| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Experimental Studies |
1 Department of Radiology, Duke University Medical Center, Box 3949, Rm 1420, Duke Hospital North, Erwin Rd, Durham, NC 27710.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Spheres with diameters of 6, 10, 13, and 22 mm were placed in the lungs of an anthropomorphic chest phantom to simulate nodules. The lungs, nodules, chest wall, and mediastinum were filled with fluorine-18 activities based on the average radionuclide concentrations in those structures from analysis of attenuation-corrected dedicated FDG PET scans. The image sets were evaluated visually and quantitatively by using contrast and signal-to-noise ratios.
RESULTS: Attenuation correction reduced the artificially high apparent uptake in the lungs, restored the spherical shape to the nodules, and provided an accurate outer body contour with appropriate intensity. Dedicated PET depicted all four nodules, camera-based PET depicted the three largest nodules, and SPECT depicted the two largest nodules. Lesion contrast was better on the attenuation-corrected images than on the non–attenuation-corrected images. The signal-to-noise ratio generally was improved with attenuation correction.
CONCLUSION: Attenuation correction results in many changes in the images and improves lesion detection.
Index terms: Emission CT (ECT), technology, 60.12162, 60.12163 • Fluorine • Glucose • Lung neoplasms, radionuclide studies, 60.3115 • Lung, nodule, 60.3115
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
At about the same time that dedicated PET scanners were being developed (2), attempts were being made to use opposed gamma cameras for annihilation coincidence detection (3). Gamma camera performance was much more limited at that time and often resulted in inadequate coincidence count rates. In addition, reconstruction of the resultant three-dimensional data has only recently become feasible. Gamma cameras with high-energy collimation have been used in the single-photon mode to image the 511-keV photons of 2-[fluorine-18]fluoro-2-deoxy-D-glucose (FDG). This technique has been demonstrated to be inadequate for tumor imaging with FDG but has been used successfully to detect viable myocardium (4,5). Recently, because of the improvements in electronics, gamma cameras are now being used for both single-photon and annihilation coincidence detection (6). The intrinsic resolution of these camera-based PET systems is comparable to that of dedicated PET scanners, but the systems are less sensitive for the detection of the annihilation radiation and are count-rate limited due to the thin sodium iodide detector and the limitations of imaging with two cameras instead of with the thousands of independent detectors typical of dedicated PET scanners (6).
Focal pulmonary opacities can be accurately characterized as benign or malignant with FDG and dedicated PET imaging (7). The accuracy for nodules with diameters of 7–15 mm is the same as that for nodules with diameters of 15–35 mm (8). Third-party payers are now reimbursing the cost of FDG PET performed to diagnose and stage lung cancer (9). Camera-based PET scanners are being used to evaluate focal pulmonary lesions, and the results of the initial clinical studies are beginning to appear in the literature (10,11). Images obtained with camera-based PET scanners have improved compared to images obtained with gamma cameras with thinner crystals, filter backpropagation algorithms, and not using attenuation correction (6). These improvements are not available with all systems at this time, but they will be incorporated into the new systems.
The purpose of this study was to compare dedicated PET, camera-based PET, and 511-keV collimated single photon emission computed tomography (SPECT) in the detection of pulmonary nodules of various sizes in a realistic chest phantom. Scans were reconstructed with and without the use of attenuation correction. This study permits the realistic evaluation of these imaging methods in the detection and characterization of lung nodules of various sizes.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
Scanning
The scanners used for the study were the Advance (GE Medical Systems, Milwaukee, Wis), in a standard configuration, and the Varicam (Elscint, Haifa, Israel), equipped with 
-inch-thick sodium iodide crystals. Varicam coincidence scanning was done with manufacturer-provided transaxial septa in front of the cameras. These septa were 3 mm thick, were spaced 13 mm center-to-center, and extended approximately 6 cm from the camera face. The septa reduce the amount of radiation that is detected from outside the field of view, limit the axial angles of photon pairs that are detected, allow for two-dimensional reconstruction of the data, and allow patients with 10 mCi (370 MBq) FDG to be imaged at a reasonable time after injection. In addition, the septa cut down on detected scattered events, as in two-dimensional dedicated PET. Between the septa and the sodium iodide detector is a graded absorber that preferentially transmits the unscattered 511-keV photons relative to the lower-energy scattered photons (13).
