitry. The two 511 keV 
-rays arising from a positron annihilation are emitted 180° apart. When detected in coincidence, the line passing through the site of annihilation is determined. This "electrical collimation" means that it is possible to acquire and reconstruct a tomographic image without any conventional collimation, although sometimes coarse axial septa are used, primarily to prevent the detection of -rays arising from outside the field of view. In either case there is a large increase in sensitivity and a significant improvement in resolution compared with conventional (single photon) imaging. In particular, there is a vast improvement over previous attempts to image PET nuclides with a gamma camera using ultra high energy collimators.
It is pertinent to compare the physical performance of GCPET with dedicated PET. Intrinsic resolution is similar, being about 4–5 mm full width at half maximum height (FWHM) in both modalities. The energy resolution of GCPET systems is superior because the scintillator used in gamma cameras, sodium iodide (NaI(Tl)), produces much more light than the bismuth germanate (BGO) used in most dedicated PET cameras. This gives the gamma camera better intrinsic scatter rejection as well as offering scope for performing scatter corrections based on multiple energy windows. In practice, the main disadvantage of GCPET is the limited number of usable counts that can be acquired in a clinical image. This is a consequence of two fundamental problems. First, NaI(Tl) has a much lower stopping power for 511 keV -rays than BGO. Second, the gamma camera, being a single crystal device, has a lower count rate capability than a dedicated PET system, which has many discrete small detectors. These two effects compound each other in that the reduced stopping power leads to a higher proportion of events where only one of the pair of -rays is detected. These unpaired events are unusable for reconstruction and can saturate the limited count rate capability of the GCPET system.
Manufacturers have been developing measures to overcome these problems. The first improvement was to offer thicker crystals to increase the stopping power. Table 1
compares the detection probabilities for a standard 9.5 mm NaI(Tl) crystal with those for a 19 mm thick crystal. Doubling the crystal thickness increases the probability of detecting a coincidence event by about a factor of four. The ratio of usable tounusable events is also increased. We have beenable to show that this increase in crystal thickness does not compromise the performance of the camera for conventional nuclear medicine imaging, at least at the 140 keV energy of 99Tcm. Table 1 also includes a comparison with a 30 mm thick BGO crystal, which shows that even with thicker crystals the gamma camera has sensitivity inferior to a typical dedicated PET system. The second main advance in gamma camera design has led to an improvement in the count rate performance. This has been achieved by improving the pulse processing electronics and by dynamically dividing the detector head into zones that allow processing of multiple events in parallel.
|
-rays. The total attenuation is a function of the product of the attenuation for each -ray. Various attenuation correction techniques can be used with GCPET. The simplest is a software correction assuming uniform attenuation. In regions where attenuation coefficients vary, for example in the chest, the only valid correction is with a measured attenuation map obtained from transmission scanning. Manufacturers have tried a number of approaches, either using transmission sources or hybrid gamma camera/CT systems. However, these add considerably to the expense. Corrections for random coincidences and scattered events have not yet been successfully implemented by all manufacturers of GCPET systems. The importance of each of these corrections will vary depending on the type of imaging being performed. Corrections are less important when the problem is simple lesion detection, more important when comparing uptake at different depths within the body, and essential if absolute quantification is required. It should also be remembered that all corrections tend to increase noise in the image. Another limitation with many GCPET systems is the type of reconstruction software supplied by the manufacturer. The reconstruction is inherently a three-dimensional (3D) problem, but often theonly solution offered has been single slicerebinning, a simplification that essentially converts the reconstruction into a two-dimensional problem. This simplification is known to degrade resolution away from the central axis of the field of view. Better options are Fourier rebinning or fully 3D reconstruction algorithms.
The next generation of GCPET systems is likely to incorporate discrete detectors using new scintillators, e.g. lutetium oxyorthosilicate. By further addressing the problems of stopping power and count rate capability, these promise performance specifications approaching those of dedicated PET whilst retaining the flexibility of being able to perform routine "single photon" nuclear medicine imaging.
