Published online before print December 20, 2002, 10.1148/radiol.2262012141
(Radiology 2003;226:577-584.)
© RSNA, 2003
Artifacts at PET and PET/CT Caused by Metallic Hip Prosthetic Material1
Gerhard W. Goerres, MD,
Sibylle I. Ziegler, PhD,
Cyrill Burger, PhD,
Thomas Berthold,
Gustav K. von Schulthess, MD, PhD and
Alfred Buck, MD
1 From the Division of Nuclear Medicine, University Hospital Zurich, Rämistrasse 100, CH-8091 Zurich, Switzerland (G.W.G., C.B., T.B., G.K.v.S., A.B.); and Nuclear Medicine Clinics, Technischen Universität München, Munich, Germany (S.I.Z.). Received January 9, 2002; revision requested March 4; revision received May 1; accepted June 5. G.W.G. supported by an award from the Research and Education Fund of the European Association of Radiology. Address correspondence to G.W.G. (e-mail: gerhard.goerres@dmr.usz.ch).
 |
ABSTRACT
|
|---|
Hip prosthetic material and a steel rod were scanned in a water bath of fluorine 18 fluorodeoxyglucose (FDG) with positron emission tomographic (PET) and PET/computed tomographic (CT) scanners to evaluate the generation of artifacts adjacent to the metal. The influences of attenuation correction (AC), positioning of the object, and image reconstruction were examined. Use of CT- and germanium 68–based AC resulted in generation of artifacts that mimicked increased FDG uptake. These artifacts were more evident when the object was moved between the emission and transmission scans. When attenuation-weighted iterative reconstruction was used, these artifacts were less evident.
© RSNA, 2003
Index terms: Images, processing • Phantoms • Positron emission tomography (PET), artifact • Stents and prostheses, 442.454 • Test objects
|
INTRODUCTION
|
|---|
During recent years, positron emission tomography (PET) with fluorine 18 fluorodeoxyglucose (FDG) has become a standard method for the evaluation of patients with heart disease, brain disease, or cancer. FDG PET has also been used for the evaluation of patients with infection and inflammation, and it has been shown that it may be a highly sensitive method for the detection of infectious foci (1). In addition, it has been suggested that FDG PET can be used successfully for the evaluation of infected prosthetic material (2,3). However, it has also been shown that metallic prosthetic material can cause artifacts in FDG PET images (4,5). Heiba et al (4) observed an artifact within the joint space of total knee metallic prostheses in two patients on attenuation-corrected PET images. Since such artifacts mimic increased FDG uptake adjacent to the metal, this pitfall can have a substantial effect on the reliability of interpretation of PET images obtained in patients suspected of having infection adjacent to metal implants.
The transmission scan is usually performed with built-in germanium 68 (68Ge) sources and is needed for the quantitative measurement of activity concentrations in tissues. With the introduction of combined in-line PET/CT cameras, it is also possible to perform attenuation correction (AC) with a computed tomographic (CT) scan (6). Since it is known that the quality of CT images is deteriorated by the presence of metal parts in the body, it can be assumed that the use of CT-based attenuation maps could result in the introduction of artifacts on PET images obtained when AC was performed with CT.
The aim of this study was to evaluate the influence of CT-based and 68Ge-based transmission imaging on the introduction of artifacts in the final PET image in the presence of metallic prosthetic material.
|
Materials and Methods
|
|---|
PET Cameras
This evaluation was conducted by means of studies of prostheses and a steel rod phantom performed with a combined in-line PET/CT scanner (Discovery LS; GE Medical Systems, Milwaukee, Wis) and a full-ring PET scanner (ECAT Exact HR+; Siemens CTI, Knoxville, Tenn). The in-line PET/CT camera has an transverse field of view of 14.6 cm and a helical multi-detector CT scanner. Transmission imaging can be performed by using either the built-in rotating 68Ge sources or a helical CT scan. With this camera, the positioning of the prostheses for transmission scanning with CT was performed by changing the table position. Attenuation maps and PET emission scan images were automatically aligned. Emission and transmission scans were performed in two-dimensional mode. One experiment was repeated with the ECAT Exact HR+ PET scanner, which consists of a three-ring system with a 16.2-cm field of view. With this scanner, transmission data were first acquired with the conventional 68Ge-based method; images were then obtained in two-dimensional mode.
Prostheses and Metallic Rod Phantom
Prosthetic material used for total hip replacement and a steel rod phantom were used in the PET measurements. Two prostheses, which consisted of a femoral stem and the corresponding acetabular component (Sulzer Orthopedics; Sulzer Medica, Muensingen, Switzerland) (Fig 1), and four femoral stems with different shapes and surface properties (Stratec Medical, Oberdorf, Switzerland) (Fig 1) were measured. Three of these prostheses were made of titanium, and three were made of steel alloys. The phantom (constructed by G.W.G.) consisted of a round stainless steel rod with a diameter of 4 cm and a length of 20 cm (Fig 1).