Dedicated PET scans were reconstructed with and without attenuation correction by using a Hann filter with a cutoff of 0.71 per centimeter. Reconstructed images were 128 x 128 x 35 and had 3.5 x 3.5 x 4.25-mm voxels. Images were transferred to off-line computers for display and analysis.
For camera-based PET, two types of events were used: events with both photons in the photopeak and events with one photon in the photopeak and one in a lower-energy window. Camera-based PET data were binned into 90 projections (two-degree sampling), a 128 x 128 matrix, and 4 x 4-mm pixels by using single-section rebinning (13). These projections were transferred to off-line computers and reconstructed by means of filtered back-projection. When these images were aligned with uncorrected PET images, an appropriate transformation was determined to scale, rotate, and translate the PET attenuation maps (reconstructed from transmission data) to spatially match the gamma camera coincidence images. Attenuation factors were then created by projecting through these reoriented attenuation maps, and the emission data were corrected with these factors. Finally, the corrected emission data were back-projected, which resulted in attenuation-corrected images. A three-dimensional Hann filter with a cutoff of 0.37 per centimeter was used to smooth the images.
For the 511-keV SPECT data, the same procedure was used to align the attenuation map with the uncorrected images. The nonuniform attenuation map was then used with the iterative Chang method (one iteration) (14) to reconstruct corrected SPECT images. A three-dimensional Hann filter (cutoff, 0.37 per centimeter) was used to smooth the images.
Three sections containing the centers of the four spheres were extracted from each of the six image sets and inspected visually and quantitatively. Quantitative analysis was done by placing an ROI on each sphere in a position and section that yielded the maximum pixel mean value. The diameters of the circular ROIs matched those of the spheres. For each sphere, 10 circular ROIs of the same diameter were placed throughout the lungs as a background region. The mean and SD were determined for each set of 10. Contrast (C) and signal-to-noise ratio (SNR) were determined for each identified lesion as follows: C = (L - µ)/µ and SNR = (L - µ)/
, where L is the lesion ROI value, µ the mean of the background ROIs, and the SD of the background ROIs.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
In these particular image sets, all four lesions are detectable on the attenuation-corrected dedicated PET images, and the 6-mm-diameter lesion is better discerned on the attenuation-corrected image than on the non–attenuation-corrected image. The smallest lesion is not seen on the camera-based PET images, and the 10-mm lesion is better discerned on the attenuation-corrected image than on the non–attenuation-corrected image. The two largest lesions are seen on the SPECT images.
Emission counts from the phantom were added axially over 10 cm. The results are as follows: dedicated PET, 9.9 million counts per 10 minutes; camera-based PET, 0.68 million counts per 30 minutes; and SPECT, 2.7 million counts per 30 minutes. Approximately 50% of the SPECT counts are from septal penetration.
Results of the ROI analysis are shown in Table 2. Regions were drawn only for spheres that could be clearly identified on either the corrected or uncorrected image set. Because of the nonuniformity of the lung region generated by the attenuation artifact, the variation in background regions was reduced considerably with attenuation correction, thereby increasing SNR. Contrast values increase (with this definition) with attenuation correction owing to the lowering of the artificially high background values.
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Until recently, only dedicated PET scanners were being used for FDG imaging. At this time, there are approximately twice as many camera-based PET scanners than dedicated PET scanners (9). Camera-based PET scanners initially used 
-inch-thick crystals, but the low sensitivity of this crystal thickness limited its utility (6). Crystals with thicknesses of 
and 
inch are now being used. The single photon sensitivity for a -inch-thick crystal is approximately twice that for a -inch-thick crystal. Even with the thicker crystal, the system sensitivity of the camera-based system is much less than that of the dedicated PET scanner. In our phantom study, the sensitivity of camera-based PET was 10,000 counts per image, whereas that for dedicated PET was 200,000 counts per image. Iterative reconstruction of the camera-based PET images has been demonstrated to improve the image quality and will be available on the clinical systems (6).