The main clinical use of dedicated PET is in oncology, and this is the area where most work has been done on the clinical usefulness of GCPET. However, it is one area where the limitations of GCPET are most exposed. Most studies comparing GCPET with dedicated PET have shown that the gamma camera detects fewer lesions, particularly below the diaphragm, and may miss lesions up to 15–20 mm in diameter [1]. In the staging of malignancy, it is vital to detect every last small deposit of tumour as this may significantly affect management of the patient. Missed lesions have led to major reservations about GCPET. However, in the staging of non-small cell lung cancer, GCPET has shown an overall sensitivity roughly equal to dedicated PET (90–98%) but a reduced sensitivity for mediastinal nodes (73–86%) [2, 3]. This must be seen in the context of CT and MRI, which have a sensitivity for mediastinal nodes of only 60–70% (and also a specificity of only 60–70%). Mediastinoscopy has a sensitivity of only 80% for mediastinal nodes owing to sampling error. Thus, FDG imaging, even with gamma camera technology, can achieve results comparable with CT plus mediastinoscopy without the need for an invasive procedure. FDG imaging also has the advantage of whole body staging and may detect metastases outside the thorax.
In recurrent colorectal cancer, PET has a role in the staging of patients who have rising serum markers and who are suitable for consideration of further radical surgery. Here, the lack of sensitivity for lesions in the abdomen could be a major concern. However, Montravers et al [4] have presented a series of 71 such patients and found that GCPET altered patient management in 44%. Thus, although dedicated PET might have been expected to change management in as many as 64%, GCPET still provided additional useful information after CT and MRI.
In lymphoma, PET is used for disease staging and also to examine residual masses after treatment to determine whether active disease is still present. The few published studies of GCPET suggest that it retains the high positive predictive value of dedicated PET for residual disease [5].
Recently there has been an important re-examination of the optimal time after injection for imaging cancer with FDG. It has been shown that at 1 h malignant lesions are still accumulating FDG at a significant rate and lesion-to-background ratios are improved by waiting longer [6, 7]. Lesion detectability is also a function of noise in the images and so there is a balance between the increasing lesion-to-background ratio and decay of the 18 F (half-life 110 min). However, with GCPET the amount of activity injected is usually limited by the count rate capability of the camera and not by regulatory controls on maximum administered activity. Hence, when waiting longer it may be possible to compensate for decay by increasing the injected activity and so further enhancing lesion detectability.
In general, in oncology GCPET is not as sensitive as dedicated PET and certainly may miss small lesions in the abdomen. However, when CT and MRI are the only alternatives, early clinical results show that GCPET supplies additional information aiding patient management. GCPET could prove to be a cost effective addition to the imaging facilities available in UK hospitals.
In applications of PET other than oncology, where the clinical usefulness is not dependent on detecting the presence of very small abnormalities, GCPET may in general compare more favourably with dedicated PET. PET detects hibernating myocardium as areas with preserved or increased FDG uptake but reduced perfusion. In this context, very small areas will probably not be clinically significant and so GCPET may be very successful. Several groups have shown encouraging results but have also highlighted the importance of performing measured attenuation correction, by means of transmission scanning, for cardiac imaging [8].
As previously mentioned, the performance of a camera in GCPET mode is better than in conventional SPECT mode owing to increased sensitivity and resolution. This means, for example, that it is possible to obtain GCPET images of the brain using FDG that are of superior quality to 99Tcm-HMPAO images. Therefore, GCPET may prove to have a role in neurology, especially as the early identification of Alzheimer's becomes increasingly important with the emergence of drugs for combating this disease.
Another area for wider application of PET with FDG is imaging inflammation. Again, the ability to detect very small lesions may not be so important. The availability of a simple alternative to white cell labelling would be a major advance for the routine nuclear medicine department.
Gamma camera PET has shown considerable promise as an additional imaging tool for routine nuclear medicine departments. However, it is important to be aware of its limitations when interpreting images. Much of the clinical data published so far for GCPET have been obtained from the earliest commercial GCPET systems. Many improvements have already been made and the technology is still advancing. There are likely to be further improvements in detector design, including the introduction of new scintillators, a wider availability of systems incorporating attenuation correction by transmission scanning, and further developments in reconstruction software. Current and future developments in technology can be expected to lead to improvements in initial clinical results.
Received for publication July 3, 2000. Revision received October 9, 2000. Accepted for publication December 21, 2000.
|
References |
|---|
|
TopIntroductionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