View larger version (53K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. Prostheses from two different manufacturers were scanned. Two (Sulzer Medica) consisted of a femoral stem and the corresponding acetabular component: A, the Allo Pro Ti-Al-Nb 920206-4 titanium; and B, the Allo Pro A 907 257 steel. Four femoral stems (Stratec Medical) with different shapes and surface properties were also measured: C, Proxilock FT Mat 190 5832-3 titanium; D, Manistream Müller Mat 190 5832-9 steel; E, Manistream 30 Mat 130 5832-9 steel; F, PPF Mat 5832-3 titanium. G, A steel rod with a circular cross-sectional profile (diameter, 4 cm) was also scanned.
|
|
Experiments
Measurements were performed with the hip prosthetic material and the steel rod to identify the influence of the following parameters on the appearance of artifacts:
1. Use of conventional AC as performed with two different types of scanners, type of metal (titanium vs steel), and shape of prosthesis.
2. Use of conventional 68Ge-based AC versus use of CT-based AC.
3. Position of the phantom within the water tank and relative to the gantry.
4. Radioactivity concentration around the metal.
5. Duration of emission and transmission scans.
6. Tube current (in milliamperes) with CT-based AC.
7. Patient movement, which was simulated by repositioning a prosthesis 3 mm to the right between emission and transmission scans.
The experiments are listed in Table 1 and were performed according to the following set-up:
1. In experiment 1, measurements of all prostheses were performed with the Discovery LS scanner by using a 36 x 30 x 16-cm cubic water tank. The prosthetic shafts and acetabular components had different surface properties and shapes. Furthermore, some of the measured prostheses were composed of titanium, and some were composed of steel. FDG was added to the water to obtain a solution with a radioactivity concentration comparable to that of soft tissue in a patient—that is, a radioactivity concentration was chosen to match the tracer activity of about 5.3 kBq/mL in soft tissues that can be expected after the injection of 370 MBq of FDG in a 70-kg patient.
Each hip prosthesis was placed in this basin, which was filled with 8 liters of water, and the shaft of the prosthesis was aligned along the z axis (corresponding to the axis of a patient’s femur) within the field of view of the scanner. This part of the study was performed with a 4-minute emission scan and a 2-minute transmission scan; this equals the imaging time for patient examinations in our institution. In addition, a 4-minute transmission scan was performed. The metal rod, which has a circular cross-sectional profile, was scanned in the same water basin with identical parameters. In this experiment, two types of hip prostheses (the Allo Pro Ti-Al-Nb 920206-4 titanium prosthesis and the Allo Pro A 907 257 steel prosthesis) and the metal rod were also imaged with the ECAT Exact HR+ scanner by using a 10-minute emission scan and a 10-minute transmission scan.
2. In experiment 2, AC for the same prostheses was performed with a CT scan.
3. In experiment 3, the same measurements were performed with the steel rod, which has a circular cross-sectional profile, by using a larger oval-shaped water basin (12 liters). This permitted positioning of the rod once in the middle of the basin and once at the border of the basin. This was performed to evaluate the influence of photon absorption by the surrounding "tissues" on the quality of the final PET image. One prosthesis was also evaluated by means of turning it 90° so that its long axis was aligned to the transverse plane through a patient; the measurement was then repeated with the same acquisition parameters. This was performed to assess the dependence of artifacts on the spatial orientation of the prosthesis in the field of view of the scanner.
4. In experiment 4, two hip prostheses consisting of a femoral stem and the corresponding acetabular component were measured in the same water bath with three different radioactivity concentrations: 1.25, 2.50, and 3.75 kBq/mL.
5. In experiment 5, different data acquisition times were evaluated. With the Discovery LS scanner, data were acquired during either a 4-minute or a 10-minute emission scan followed by a 2-, 4-, 10-, or 180-minute conventional transmission scan.
6. In experiment 6, AC data were obtained by means of CT scans, which were always performed with 140 kV but different tube currents: 10, 40, 80, and 120 mA.
7. In experiment 7, the transmission scan was performed after the prosthesis had been repositioned 3 mm to the right. In this way, patient movement between the emission and transmission scans was simulated. Transmission scanning was performed with the conventional method as well as with CT scanning. CT scans were again performed with four different tube currents (10, 40, 80, and 120 mA).
Image Reconstruction
With both scanners, image reconstruction was performed with the same protocols as those used for patient studies in the two institutions taking part in this evaluation. In all experiments, the attenuation data were reconstructed by using FBP. The 68Ge transmission images were segmented, and predefined attenuation values for tissue and lung were assigned to the corresponding areas. For CT-based attenuation data, no segmentation was applied. All emission images corrected for attenuation were reconstructed with FBP and with an iterative ordered-subset expectation maximization (OSEM) algorithm programmed into the PET camera for routine clinical use.
With the GE scanner, images with a zoom of 1.0 were reconstructed with FBP and an iterative OSEM-based algorithm that included 28 subsets and two iterations. IR was performed with post filtering of 5.42 mm and a loop filter of 3.91 mm. FBP was performed with a Hanning filter (cutoff, 7.7 mm). The transmission data were reconstructed and smoothed with the same filters that were used with the emission data to reduce statistical noise in the images from these short transmission scans (7). After segmentation, the images were forward projected and saved as attenuation factor sinograms. With the GE camera, emission sinograms were corrected for dead time, random coincidences, and attenuation before FBP or IR was performed.