Our results demonstrate that attenuation-corrected dedicated PET can depict nodules as small as 6 mm in diameter in a realistic chest phantom with radionuclide concentrations that were determined from normal tissues and nodules on attenuation-corrected dedicated PET scans. Non–attenuation-corrected dedicated PET images, camera-based PET images, and SPECT images did not provide adequate contrast for demonstrating the smallest nodule. Lesion contrast was improved with attenuation correction for all nodule sizes and all three modalities. Our results demonstrate the importance of attenuation correction for detecting small nodules. With use of the average radionuclide concentrations from the attenuation-corrected dedicated PET scans obtained in 11 patients with malignant nodules, these results suggest that nodules approximately 10 mm in diameter could be detected with clinical FDG studies by using attenuation-corrected camera-based PET. Preliminary clinical studies have demonstrated similar results (10,11).
With dedicated PET scanners, Bengel et al (18) evaluated the effects of attenuation correction on lesion-to-background count rates in a body phantom containing a cylinder with spheres of various sizes and in patients. The lesion-to-background count rates were used as contrast values. The authors reported higher contrast values for the spheres in the phantom and for areas of focal FDG uptake in patients on the non–attenuation-corrected images than on the attenuation-corrected images. The clinical results included 30 focal abnormalities in the lungs.
There are several methodologic differences between our study and the study of Bengel et al (18). First, the phantoms were different. Our phantom contained material to mimic lung attenuation, and the spheres were placed in this low-attenuation region. Bengel et al placed the spheres into a water-filled cylinder, which was placed in a body phantom. Their phantom study used a lesion-to-background activity ratio of 4.0 (15/3.7), whereas a ratio of 30.5 (1.037/0.034) was used in our study. They used a "ringlike" ROI around the focal abnormality to determine the background count rates, whereas we used 10 circular ROIs throughout the lung. With our method, there are more regions throughout the lungs that might have greater variability in counts. When we calculated contrast values with their formula for lesion-to-background count rates, contrast values were higher for the attenuation-corrected images than for the non–attenuation-corrected images. Our results do not confirm their conclusion that uncorrected images provide higher lesion contrast. We would expect the results of our phantom study to differ from their results because their phantom study did not simulate the lungs and would not have the artificially high lung uptake without attenuation correction.
Although the attenuation characteristics of our phantom are similar to those of the human chest and the radionuclide concentrations used in normal tissues and nodules were based on clinical data, the image quality and lesion detection rate in clinical studies may differ from those obtained in our phantom study. The radioactivity concentrations used in our study were meant to match typical FDG distribution. Previous studies have reported various values for tumor-lung uptake ratios. Knight et al (19) reported a mean ratio of 16.6 (range, 5.3–51.0) in 29 malignant lung nodules, and Lowe et al (20) reported a similarly wide range in 10 malignant nodules. Although the 30.5:1 ratio used for this phantom study is somewhat high compared to that used by Bengel et al (18) and to these other measurements, higher lesion values should be expected for our study because of the differences in methods. Scanner spatial resolution effects tend to yield low estimates for radioactivity concentrations of small lesions (12). Use of a high-resolution scanner and a maximum-pixel ROI method (compared with the use of ROI means) (21), however, minimized this error for the 11 patients analyzed. Factors such as FDG accumulation time (20) and blood glucose levels (22) lead to variation in actual tumor-lung ratios. Because lesion detectability is related to image contrast, which, in turn, is dependent on the uptake ratios, actual detectability of various nodule sizes will differ from patient to patient.
