| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, University of Pittsburgh Medical Center, 200 Lothrop
St., Pittsburgh, PA 15213.
2 Present address: Institute of Diagnostic Radiology, University of
Erlangen-Nuremberg, Maximiliansplatz 1, Erlangen 91054, Germany.
3 Present address: Department of Medicine, University of Tennessee Medical
Center, 1924 Alcoa Hwy., Knoxville, TN 37920.
Received September 8, 2003; accepted after revision November 25, 2003.
Address correspondence to N. Avril
(AvrilNE{at}msx.upmc.edu).
OBJECTIVE. CT in positron emission tomography (PET)–CT imaging is often performed as a single scan from the base of the skull to the groin, potentially resulting in degradation of the quality of CT scans depending on the position of the patient's arms and mode of breathing and the use and timing of IV contrast injection. The aim of our study was to assess the impact of artifacts on the diagnostic quality of CT scans using a single-detector helical CT scanner in PET–CT imaging.
MATERIALS AND METHODS. Two radiologists retrospectively evaluated the diagnostic image quality of CT scans obtained with PET–CT in 81 patients with lymphoma. The severity of the artifacts related to the position of the patient's arms beside the body, the influence of breathing motion, and the presence of contrast material in the upper thoracic veins were ranked using a 4-point scale.
RESULTS. Performing CT with the patient's arms positioned beside the body resulted in streak artifacts, predominantly in the upper abdomen, that were graded as mild in 22%, moderate in 40%, and severe in 38% of the scans. A patient's weight significantly correlated with the degree of severity of the artifacts (p < 0.05). Shallow breathing by the patient during scanning caused blurring and double-imaging, again predominantly in the upper abdomen, that were graded as mild in 23%, moderate in 49%, or severe in 28% of the scans. In 84% of the CT scans obtained with IV contrast material, the image quality of the upper thoracic region was moderately (27%) or severely (57%) degraded by streak artifacts from highly concentrated contrast material in the upper thoracic veins.
CONCLUSION. The use of a single-detector CT scanner in whole-body PET–CT decreases the image quality of CT scans because of streak artifacts that occur predominantly in scans of the upper abdomen. Scanning with the patient's arms raised eliminates the streak artifacts in scans of the abdominal region. With the new generation of PET–CT devices equipped with MDCT scanners, breathing motion artifacts can be expected to be eliminated if protocols for breath-hold CT are applied. Reversing the direction of CT scanning allows one to avoid imaging the thoracic region at a time when undiluted IV contrast material is still present in the upper thoracic veins.
Using functional data from FDG positron emission tomography (PET) has gained increasing acceptance as a means for the staging and restaging of many types of tumors as well as for monitoring therapeutic effects [1, 2]. However, identifying the exact anatomic location of metabolic abnormalities is often difficult on PET scans, specifically on those obtained in the head and neck region, mediastinum, abdomen, or pelvis. Recently, a combined PET–CT system was developed that can provide coregistration of both functional and morphologic information in a single imaging procedure [3]. Two different approaches may be used with in-line CT in PET–CT scanners [4, 5]. In the first, low-radiation CT is performed primarily to obtain transmission data for attenuation correction by extrapolating the CT data to 511 keV [6]. With this approach, oral or IV contrast material is typically not used as part of the CT protocol. The second approach involves diagnostic CT performed with administration of IV and oral contrast materials, as proposed by Antoch et al. [7]. However, in conventional CT, a whole-body protocol requires imaging several anatomic segments sequentially with the appropriate timing of IV contrast administration. Typically, the thoracic, abdominal, and pelvic CT scans are obtained with the patient's arms raised to avoid streak artifacts. In contrast, we performed CT as a single scan from the base of the skull to the groin with the patient's arms at the side of the body because many patients find it uncomfortable to keep their arms raised above the head for the entire PET–CT examination. The optimal protocol for PET–CT remains a subject of scientific investigation.
We used a PET–CT device that consists of a single-detector helical Somatom Emotion CT scanner (Siemens Medical Solutions) and a modified ECAT HR+ PET scanner (CPS Innovations). The region of the body being imaged determines the length of the CT scanning delay required to optimize soft-tissue contrast with IV iodinated contrast material [8]. Because we scanned the patients in the craniocaudal direction from the skull to the groin, IV contrast material was administered with a scanning delay of 30 sec to optimize portal venous liver contrast, resulting in a scanning delay for the liver of approximately 70–80 sec. However, this procedure often resulted in undiluted vascular contrast material in the upper thoracic veins, impairing the diagnostic image quality. Single-detector helical CT scanning from the base of the skull to the groin takes about 80–100 sec; the length of the scanning time precludes the use of breath-hold imaging. PET scanning, on the other hand, averages the effect of normal breathing motion because scans are acquired over a period of 5 min per bed position. Thus, our protocol included shallow breathing for the CT portion of combined PET–CT imaging.
