Applied Radiology

Almost one year ago, Dr. Henry Wagner, in this space, presented very early data on gamma camera coincidence imaging in several different entities. During the ensuing year, a good deal more information has become available as over 70 coincidence cameras are in the field. The purpose of this article is to provide an update for the practicing radiologist and nuclear medicine physician on the current capabilities of coincidence imaging on a gamma camera.

Principles

The principles of coincidence imaging are well known and have been available in medicine for many years. The original positron cameras existed as long as 30 years ago.1 Beginning in the 1970s, significant research time and effort was devoted to the development of an imaging device that could effectively display the distribution of positron radiopharmaceuticals in patients. The result of this effort was the positron emission tomographic scanner (PET).2 This device takes advantage of the annihilation radiation emitted when a positron and electron fuse and annihilate one another. From this interaction, two 511 keV photons are emitted that are 180° apart. By timing the interaction of photons in opposing detectors, coincident events can be identified. Though some non-coincidence events contain useful information, the focus of coincidence detection has always been to deal as much as possible with coincidence events.

As it was recognized early on that counting statistics were crucial in coincidence imaging, special crystals were developed for dedicated PET scanners. These crystals permitted a much higher sensitivity than could be obtained with sodium iodide, but they also decreased the resolution of the system while increasing its cost. The cost of a dedicated PET scanner, despite its apparent clinical efficacy, has limited the widespread introduction of these devices, with only about 60 being used in the United States at present.

The development of the sodium iodide-based PET scanner at the University of Pennsylvania in the late 1980s renewed interest in the use of sodium iodide as a detector system for coincidence events.3 The cost of this device was lower than the standard PET scanner, though its sensitivity was lower as well. However, due to a higher energy resolution than dedicated PET devices (12% for the camera versus 30% for dedicated PET scanners) it demonstrated decreased scatter acceptance. In retrospect, perhaps the major impact of this device was to raise considerations as to whether standard gamma cameras using sodium iodide detectors could be acceptable devices for coincidence detection.

Sodium iodide gamma camera as a coincidence device

Over the last several years, significant attention and research has been dedicated to optimizing the performance of the standard gamma camera in the coincidence mode. Several important issues have emerged with regard to whether or not gamma camera detectors can adequately serve as coincidence devices. In order to look at this issue, we must consider some of the problems associated with the sodium iodide detector.

Over the last 25 years, the gamma camera has progressively been optimized for low energy imaging. Thinner crystals have been developed and low-energy collimator systems have been introduced. Detector energy and linearity mapping have been optimized so that low energy isotopes are better imaged. The thinner crystals intended to improve low energy imaging have sacrificed sensitivity for higher energy isotopes.

A major factor in dealing with coincidence events is the sensitivity of the gamma camera detector at 511 keV. In order to improve the sensitivity, the option of thicker detectors has been employed. Currently, commercially available systems have sodium iodide detectors ranging up to 3/4" in thickness. However, detector thickness as great as 1" have been considered and may ultimately prove desirable in this application. The optimal sodium iodide crystal thickness in a theoretical situation might be as much as 2.5 inches, but practical considerations may well limit units to thinner crystals.

A potential implication of using thicker crystals is decreased resolution at other lower clinical energies. While this would have been enough to halt further consideration a number of years ago, the more recent development of resolution recovery algorithms for use in digital gamma cameras has made the utilization of thicker crystals a more attractive option.

Another issue, that of detector count rate, is one that is somewhat misunderstood. There are effectively no collimators in place in the coincidence mode. Because the collimator typically screens up to 99.99% of the photons that would impact the crystal, the count rates for coincidence studies can be remarkably high without a collimator. When acquiring a patient study at high count rates, one must be concerned with the dead- time losses and potential paralysis of the detector. Typically, in an FDG study begun one hour after administration of

5 mCi of FDG to the patient, the count rate is on the order of 1.3 million per second. If the detector is not capable of handling this count rate, then artifacts in the clinical images may be generated.

If one attempts to do a measured attenuation correction by transmission imaging of the patient, using radioactive sources mounted on opposing detectors to generate transmission data, the count rate will increase further. While the impact of measured attenuation correction will be discussed below, the detector requirements are further increased with regard to count rate when this option is employed. Because lower counts in the study create a major difference when comparing dedicated PET and gamma camera systems, one should investigate systems that yield the highest trues (true coincidence events) rates.

Attenuation correction

Before considering attenuation correction for coincidence imaging, a few observations should be made relative to single photon imaging. Attenuation correction for single photon studies has received a great deal of attention in recent years. Particularly where cardiac single photon emission computed tomography (SPECT) is involved, many users feel that correcting for body thickness and intervening structures will decrease or eliminate inferior wall and breast artifacts.4,5 Clinical attenuation correction of SPECT studies is common. However, problems exist with regard to methods that do not employ measured attenuation. In measured attenuation correction, a transmission study of the patient is performed and the attenuation correction maps are generated from that data. Unfortunately, when estimating correction by computing the thickness of the patient and applying the linear attenuation coefficient, errors tend to result.