CT data were obtained in a 512 x 512 image matrix and redimensioned to an image matrix of 128 x 128 for adaptation to the PET emission scans. The CT-based attenuation data were also smoothed by using the same Gaussian filter. The CT pixel values in Hounsfield units were transformed into linear attenuation coefficients in units of cm-1 at 511 keV by means of a bilinear function defined by three coordinate points: -1,000 HU, 0 cm-1; 0 HU, 0.093 cm-1; and 1,326 HU, 0.172 cm-1. These attenuation images were then forward projected according to the geometry of the PET scanner, and the calculated line integrals were exponentiated to obtain the AC factors. The resulting attenuation factor sinograms were smoothed with an 8-mm Gaussian filter to adjust the correction data to the PET resolution. Image reconstruction was again performed after the sinograms were corrected for attenuation. All images were reconstructed by using the software provided by the manufacturer (ie, GE software release 5.0).
Transmission data from the ECAT scanner were also reconstructed with FBP. In contrast to the approach of the GE image reconstruction software, an attenuation-weighted OSEM algorithm (eight iterative steps, four subsets, zoom 1.5) was used to reconstruct emission data with the ECAT scanner. Sinograms were not precorrected for attenuation, but attenuation factors from the segmented transmission image were included in the system matrix for the statistical OSEM algorithm. It has been shown that this approach avoids the generation of artifacts in regions of extreme attenuation factors because it incorporates the use of an accurate statistical model (8).
Data Analysis
All images were displayed and visually assessed by two nuclear medicine physicians (G.W.G. and A.B.) and by two physicists (S.I.Z. and C.B.). The intensity of artifacts in the attenuation-corrected PET images and differences in artifact appearance between CT-based and 68Ge-based AC images were evaluated. For all experiments, a consensus was reached among the readers in the definition of differences in the visual intensity of artifact appearance. Additionally, activity profiles were measured in all experiments by using commercially available software (PMOD V2.3; Division of Nuclear Medicine of University Hospital Zurich, Zurich, Switzerland, available at www.pmod.com) (9).
|
Results
|
|---|
Results of the experiments are listed in Table 2. All figures are shown as inverted black-and-white images and are leveled to the same gray scale with a lower level of 0. Throughout the Results section, the term artifact describes an increase in measured radioactivity concentration within or adjacent to the metal of the phantom or prosthesis.
The results of the experiments were as follows:
1. Measurements of all prostheses in the FDG water bath revealed that artifacts arose in non–attenuation-corrected PET images when FBP was used for image reconstruction (Fig 2). These artifacts were located within the prosthetic material itself. Artifacts adjacent to the metallic implants were generated only in the attenuation-corrected images and were evident in areas with steep differences in attenuation values between the metal and the surrounding water (Fig 2). Therefore, the shape of a prosthesis was an important factor in the generation of artifacts. Small (approximately 1-mm) irregularities of the surface, such as in the prostheses depicted in C and D in Figure 1, did not influence the appearance of artifacts. In contrast, the uneven shape (diameter, 5–10 mm) of prosthesis B in Figure 1 resulted in an artifact being generated at the anterior and posterior surfaces (Fig 2). Artifacts were visible with both titanium and steel prostheses (Fig 2).
View larger version (80K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2. Coronal PET images of the prosthesis shown in Figure 1, B, in an FDG water bath. A, Non-attenuation-corrected PET image reconstructed with FBP (from a 4-minute emission scan) shows an artifact located within the prosthetic material itself. The activity profile below the PET image, obtained along the horizontal line in the PET image, shows the measured radioactivity concentration in relation to the background in the water tank. B, PET image obtained after AC was applied with the 68Ge sources (during a 4-minute transmission scan) that was reconstructed with segmentation shows that artifact has been generated adjacent to the metallic implant. Artifact is evident in areas with steep differences in attenuation values between the metal and the surrounding water—for example, the shoulder of the prosthesis (long arrow) or the anterior and posterior surfaces (short arrows) of the prosthesis, which have an uneven shape. The activity profile, obtained along the horizontal line in the PET image, shows that the measured FDG concentration in the artifact is approximately twice that of the water basin. C, On a PET image obtained after AC was applied (during a 4-minute emission scan) and with IR, no artifacts are visible. D, On a PET image obtained after 68Ge-based AC was applied and with IR, artifact (arrows) due to partial volume mapping is seen adjacent to the metal. The activity profile, obtained along the horizontal line in the PET image, shows that the measured FDG concentration is about twice that of the surrounding water. E, PET image obtained with the ECAT Exact HR+ scanner also shows artifact (arrows), underlining the fact that the generation of these artifacts is an inherent problem of AC. However, artifact is less marked on this image; this is due to the different method of image reconstruction (ie, an attenuation-weighted IR algorithm) used with this scanner.
|
|
Measurements of the prostheses and the steel rod with the ECAT Exact HR+ scanner revealed that, when this scanner was used, artifacts were visible at the same locations adjacent to the prosthesis at which artifacts were observed when the Discovery LS scanner was used (Fig 2). Therefore, the generation of artifacts is an inherent problem of AC. However, we found that an artifact was somewhat less evident when the ECAT Exact HR+ scanner was used.
2. With the use of CT-based AC, artifacts were also evident. In all experiments, background noise was much more evident when FBP was used, and, therefore, artifacts were somewhat less visible with FBP because of noise in the image (Fig 2).