Other factors are different between the phantom and clinical studies. Motion is present in clinical studies but not in phantom studies. In clinical studies, the lungs, chest wall, and nodules move, which results in degradation of the image and in the detectability of small nodules—particularly at the lung bases where motion is greatest. The patient may move between the emission and transmission scan, and this motion may result in degradation of lesion contrast on the image. In the phantom study, the lesion and background radionuclide concentrations are uniform, whereas in patients, the radionuclide accumulation is frequently nonuniform in the lesion and background. Even with these limitations of the phantom study, the results do provide information comparing the relative merits of dedicated PET, camera-based PET, and 511-keV SPECT of suspected lung tumors.
The results were obtained by using acquisition geometries and image reconstructions that can be improved. For these systems, camera-based PET stands to improve the most with reconstruction algorithms that better account for the three-dimensional nature of the data. All three modalities may show substantial improvement with appropriate iterative reconstruction algorithms. Our results suggest the following: FDG SPECT can depict nodules as small as 13 mm in diameter, camera-based PET can depict nodules as small as 10 mm in diameter, and dedicated PET can depict nodules as small as 6 mm in diameter. The importance of attenuation correction in the detection of small lesions is demonstrated in this phantom study. The results may differ from those of clinical studies because of patient motion and lack of perfect registration between the transmission and emission images.
Practical Applications: These data demonstrate that dedicated PET can depict lung nodules at least 6 mm in diameter. A multicenter study of indeterminate focal pulmonary lesions demonstrated no difference in the accuracy of dedicated PET images in the characterization of lesions 0.7–1.5 cm in diameter compared to those 1.6–4.0 cm in diameter (8). In this experimental study, attenuation-corrected camera-based PET depicted nodules at least 10 mm in diameter, whereas non–attenuation-corrected camera-based PET depicted nodules at least 13 mm in diameter. In a study comparing non–attenuation-corrected camera-based PET scans to attenuation-corrected dedicated PET scans, 84% of lesions at least 1 cm in diameter and 90% of lesions at least 1.5 cm in diameter were detected with camera-based PET (10). These preliminary clinical data and experimental results suggest that, at this time, camera-based PET systems be used only to characterize nodules with diameters of 1.5 cm or more. When attenuation correction becomes routinely available on camera-based PET systems, clinical studies will be needed to document the smallest lesion size detectable. FDG SPECT of lung nodules requires nodules to be at least 1.3 cm in diameter. Martin et al (4) concluded that lesions larger than 1.5 cm in diameter could be routinely detected with FDG SPECT.
|
Footnotes |
|---|
Address reprint requests to R.E.C.
R.E.C. is a stockholder in PETNet Pharmaceutical, a supplier of FDG.
Abbreviations: FDG = 2-[fluorine-18]fluoro-2-deoxy-D-glucose ROI = region of interest SNR = signal-to-noise ratio
Author contributions: Guarantor of integrity of entire study, R.E.C.; study concepts, T.G.T., R.E.C.; study design, all authors; definition of intellectual content, T.G.T., R.E.C.; literature research, T.G.T., R.E.C.; clinical studies, all authors; experimental studies, C.M.L., T.G.T.; data acquisition, C.M.L., T.G.T.; data analysis, all authors; manuscript preparation, editing, and review, T.G.T., R.E.C.