The chest is a particularly complex area for attenuation correction. Multiple attenuating structures exist. These include water density, bone density, and air density structures. The best method that has been found to accurately compute attenuation for the chest is to measure it by transmission imaging.6 As noted above, when attenuation correction is applied to coincidence imaging, detector performance becomes critical.

Is attenuation correction necessary?

On initial thought, it would seem that attenuation correction for a 511 keV photon would be unnecessary. This high-energy emission should be less affected by attenuation. However, for coincidence imaging, attenuation correction is actually more important than for many single-photon applications. The combined path length of the two gamma rays generated from an annihilation interaction is twice that of a single gamma ray of the same energy arising from the same location. Attenuation is therefore magnified in this setting.

This debate over attenuation correction has been going on for a number of years in both the PET community and, more recently, in the coincidence camera community. There are advocates who present data on both sides of the question. For PET, the effort required in the most common methodology makes attenuation correction a cumbersome procedure. In the gamma camera community, however, the procedure is highly automated and adds no more than five minutes to the data acquisition and two minutes to the processing. Routine attenuation correction of coincidence data on the gamma camera is now available.

With such a small investment in time and money, there is increasing interest in the use of attenuation correction on the gamma camera. In a study of 25 patients in our laboratory, data was reconstructed from the same data set both with and without measured attenuation correction. Three experienced nuclear medicine physicians were asked to review the data and grade it in regard to diagnostic quality, the presence of artifacts, and overall image quality. The attenuation-corrected images consistently graded statistically significantly higher than the non-attenuation corrected images, particularly in the region of the chest. Lesion visibility, location of the lesion, and the general accurate representation of isotopic distribution were improved after attenuation correction (figure 1). In addition, five lesions were identified that were not seen on the non-attenuation-corrected data or were felt to be part of a normal organ on the non-attenuation-corrected images (figure 2).

Clinical applications

There are a number of clinical applications that have been advocated for gamma camera coincidence imaging. Predominantly, these are based on information developed in the PET literature.7 The first of these studies to accrue any significant number of patients has been the multi-center cooperative trial on imaging lung cancer. In this protocol, patients who were surgical candidates or mediastinoscopy candidates were studied with coincidence imaging on a gamma camera. Some of these patients also were studied by dedicated PET scanner. The results of the study were correlated with the surgical/pathologic findings. To date, approximately 150 patients have been studied in this protocol.

When the correlation was made between independent reading of the coincidence images and the pathologic data, a sensitivity of approximately 94% was obtained; specificity in the same patient set was approximated at 84%.8 About one-half the lesions in the group were smaller than 3 cm in size. This data is comparable to previously published data for dedicated PET scanners.

In a small group of patients, success has been reported in detecting malignant lymphoma with coincidence imaging.7 At the moment, one of the most significant applications at our institution is the use of the FDG study for the growing lymphoma population. As long time advocates of gallium-67 for imaging of suspected lymphoma, we had the opportunity at our institution to look at the crossover in a limited number of patients. While gallium remains a viable lymphoma imaging radiopharmaceutical, FDG appears to avoid a number of the problems inherent in gallium-67 imaging. For example, intraluminal bowel activity is not identified on the FDG study. However, when imaging the pelvis it is necessary that the urinary bladder be drained during the examination. In one of our pediatric patients with non-Hodgkin's lymphoma, regrown thymus raised questions as to the presence of disease on the gallium scan as residual disease (figure 3). The FDG study in that patient was normal. At biopsy, only normal thymus was identified.

There appears to be great promise in the use of FDG for the detection of recurrent lymphoma. Again, in anecdotal situations, a number of patients have been identified who demonstrate more extensive disease on the FDG study than on the CT scan. At least two patients at our institution have been restaged purely on the basis of FDG. However, a more systematic appraisal of the role of gamma camera FDG in lymphoma is required based on this early data.

Other areas where FDG imaging is of interest include breast imaging and ovarian cancer. At the present time, there is insufficient data to support the routine use of gamma camera coincidence imaging in either of these applications. However, research studies are under way in an effort to further determine the accuracy in both of these conditions. Ovarian cancer is of special interest to the oncology community because the detection of recurrence is extremely difficult by standard means. As of this writing, however, we cannot routinely advocate the use of FDG coincidence imaging in either breast cancer or ovarian cancer.

Recurrent brain tumor studies

As a relatively small homogeneous organ, brain imaging on a coincidence gamma camera is much less fraught with technical issues than the imaging of chest lesions. Attenuation correction is less of a problem in the brain, and the major consideration for both dedicated PET and gamma camera PET becomes the appropriate correlation with other imaging modalities to precisely localize FDG uptake. As "fusion imaging" continues to develop, there is every reason to believe that the overlay of CT and MR data with FDG will improve the diagnostic accuracy of all these methods in the patient with recurrent brain tumor.

Cardiac FDG studies

Estimates at our institution have indicated that up to 15% of patients undergoing cardiac revascularization do so with unclear indications. In some patients, the decision as to whether to perform a heart transplant or revascularize the heart is difficult. In these cases, FDG cardiac imaging may prove of use. For example, discovery of a hypoperfused section of myocardium that has significant FDG concentration may push the clinician to try revascularization for this patient, if it is technically feasible, rather than transplantation.