3. When a CT scan was used for AC, repositioning of the metal rod to the border of the water basin led to a different appearance of the artifacts (Fig 3). The difference between CT-based and 68Ge-based AC is evident in a review of images from transmission scans (Fig 3). Rotation of a prosthesis 90° in the horizontal plane demonstrated an influence of the geometric relationship between the prosthesis and the transmission source (Fig 3). Artifacts changed in position along the prosthesis to the region with the steepest change in attenuation values between metal and surrounding water according to the axis in which the transmission scan was performed (Fig 3).
View larger version (53K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a. PET images of the steel rod phantom. (a) On images obtained with 4-minute emission scan and 180-minute transmission scan and reconstructed with segmentation and IR, a black circlelike artifact is visible around the steel rod in the large water basin. With longer-duration conventional transmission scan, image noise is reduced and artifact is somewhat more visible (see c). (b) On images from 4-minute emission scan with CT-based AC and IR, absorption of photons by the water tank has created inhomogeneities in the attenuation map that have influenced the appearance of artifacts adjacent to the metal. (c) Images obtained with 4-minute emission scan and 4-minute transmission scan and reconstructed with IR shows that repositioning of the steel rod within the water tank did not affect artifacts generated with conventional AC. (d) Images from the corresponding 68Ge-based transmission scan illustrates quality of attenuation map. (e) In contrast to c, these images show that repositioning of the steel rod within the water basin led to a change in artifact appearance when CT-based AC was used. Therefore, it can be expected that attenuation properties in a patient will be more important when CT-based AC is used. (f) Images from the corresponding transmission scan (40 mAs) with CT-based AC show that CT-based AC creates a high-quality attenuation map with less noise. However, adjacent to the metal is an artifact (arrows) with a spiral-like appearance. In addition, the absorption of photons in the water tank has created inhomogeneities in the attenuation map that are visible as white stripes. They influenced artifact appearance on the final reconstructed PET image in e. (g) Images obtained with a 4-minute emission scan and a 2-minute transmission scan and reconstructed with segmentation before (left) and after (right) rotation of the hip prosthesis show that rotation of the item within the same coronal plane has changed the position of the artifact (arrows) along the hip prosthesis because the geometric relationship between the radiation source and the steep edges of the prosthesis has changed. These images are set at a very dark gray-scale level to show that after rotation, the artifact is more visible at the shoulder of the prosthesis than at the femoral shaft.
|
|
View larger version (57K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b. PET images of the steel rod phantom. (a) On images obtained with 4-minute emission scan and 180-minute transmission scan and reconstructed with segmentation and IR, a black circlelike artifact is visible around the steel rod in the large water basin. With longer-duration conventional transmission scan, image noise is reduced and artifact is somewhat more visible (see c). (b) On images from 4-minute emission scan with CT-based AC and IR, absorption of photons by the water tank has created inhomogeneities in the attenuation map that have influenced the appearance of artifacts adjacent to the metal. (c) Images obtained with 4-minute emission scan and 4-minute transmission scan and reconstructed with IR shows that repositioning of the steel rod within the water tank did not affect artifacts generated with conventional AC. (d) Images from the corresponding 68Ge-based transmission scan illustrates quality of attenuation map. (e) In contrast to c, these images show that repositioning of the steel rod within the water basin led to a change in artifact appearance when CT-based AC was used. Therefore, it can be expected that attenuation properties in a patient will be more important when CT-based AC is used. (f) Images from the corresponding transmission scan (40 mAs) with CT-based AC show that CT-based AC creates a high-quality attenuation map with less noise. However, adjacent to the metal is an artifact (arrows) with a spiral-like appearance. In addition, the absorption of photons in the water tank has created inhomogeneities in the attenuation map that are visible as white stripes. They influenced artifact appearance on the final reconstructed PET image in e. (g) Images obtained with a 4-minute emission scan and a 2-minute transmission scan and reconstructed with segmentation before (left) and after (right) rotation of the hip prosthesis show that rotation of the item within the same coronal plane has changed the position of the artifact (arrows) along the hip prosthesis because the geometric relationship between the radiation source and the steep edges of the prosthesis has changed. These images are set at a very dark gray-scale level to show that after rotation, the artifact is more visible at the shoulder of the prosthesis than at the femoral shaft.