The aim of our study was to assess the diagnostic quality of CT images acquired in a single scan during PET–CT scanning with a single-detector helical CT scanner. The CT scans from PET–CT studies were retrospectively reviewed by two radiologists who evaluated the diagnostic image quality of scans acquired in 81 consecutive patients with lymphoma. Specifically, the radiologists analyzed the streak artifacts caused by scanning with the patient's arms beside the body and streak artifacts caused by highly concentrated contrast material in the upper thoracic veins. Furthermore, we assessed the influence of shallow breathing during CT on the diagnostic quality of the CT scans.
Materials and Methods
Patients
PET–CT was performed in 81 consecutive patients (44 males, 37
females; age range, 9–89 years; mean age, 53 years) for the staging or
restaging of lymphoma. In 75 of the 81 patients, PET–CT was performed
using IV iodinated nonionic contrast material. We did not administer IV
contrast material to patients with a history of adverse allergic reactions,
hyperthyroidism, or renal failure. No adverse reactions to IV contrast
material were observed among our patients. Positive iodinated oral contrast
material (MD-Gastroview [diatrizoate meglumine and diatrizoate sodium solution
U.S.P.], Mallinckrodt; diluted with water to 2.4% weight/volume) was used for
all studies except one.
PET–CT
Our PET–CT scanner is a modified ECAT HR+ scanner combined with a
single-detector Somatom Emotion CT scanner. The modified ECAT HR+ PET scanner
consists of 32 rings of bismuth germanate detectors that yield 63 transverse
slices, each 2.425-mm apart (axial field of view, 15.3 cm). This scanner
allows PET data to be acquired in 3D-mode only. The patient port was increased
from 56 cm (used in the standard ECAT HR+ PET scanner) to 70 cm to match the
opening of the CT scanner.
Patients were asked to fast for at least 4–6 hr before the injection
of 163–487 MBq of FDG into a cubital vein with a 20-gauge needle. After
the FDG injection, patients were seated in a comfortable chair and asked to
rest for approximately 1 hr. In addition, for optimal delineation of bowels,
the patients were asked to ingest 400–600 mL of oral contrast material 1
hr before the initiation of the CT examination. CT was performed before PET in
a single step, with the patients in the supine position with their arms beside
the body and their palms in contact with the gluteal region. An
anteroposterior CT scout scan (topogram) was obtained to plan the whole-body
CT data acquisition as well as the whole-body PET examination. IV contrast
material using 125 mL of Optiray 350 ([ioversol] 350 mg/mL of organically
bound iodine, Mallinckrodt) was administered IV at a flow rate of 2.5 mL/sec
with a 30-sec scanning delay (resulting in a scanning delay for the liver of
70–80 sec).
The scanning parameters for whole-body CT craniocaudal scanning were 130 kV; 130 mAs; rotation time, 0.8 sec; collimation, 5 mm; and pitch, 1.5. During the scanning, patients were asked to breathe shallowly.
Evaluation of Scans
For image interpretation, the CT and PET scans were transferred to a Syngo
viewing station (Siemens Medical Solutions) where they were evaluated
interactively on the monitor by two experienced radiologists who scrolled
through each slice. The whole-body scans were displayed at soft-tissue window
settings (level, 50 H; width, 350 H), whereas the evaluation of the thorax was
displayed at lung window settings (level, –600 H; width, 1,700 H).
Several aspects of image quality were evaluated. The radiologists rated the streak artifacts seen across the midline soft-tissue anatomy that resulted from the position of a patient's arms (beside the body) during scanning at the level of the upper thorax, lung and mediastinum, liver and upper abdomen, middle abdomen, and pelvis. As suggested by Hu et al. [9], a 4-point scale was used to rate the artifacts: 0, absent; 1, mild; 2, moderate; and 3, severe. Artifacts were rated as being absent (0) if the scan was virtually free of image degradation. Artifacts were rated mild (1) if less than 5% of the evaluated region had streaks that did not obscure any critical structures or result in the image degradation that precluded correct interpretation of the scan. Artifacts were rated as moderate (2) on the basis of each radiologist's subjective judgment of a degree of streaking that was between minimal and severe. Artifacts were rated as severe (3) if streaks appeared on more than 10% of the evaluated region, making diagnostic interpretation impossible.
Undiluted contrast material in the axillary, subclavian, and brachiocephalic veins resulted in streak artifacts that degraded image quality at the level of the upper thorax. These artifacts were rated on the same 4-point scale previously described as was the effect of breathing on CT image quality on the scans of the lung and mediastinum, liver and upper abdomen, and middle abdomen [10]. To correlate the artifact ratings with the diameter of the patient's body habitus, we measured the transverse diameter of the upper abdomen at the level of the upper pole of the right kidney with a software-implemented caliper. The transverse diameter of the pelvis was measured at the level of the hip-joint space. The artifact ratings were also correlated with the patient's weight.