Collimator-based FDG cardiac studies appear to offer information that is equivalent to that of coincidence images. However, they do not deal with issues of attenuation correction. In collimated studies, attenuation correction is less of a problem than in the coincidence mode (figure 4).

Collimator FDG studies

As an interim step in the evolution of FDG imaging using the gamma camera, high energy collimators have historically been employed to image FDG. Favorable results have been attained for cardiac studies, although attenuation correction usually was not employed. However, some problems are encountered when attempting to apply collimator FDG to oncology applications. Perhaps the most significant issue is the low count sensitivity encountered in collimator studies. If the argument is made that gamma camera coincidence imaging has decreased count sensitivity, as compared to dedicated PET, the collimator studies have an order of further magnitude decrease in count sensitivity. Because the acquisition time cannot be lengthened in order to try to overcome this issue, due to the short half-life of FDG, collimated studies remain count poor in comparison to coincidence detection.9

Reimbursement

In the ideal scenario, reimbursement would not affect the choice of technology to be employed for imaging. In

the real word, however, this often

has not been the case. The cost of FDG and coincidence devices was hard to justify when no reimbursement was available. However, with the recent decision by the Health Care Financing Administration (HCFA) to reimburse for coincidence imaging related to staging of non-small cell lung cancer and evaluation of solitary pulmonary nodules, and a willingness to look at other applications, a major barrier to the introduction of these devices has been lowered.

Summary

Judging from the current trends, it is becoming clear that gamma camera coincidence imaging will play a significant role in nuclear medicine's future. In order to properly utilize these devices, however, an understanding of the biodistribution of FDG and the physics of gamma camera coincidence operation are imperative. The performance of these devices, and the additional diagnostic information they bring to bear in various clinical settings is starting to appear. While debate still exists over the method and the necessity for attenuation correction, our clinical experience indicates that attenuation correction improves image quality and diagnostic yield.

The number of coincidence gamma cameras from all vendors is likely to exceed the number of PET scanners in this country in a short period of time, bringing the benefits of FDG imaging to the community at large. Gamma cameras should not be held to a comparison with dedicated PET. Rather, they should stand on their own in determining whether this technology offers sufficient additional clinical information to be routinely useful. In the study noted above on non-small cell lung cancer, coincidence gamma cameras would appear to meet this criterion. When clinical FDG data is not available, management decisions often are made without full knowledge of the status of the patient.

Finally, there is the question of timing. Should you wait for the optimal device or start FDG imaging immediately? Analogous areas exist elsewhere in medical imaging. For example, computed tomography (CT) studies are performed on a wide variety of instruments. While spiral CT may be the current state of the art, CT examinations are not denied to patients when there is no spiral CT locally available. Non-spiral examinations still provide information that can be used in patient management. Dedicated PET scanners may not be available in all communities; however, gamma camera PET could be in the very near future. Additional, useful clinical information is added to patient management by these studies and we should not allow honest disagreements over technologic issues to impede the acquisition of vital patient management data. AR

References

1. Cassen B: Principles of Instrumentation. In: Blahd WH (ed): Nuclear Medicine, pp 58-60. New York, McGraw-Hill, 1965.

2. Ter-Pogossian MM, Phelps ME, Hoffman EJ, Mullani NA: A positron-emission transaxial tomograph for nuclear imaging. Radiology 114(1):89-98, 1975.

3. Muehllehner G, Karp JS: A positron camera using position sensitive detectors: PENN-PET.

J Nucl Med 27(1):90-98, 1986.

4. He ZX, Scarlett MD, Mahmarian JJ, Verani MS: Enhanced accuracy of defect detection by myocardial single-photon emission computed tomography with attenuation correction with gadolinium-153 line sources: Evaluation with a cardiac phantom.

J Nucl Cardiol 4(3):202-210, 1997.

5. Bacharach SL, Bivat I: Attenuation correction in cardiac positron emission tomography and single-photon emission tomography. J Nucl Cardiol 2(3):246-255, 1995.

6. Hashimoto J, Ogawa K, Ichihara T, et al: Application of transmission scan-based attenuation compensation to scatter-corrected Thallium-201 myocardial single-photon emission tomographic images. Eur J Nucl Med 25(2):120-127, 1998.

7. Valk PE, Pounds TR, Tesar RD, et al: Cost-effectiveness of PET imaging in clinical oncology. Nucl Med Biol 23(6):737-743, 1996.

8. Henry N. Wagner, MD: Personal communication, January, 1998.

9. Grossman ZD, Lamonica MD, Klippenstein MD, et al: A comparative study of 511 KeV SPECT and PET, for malignancy detection using separate 370 MBq Fluorine-18-FDG on different days. Radiology (Suppl) 205(P):221, 1997.

Dr. Henkin is Director of Nuclear Medicine and Professor of Radiology at

Loyola University Medical Center in Maywood, IL. He is also a member of the editorial advisory board of this journal.