|
|
View larger version (54K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3c. PET images of the steel rod phantom. (a) On images obtained with 4-minute emission scan and 180-minute transmission scan and reconstructed with segmentation and IR, a black circlelike artifact is visible around the steel rod in the large water basin. With longer-duration conventional transmission scan, image noise is reduced and artifact is somewhat more visible (see c). (b) On images from 4-minute emission scan with CT-based AC and IR, absorption of photons by the water tank has created inhomogeneities in the attenuation map that have influenced the appearance of artifacts adjacent to the metal. (c) Images obtained with 4-minute emission scan and 4-minute transmission scan and reconstructed with IR shows that repositioning of the steel rod within the water tank did not affect artifacts generated with conventional AC. (d) Images from the corresponding 68Ge-based transmission scan illustrates quality of attenuation map. (e) In contrast to c, these images show that repositioning of the steel rod within the water basin led to a change in artifact appearance when CT-based AC was used. Therefore, it can be expected that attenuation properties in a patient will be more important when CT-based AC is used. (f) Images from the corresponding transmission scan (40 mAs) with CT-based AC show that CT-based AC creates a high-quality attenuation map with less noise. However, adjacent to the metal is an artifact (arrows) with a spiral-like appearance. In addition, the absorption of photons in the water tank has created inhomogeneities in the attenuation map that are visible as white stripes. They influenced artifact appearance on the final reconstructed PET image in e. (g) Images obtained with a 4-minute emission scan and a 2-minute transmission scan and reconstructed with segmentation before (left) and after (right) rotation of the hip prosthesis show that rotation of the item within the same coronal plane has changed the position of the artifact (arrows) along the hip prosthesis because the geometric relationship between the radiation source and the steep edges of the prosthesis has changed. These images are set at a very dark gray-scale level to show that after rotation, the artifact is more visible at the shoulder of the prosthesis than at the femoral shaft.
|
|
View larger version (50K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3d. PET images of the steel rod phantom. (a) On images obtained with 4-minute emission scan and 180-minute transmission scan and reconstructed with segmentation and IR, a black circlelike artifact is visible around the steel rod in the large water basin. With longer-duration conventional transmission scan, image noise is reduced and artifact is somewhat more visible (see c). (b) On images from 4-minute emission scan with CT-based AC and IR, absorption of photons by the water tank has created inhomogeneities in the attenuation map that have influenced the appearance of artifacts adjacent to the metal. (c) Images obtained with 4-minute emission scan and 4-minute transmission scan and reconstructed with IR shows that repositioning of the steel rod within the water tank did not affect artifacts generated with conventional AC. (d) Images from the corresponding 68Ge-based transmission scan illustrates quality of attenuation map. (e) In contrast to c, these images show that repositioning of the steel rod within the water basin led to a change in artifact appearance when CT-based AC was used. Therefore, it can be expected that attenuation properties in a patient will be more important when CT-based AC is used. (f) Images from the corresponding transmission scan (40 mAs) with CT-based AC show that CT-based AC creates a high-quality attenuation map with less noise. However, adjacent to the metal is an artifact (arrows) with a spiral-like appearance. In addition, the absorption of photons in the water tank has created inhomogeneities in the attenuation map that are visible as white stripes. They influenced artifact appearance on the final reconstructed PET image in e. (g) Images obtained with a 4-minute emission scan and a 2-minute transmission scan and reconstructed with segmentation before (left) and after (right) rotation of the hip prosthesis show that rotation of the item within the same coronal plane has changed the position of the artifact (arrows) along the hip prosthesis because the geometric relationship between the radiation source and the steep edges of the prosthesis has changed. These images are set at a very dark gray-scale level to show that after rotation, the artifact is more visible at the shoulder of the prosthesis than at the femoral shaft.
|
|
View larger version (41K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3e. PET images of the steel rod phantom. (a) On images obtained with 4-minute emission scan and 180-minute transmission scan and reconstructed with segmentation and IR, a black circlelike artifact is visible around the steel rod in the large water basin. With longer-duration conventional transmission scan, image noise is reduced and artifact is somewhat more visible (see c). (b) On images from 4-minute emission scan with CT-based AC and IR, absorption of photons by the water tank has created inhomogeneities in the attenuation map that have influenced the appearance of artifacts adjacent to the metal. (c) Images obtained with 4-minute emission scan and 4-minute transmission scan and reconstructed with IR shows that repositioning of the steel rod within the water tank did not affect artifacts generated with conventional AC. (d) Images from the corresponding 68Ge-based transmission scan illustrates quality of attenuation map. (e) In contrast to c, these images show that repositioning of the steel rod within the water basin led to a change in artifact appearance when CT-based AC was used. Therefore, it can be expected that attenuation properties in a patient will be more important when CT-based AC is used. (f) Images from the corresponding transmission scan (40 mAs) with CT-based AC show that CT-based AC creates a high-quality attenuation map with less noise. However, adjacent to the metal is an artifact (arrows) with a spiral-like appearance. In addition, the absorption of photons in the water tank has created inhomogeneities in the attenuation map that are visible as white stripes. They influenced artifact appearance on the final reconstructed PET image in e. (g) Images obtained with a 4-minute emission scan and a 2-minute transmission scan and reconstructed with segmentation before (left) and after (right) rotation of the hip prosthesis show that rotation of the item within the same coronal plane has changed the position of the artifact (arrows) along the hip prosthesis because the geometric relationship between the radiation source and the steep edges of the prosthesis has changed. These images are set at a very dark gray-scale level to show that after rotation, the artifact is more visible at the shoulder of the prosthesis than at the femoral shaft.