Received May 20, 1998;
revision requested July 14, 1998; revision received August 26, 1998;
accepted October 7, 1998.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  Y. Wang, R. J. van Klaveren, H. J. van der Zaag-Loonen, G. H. de Bock, H. A. Gietema, D. M. Xu, A. L. M. Leusveld, H. J. de Koning, E. T. Scholten, J. Verschakelen, et al. Effect of Nodule Characteristics on Variability of Semiautomated Volume Measurements in Pulmonary Nodules Detected in a Lung Cancer Screening Program Radiology, August 1, 2008; 248(2): 625 - 631. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Suzuki, H. Jacobsson, T. Hatschek, M. R. Torkzad, K. Boden, Y. Eriksson-Alm, E. Berg, H. Fujii, A. Kubo, and L. Blomqvist Radiologic Measurements of Tumor Response to Treatment: Practical Approaches and Limitations RadioGraphics, March 1, 2008; 28(2): 329 - 344. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Bunyaviroch and R. E. Coleman PET Evaluation of Lung Cancer J. Nucl. Med., March 1, 2006; 47(3): 451 - 469. [Full Text] [PDF]
|
|
|
|
 |
|
 M. R. Acker and S. C. Burrell Utility of 18F-FDG PET in Evaluating Cancers of Lung J. Nucl. Med. Technol., June 1, 2005; 33(2): 69 - 74. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R.-F. Yen, R.-L. Hong, K.-Y. Tzen, M.-H. Pan, and T. H.-H. Chen Whole-Body 18F-FDG PET in Recurrent or Metastatic Nasopharyngeal Carcinoma J. Nucl. Med., May 1, 2005; 46(5): 770 - 774. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Even-Sapir, B. Yuzefovich, E. Miller, J.-P. Bouhnik, O. Zak, H. Lerman, G. Lievshitz, and C. Levin Coincidence Imaging Using 2 Dual-Head {gamma}-Camera Systems, With and Without Attenuation Correction J. Nucl. Med. Technol., December 1, 2004; 32(4): 190 - 197. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.-P. Revel, A. Bissery, M. Bienvenu, L. Aycard, C. Lefort, and G. Frija Are Two-dimensional CT Measurements of Small Noncalcified Pulmonary Nodules Reliable? Radiology, May 1, 2004; 231(2): 453 - 458. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.-P. Revel, C. Lefort, A. Bissery, M. Bienvenu, L. Aycard, G. Chatellier, and G. Frija Pulmonary Nodules: Preliminary Experience with Three-dimensional Evaluation Radiology, May 1, 2004; 231(2): 459 - 466. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. W. Goerres, C. Burger, E. Kamel, B. Seifert, A. H. Kaim, A. Buck, T. C. Buehler, and G. K. von Schulthess Respiration-induced Attenuation Artifact at PET/CT: Technical Considerations Radiology, March 1, 2003; 226(3): 906 - 910. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. J. Kadrmas and P. E. Christian Comparative Evaluation of Lesion Detectability for 6 PET Imaging Platforms Using a Highly Reproducible Whole-Body Phantom with 22Na Lesions and Localization ROC Analysis J. Nucl. Med., November 1, 2002; 43(11): 1545 - 1554. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. A. D'Amico, T. Z. Wong, D. H. Harpole, S. D. Brown, and R. E. Coleman Impact of computed tomography-positron emission tomography fusion in staging patients with thoracic malignancies Ann. Thorac. Surg., July 1, 2002; 74(1): 160 - 163. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.F. Vansteenkiste Imaging in lung cancer: positron emission tomography scan Eur. Respir. J., February 1, 2002; 19(35_suppl): 49S - 60s. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Imdahl, S. Jenkner, I. Brink, E. Nitzsche, E. Stoelben, E. Moser, and J. Hasse Validation of FDG positron emission tomography for differentiation of unknown pulmonary lesions Eur. J. Cardiothorac. Surg., August 1, 2001; 20(2): 324 - 329. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J.F. Vansteenkiste and S.G. Stroobants The role of positron emission tomography with 18F-fluoro-2-deoxy-D-glucose in respiratory oncology Eur. Respir. J., April 1, 2001; 17(4): 802 - 820. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Tatsumi, H. Kitayama, H. Sugahara, N. Tokita, H. Nakamura, Y. Kanakura, and T. Nishimura Whole-Body Hybrid PET with 18F-FDG in the Staging of Non-Hodgkin's Lymphoma J. Nucl. Med., April 1, 2001; 42(4): 601 - 608. [Abstract] [Full Text]
|
|
|
|
 |
|
 G.-U. Hung, Y.-C. Shiau, S.-C. Tsai, Y.-J. Ho, C.-H. Kao, and R.-F. Yen Differentiation of Radiographically Indeterminate Solitary Pulmonary Nodules with [18F]Fluoro-2-deoxyglucose Positron Emission Tomography Jpn. J. Clin. Oncol., February 1, 2001; 31(2): 51 - 54. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