Statistical Analysis
The ordered logistic regression model was used to analyze the correlation
of a patient's weight and diameter with the ratings of streak artifacts caused
by the placement of the patient's arms. We used the logistic model to predict
the probability that severe streak artifacts would appear on the scans
depending on the patient's weight.
Results
Artifacts caused by the position of a patient's arms resulted in decreased image quality in the single-detector whole-body CT scans of all the patients (Table 1). The most pronounced image degradation was observed in scans of the upper abdomen, specifically in the liver. Moderate streak artifacts in 40% of the CT scans and severe streak artifacts in 38% impaired the diagnostic value of the images. Figure 1A, 1B illustrates the image degradation from streak artifacts in the upper abdomen. In scans of this region, important anatomic structures such as the right lobe of the liver, the spleen, both adrenal glands, and the kidneys were obscured. In the mid abdomen, artifacts were graded predominantly as mild (57%); severe artifacts were observed in only 8% of the CT scans. Similar results were found in the scans of the pelvis (mild, 58%; moderate, 32%; and severe, 8%) and the upper thorax (mild, 56%; moderate, 25%; and severe, 6%). For the mediastinum and lungs, 38% of the scans were not affected by artifacts. The artifacts in 37% of the scans of this region were graded as mild; artifacts in 17%, as moderate; and 9%, as severe. Because the patients were positioned in a standardized way (with the palms contacting the gluteal region), artifacts were predominately observed in the middle and posterior parts of the trunk.
|
|
|
In the upper thorax, the mediastinum and lungs, the liver and upper abdomen, and the mid abdomen, the correlation between the intensity of the streak artifacts and the patient's weight was statistically significant (p < 0.05). However, no correlation was found between the patient's weight and the level of image degradation in the pelvis. The correlation between the transverse diameter of the upper abdomen and the intensity of streak artifacts was statistically significant (p < 0.05). In patients weighing more than 100 kg (220.5 lb), there was an approximately 90% probability of severe image quality degradation at the level of the liver and upper abdomen due to streak artifacts.
Streak artifacts caused by undiluted contrast medium in the axillary, subclavian, or brachiocephalic veins were evaluated in the scans of the 75 patients who received IV contrast material. Severe artifacts through the upper thorax were noted in 57% of these scans (Table 2 and Fig. 2A, 2B, 2C, 2D, 2E, 2F). Mild streak artifacts were observed in 10%, and moderate artifacts were seen in 27%. In the 6% of CT scans with no artifacts from contrast material, the start of scanning had been delayed for technical reasons, allowing time for the contrast material to become diluted or to wash out of the upper thoracic veins.
|
|
|
|
|
|
|
The diagnostic quality of the CT scans was also strongly affected by artifacts from breathing motion during the scanning (Table 3). Especially in liver scans, the diagnostic image quality was impaired by moderate (49%) or severe (28%) double contours and blurring as well as discontinuities caused by breathing motion. In images of the lungs and the mediastinum, moderate image quality degradation from breathing motion artifacts was observed in 49% of the CT scans, and severe breathing artifacts were found in 9%. Visualization of anatomic structures in the mid abdomen— such as the pancreas, the celiac trunk, and the superior mesenteric artery and vein—was less impaired, with 23% of the CT scans showing moderate artifacts and 3% showing severe artifacts. Breathing motion resulted in misregistration between the lesions on CT scans and the lesions on PET scans. Misregistration of lesions in the lower lungs and the liver was especially severe (Fig. 3A, 3B, 3C).
|
|
|
|
Discussion
Performing whole-body CT in one step with a patient's arms positioned beside the body resulted in a considerable number of streak artifacts, particularly in scans of the upper abdomen. Of 81 consecutive patients with lymphoma, 78% had scans with moderate or severe artifacts that impaired the diagnostic evaluation of the liver. The upper abdomen was the most commonly affected region because at this level, X-ray beams traverse across several high-density structures such as the elbows and the spine. The intensity of streak artifacts was correlated with the patient's weight. In patients weighing more than 100 kg (220.5 lb), we found an approximately 90% probability of substantial artifacts appearing on scans obtained with the patient's arms beside the body. The image quality degradation in CT scans of the upper abdomen could be prevented by scanning with the patient's arms raised. However, for PET–CT scanning of patients with suspected head and neck cancer, the arms-raised position is likely to be inadequate. In addition, imaging patients with melanoma whose primary tumor is in the upper extremities still requires positioning the arms beside the body.