|
|
View larger version (33K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3f. PET images of the steel rod phantom. (a) On images obtained with 4-minute emission scan and 180-minute transmission scan and reconstructed with segmentation and IR, a black circlelike artifact is visible around the steel rod in the large water basin. With longer-duration conventional transmission scan, image noise is reduced and artifact is somewhat more visible (see c). (b) On images from 4-minute emission scan with CT-based AC and IR, absorption of photons by the water tank has created inhomogeneities in the attenuation map that have influenced the appearance of artifacts adjacent to the metal. (c) Images obtained with 4-minute emission scan and 4-minute transmission scan and reconstructed with IR shows that repositioning of the steel rod within the water tank did not affect artifacts generated with conventional AC. (d) Images from the corresponding 68Ge-based transmission scan illustrates quality of attenuation map. (e) In contrast to c, these images show that repositioning of the steel rod within the water basin led to a change in artifact appearance when CT-based AC was used. Therefore, it can be expected that attenuation properties in a patient will be more important when CT-based AC is used. (f) Images from the corresponding transmission scan (40 mAs) with CT-based AC show that CT-based AC creates a high-quality attenuation map with less noise. However, adjacent to the metal is an artifact (arrows) with a spiral-like appearance. In addition, the absorption of photons in the water tank has created inhomogeneities in the attenuation map that are visible as white stripes. They influenced artifact appearance on the final reconstructed PET image in e. (g) Images obtained with a 4-minute emission scan and a 2-minute transmission scan and reconstructed with segmentation before (left) and after (right) rotation of the hip prosthesis show that rotation of the item within the same coronal plane has changed the position of the artifact (arrows) along the hip prosthesis because the geometric relationship between the radiation source and the steep edges of the prosthesis has changed. These images are set at a very dark gray-scale level to show that after rotation, the artifact is more visible at the shoulder of the prosthesis than at the femoral shaft.
|
|
View larger version (72K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3g. PET images of the steel rod phantom. (a) On images obtained with 4-minute emission scan and 180-minute transmission scan and reconstructed with segmentation and IR, a black circlelike artifact is visible around the steel rod in the large water basin. With longer-duration conventional transmission scan, image noise is reduced and artifact is somewhat more visible (see c). (b) On images from 4-minute emission scan with CT-based AC and IR, absorption of photons by the water tank has created inhomogeneities in the attenuation map that have influenced the appearance of artifacts adjacent to the metal. (c) Images obtained with 4-minute emission scan and 4-minute transmission scan and reconstructed with IR shows that repositioning of the steel rod within the water tank did not affect artifacts generated with conventional AC. (d) Images from the corresponding 68Ge-based transmission scan illustrates quality of attenuation map. (e) In contrast to c, these images show that repositioning of the steel rod within the water basin led to a change in artifact appearance when CT-based AC was used. Therefore, it can be expected that attenuation properties in a patient will be more important when CT-based AC is used. (f) Images from the corresponding transmission scan (40 mAs) with CT-based AC show that CT-based AC creates a high-quality attenuation map with less noise. However, adjacent to the metal is an artifact (arrows) with a spiral-like appearance. In addition, the absorption of photons in the water tank has created inhomogeneities in the attenuation map that are visible as white stripes. They influenced artifact appearance on the final reconstructed PET image in e. (g) Images obtained with a 4-minute emission scan and a 2-minute transmission scan and reconstructed with segmentation before (left) and after (right) rotation of the hip prosthesis show that rotation of the item within the same coronal plane has changed the position of the artifact (arrows) along the hip prosthesis because the geometric relationship between the radiation source and the steep edges of the prosthesis has changed. These images are set at a very dark gray-scale level to show that after rotation, the artifact is more visible at the shoulder of the prosthesis than at the femoral shaft.
|
|
4. FDG concentration had no influence on either the quality of the final reconstructed PET image or the generation of artifacts (no figure shown).
5. Duration of emission scanning had no influence on either the quality of the final reconstructed PET image or the generation of artifacts. An example of a reconstruction from a 180-minute transmission scan is shown in Figure 3. The time of transmission scanning had only a marginal influence on the appearance of artifacts, although use of a longer transmission time resulted in an attenuation map that was less noisy. In FBP images, this did not visibly influence the appearance of artifacts (no figure shown). In contrast, after IR, the artifacts became slightly more evident due to the reduced influence of noise on image quality (Fig 3).
6. Tube current (in milliamperes) had no influence on final image quality (no figure shown).
7. After the prosthesis was repositioned 3 mm to the right between emission and transmission scanning without changing the axis in the field of view, a strong increase in the visibility of the artifact along the hip prosthesis was observed (Fig 4). This artifact was slightly more visible when the CT-based attenuation map was used than when the conventional attenuation map was used (Fig 4).
View larger version (124K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4. Coronal PET images illustrate the effect of repositioning the prosthesis 3 mm to the right between emission and transmission scanning, which was performed to mimic patient movement. Misalignment between emission scanning and the attenuation map creates a black stripe that is evident when a 4-minute emission scan, a 10-minute transmission scan, and IR and segmentation are used. Images obtained, A, before and, B, after the change of position show the increase in visibility of the artifact due to repositioning of the prosthesis. The same phenomenon is evident when CT-based AC is used on images obtained before (C) and after (D) the change of position. With CT-based AC, this artifact is slightly more evident (as seen in D compared with B). The artifact mimics increased FDG uptake adjacent to the femoral stem in this experiment, but it could have a more focal appearance depending on what movement is performed. The activity profiles show that measured radioactivity concentration is two to three times that of the surrounding water. (The left and right activity profiles were obtained at the levels of the horizontal lines in B and D, respectively.)
|
|
|
Discussion
|
|---|
The results of the experiments performed in this study show that PET imaging of prosthetic material can generate artifacts. All notable artifacts manifest as artificial increases in activity. On non–attenuation-corrected emission scan images, artifacts are visible within the imaged confines of the metal if FBP is used for image reconstruction. This is due to use of the FBP algorithm, which tends to fill in lacking data. When IR is used, this kind of artifact is not visible. However, because this kind of artifact is located within the imaged confines of the metal, it can easily be identified and is not likely to be mistaken for increased FDG uptake in a diseased area adjacent to a metallic implant.
In contrast, AC results in the generation of artifacts adjacent to metal implants like hip prostheses. It can be assumed that such artifacts are visible on images obtained with all types of PET scanners. In this study, measurements were performed with the GE in-line PET/CT scanner Discovery LS and with the Siemens ECAT Exact HR+ scanner. In a report by Heiba et al (4), an AC-induced artifact was also described. These researchers scanned their patients with a Vertex MCD/AC scanner (ADAC, Milpitas, Calif) and performed three-dimensional data acquisition, with 32 projections (at 40 seconds per projection) in emission scanning and 96 projections (at 2 seconds per projection) in transmission scanning. They observed an artifact that arose between the femoral and tibial parts of the prosthesis. Our data suggest that artifacts can be seen at the borders between normal tissue and prosthetic material, not just between two adjacent metal parts.
The shape of a prosthesis is important, since artifacts are only generated when differences between high-density metal objects and surrounding tissues are present. Metal has a high density, and at the boundaries where the metal meets surrounding tissues, inconsistencies in data can arise. This is inherently a problem of spatial resolution and partial volume effects. Because of the limitations of finite sampling, a scanner records information at the interface of structures of low to high activity or density that is an average of the two adjacent activities or densities. This leads to overcorrection of low-density areas around a high-density object and becomes relatively more important when one is evaluating very small and/or dense objects. In this study, only hip prostheses and a large metal rod were scanned with conditions similar to those encountered in clinical studies. Therefore, the problem of partial volume mapping of very small and/or dense objects was not addressed with this study.
Generation of observed artifacts is strongly influenced by the method (ie, CT based or 68Ge based) of transmission scanning. The 68Ge sources used for AC have a length of 30 cm and "irradiate" the whole field of view in a homogeneous manner. In contrast, the helical CT scan used for transmission scanning with the Discovery LS unit produces a narrow beam of radiation, which allows creation of an attenuation map with a smaller voxel size. The CT scan creates an inhomogeneity in the attenuation map that looks like a spiral of bright and dark stripes adjacent to the round metal rod. The conventional transmission scan generates a noisy image, but the attenuation map produced by the CT scan is visually much better because the CT scan is performed with a 512 x 512 image matrix which, in a second step, is redimensioned to 128 x 128 to adapt CT scanning to PET emission scanning.
Additionally, even when a low-dose CT scan performed with only 10 mA is used, image statistics are better, rendering the attenuation map less noisy. However, increasing the milliamperes for the CT scan does not change the visual quality of the final image. On the other hand, CT scanning is more influenced by surrounding tissues than is a 68Ge-based attenuation map; this difference is explained by the different energies of the photons used in transmission scanning.
Therefore, it can be expected that use of CT-based AC has the potential to result in different appearances of artifacts adjacent to metallic implants depending on the tissues surrounding the prostheses. In addition, it is possible that artifacts look different on images obtained in obese and thin people, depending on the method of AC used in imaging. Furthermore, it is possible that there is a difference in artifact appearance between cemented and noncemented prostheses. The influence of surrounding tissues and cement on artifact appearance should be addressed in future studies.
Movement of the prosthesis between emission and transmission scans leads to a clear increase in visibility of the artifact: A 3-mm movement of the prosthesis to the left side, parallel to the z axis, creates a lack of attenuation in an area where emission data are available; this results in overcorrection and generation of a black stripe along the shaft. This phenomenon is almost equally well evident with CT-based AC and 68Ge-based AC and is also present when both methods of reconstruction are used. However, movement-induced artifact is somewhat better visible with CT-based AC than with 68Ge-based AC; this reflects the better quality of the CT-based attenuation map.
Furthermore, our data demonstrate that the algorithm used for image reconstruction has an important effect on the visibility of artifacts in the final image: If IR is based on precorrected sinograms, as was the case when the Discovery LS PET camera was used, inhomogeneities in the attenuation map adjacent to the metallic implant are amplified; in the final images, this becomes evident along the shaft of a steel prosthesis, for example. In a noisy image, artifacts tend to be less evident, although they are present.
The results of this study suggest that artifact appearance will not be a constant finding in clinical studies involving imaging of metallic hip prostheses, but when artifacts do occur, they are a product of interactions between the size and shape of the prosthesis, the absorption properties of surrounding tissues, patient movement, and the image reconstruction method used.
The difference in the visibility of artifacts between images obtained with each of the two scanners used in this study is explained by the different methods of image reconstruction. In a more recent study, Van Acker et al (5) showed, in a phantom experiment performed with an ECAT Exact HR+ PET scanner, that artifacts mimicking FDG uptake adjacent to a knee prosthesis can arise in attenuation-corrected images obtained with different methods of image reconstruction. The weighted IR algorithm programmed into this camera seems to result in a reduction of the generation of artifacts adjacent to metallic implants. However, the use of this algorithm is not expected to result in an improvement in image quality if patient movement leads to a mismatch between emission and transmission data.