An alternative solution to avoid streak artifacts was proposed by Hany et al. [11] who performed PET–CT with the patients' arms elevated above the abdomen. Another approach requires a higher tube current (adapted to the patient's weight) than that normally used. Protocols for thoracic and abdominal CT scanning using 220–260 mAs have been reported [12]. However, the tube current is limited in whole-body CT scanning because of the need for tube cooling, especially in single-detector CT scanners. A higher tube current results in higher radiation exposure to the patient. The radiation exposure may be reduced using attenuation-based online modulation of the tube current, a technologic advance that is now commercially available on some CT scanners [13]. In addition, different adaptive filtering approaches may be used to reduce streak artifacts by appropriate postprocessing [14, 15].
Scanning the whole body from the base of the skull to the groin on a single-detector scanner takes approximately 100 sec. Obviously, breath-hold scanning is not possible. We found that the image quality of single-detector CT scans acquired during shallow breathing was greatly impaired by breathing motion artifacts in the lung and mediastinum, the liver and upper abdomen, and the mid abdomen. In 77% of the cases, moderate or severe blurring or double contours occurred at the level of the liver and the upper abdomen. In addition, breathing motion during the data acquisition resulted in discontinuities in the structures imaged. Thus, small lesions in the liver or the lower lungs may be missed. Recently, Goerres et al. [16, 17] found CT scans acquired at normal expiration fit best with PET scans acquired during shallow breathing. However, their studies were performed with a 4-MDCT scanner, so patients were asked to suspend breathing during a scanning period of approximately 20 sec.
Equipping combined PET–CT units with MDCT will reduce the limitations encountered in conventional CT. MDCT techniques allow significant reduction in overall scanning time per patient. For example, using a 4-MDCT scanner at half the rotation speed in PET–CT reduces the time required for scanning by a factor of 8 [18] and should significantly decrease breathing artifacts. These are important benefits to note because the use of MDCT in PET–CT will increase the cost. However, we believe that with MDCT, PET–CT can be used for diagnosis, obviating an additional CT scan.
In our PET–CT protocol, the administration of contrast material was timed so that the liver could be imaged during the portal venous phase. Scanning in the craniocaudal direction, we used a 30-sec scanning delay, resulting in a scanning delay for the liver of approximately 70–80 sec. Streak artifacts from undiluted contrast material in the upper thoracic veins degraded the diagnostic image quality. In 84% of the PET–CT studies, moderate or severe artifacts marred the depiction of the upper thorax to a degree that made adequate evaluation impossible. Furthermore, because the CT portion of PET–CT is used for attenuation correction of the PET data, high-density areas from IV contrast material in CT scans can cause artifacts in PET images. Thus, overestimation of the attenuation in PET scans resulted in an artificially increased FDG uptake on attenuation-corrected PET scans that was not seen on the non–attenuation-corrected PET scans (Fig. 2A, 2B, 2C, 2D, 2E, 2F). Such artifacts were previously described by Antoch et al. [19] in the scans obtained in four of 30 patients studied with IV contrast material. However, those authors did not analyze the level of degradation of CT image quality attributable to streak artifacts. Changing the scanning direction from craniocaudal to caudocranial provides a reasonable solution, although the PET–CT scanner used for our study did not permit this scanning mode. A scanning delay of approximately 40 sec starting at the symphysis pubis results in optimal portal venous contrast in the liver; the contrast material in the axillary, subclavian, and brachiocephalic veins is diluted by venous blood flow because contrast injection ends approximately 30 sec before the upper thorax is scanned. As an alternative, a saline flush after contrast injection may be used to clear the hyperdense contrast material from the axillary, subclavian, and brachiocephalic veins [20].
In conclusion, our initial PET–CT protocol using a single-detector CT scanner resulted in CT image quality degradation from streak artifacts and breathing motion. We speculate that many of these problems can be eliminated with MDCT, but this supposition remains to be proven. Artifacts resulting from the patient's arms being positioned beside the body can be eliminated either by elevating the arms above the head or, after the implementation of the MDCT, using higher tube currents. Artifacts from undiluted contrast material in the upper thoracic veins can be prevented by scanning the patient in the caudocranial direction. Otherwise, modification of contrast injection parameters combined with a saline flush could eliminate this problem. Breath-hold CT scans obtained at the level of normal expiration are likely to match the PET scans obtained with shallow breathing. Only MDCT scanners capable of acquiring whole-body scans in less than 30 sec can allow the introduction of breath-hold techniques.
Acknowledgments
We thank Jonathan P. J. Carney of the Department of Medicine at the University of Tennessee, Knoxville, and Marc Kachelriess of the Institute of Medical Physics at the University of Erlangen-Nuremberg, for helpful discussions and for critical comments on this article.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  D. W. Townsend Dual-Modality Imaging: Combining Anatomy and Function J. Nucl. Med., June 1, 2008; 49(6): 938 - 955. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