In our clinical experience, we have observed artifacts that mimic the presence of increased FDG uptake in patients with hip and knee prostheses. In this study, only slight FDG uptake was mimicked by artifacts, as was shown in the activity profiles (ie, the measured activity values reached about twice the value of the background). Therefore, one could conclude that such artifacts should be easily distinguishable from metabolically active disease adjacent to a hip prosthesis.
However, because patient movement of only a few millimeters will increase the dark region adjacent to a metal object, it could well be possible that such artifacts could mimic pathologic FDG uptake in patients with clinical evidence of infection or inflammation. Therefore, it seems prudent to verify an increased FDG uptake adjacent to prosthetic material by also reviewing the non–attenuation-corrected emission scan images. Because they are due to AC, artifacts are most easily identified as such on uncorrected images. It is of paramount importance to interpret PET studies obtained in patients suspected of having inflammation or infection adjacent to metallic foreign bodies carefully to make full use of FDG PET in imaging foci of infection (5,10–13), because one of the most important clinical problems in infection imaging is the identification of infected prosthetic metallic implants.
The data of this study suggest that the development of optimized reconstruction software may help reduce the problem of artifacts in imaging metallic prostheses. Future studies must be performed to evaluate the influence of other imaging reconstruction factors such as segmentation, the number of iterative steps, or preprocessing of attenuation data on artifact reduction. In addition, the possible influence of metal-induced artifacts on the diagnostic accuracy of clinical studies should be evaluated.
In conclusion, the findings of this study indicate that the use of AC can result in the generation of artifacts of apparently increased FDG concentration around hip prostheses. Such artifacts are visible with all types of scanners because of the inherent problem of partial volume mapping at the borders of metal and adjacent tissues. The shape of the prosthesis, the absorption properties of the surrounding tissues, and the method of transmission scanning (ie, whether it is CT or 68Ge based) influence the appearance of such artifacts. Since patient movement worsens these artifacts, it is important to verify attenuation-corrected images against non–attenuation-corrected images to avoid false-positive results in patients suspected of having infection or inflammation adjacent to prosthetic material. It appears that the use of attenuation-weighted IR can reduce the visibility of such artifacts.
|
FOOTNOTES
|
|---|
Abbreviations: AC = attenuation correction,
FBP = filtered back projection,
FDG = fluorine 18 fluorodeoxyglucose,
IR = iterative reconstruction,
OSEM = ordered-subset expectation maximization
Author contributions: Guarantor of integrity of entire study, G.W.G.; study concepts and design, G.W.G., A.B., C.B.; literature research, G.W.G., S.I.Z.; experimental studies, G.W.G., T.B., C.B., S.I.Z., A.B.; data acquisition, G.W.G., T.B., C.B., S.I.Z.; data analysis/interpretation, all authors; manuscript preparation, G.W.G.; manuscript definition of intellectual content, G.W.G., C.B., S.I.Z., A.B., G.K.v.S.; manuscript editing, G.W.G., G.K.v.S.; manuscript revision/review, G.W.G., C.B., S.I.Z., A.B., G.K.v.S.; manuscript final version approval, G.W.G., G.K.v.S.
|
REFERENCES
|
|---|
- Stumpe KD, Dazzi H, Schaffner A, von Schulthess GK. Infection imaging using whole-body FDG-PET. Eur J Nucl Med 2000; 27:822-832.[CrossRef][Medline]
- Zhuang H, Duarte PS, Pourdehnad M, et al. The promising role of 18F-FDG PET in detecting infected lower limb prosthesis implants. J Nucl Med 2001; 42:44-48.[Abstract/Free Full Text]
- Seabold JE, Nepola JV. Imaging techniques for evaluation of postoperative orthopedic infections. Q J Nucl Med 1999; 43:21-28.[Medline]
- Heiba S, Luo J, Sadek S, et al. Attenuation-correction induced artifact in F-18 FDG PET imaging following total knee replacement. Clin Positron Imaging 2000; 3:237-239.[CrossRef][Medline]
- Van Acker F, Nuyts J, Maes A, et al. FDG-PET, 99mTc-HMPAO white blood cell SPET and bone scintigraphy in the evaluation of painful total knee arthroplasties. Eur J Nucl Med 2001; 28:1496-1504.[CrossRef][Medline]
- Kinahan P, Townsend D, Beyer T, Sashin D. Attenuation correction for a combined 3D PET/CT scanner. Med Phys 1998; 25:2046-2053.[CrossRef][Medline]
- Bettinardi V, Pagani E, Gilardi MC, et al. An automatic classification technique for attenuation correction in positron emission tomography. Eur J Nucl Med 1999; 26:447-458.[CrossRef][Medline]
- Comtat C, Kinahan PE, Defrise M, Michel C, Townsend DW. Fast reconstruction of 3D PET data with accurate statistical modeling. Trans Nucl Sci 1998; 45:1083-1089.[CrossRef]
- Mikolajczyk K, Szabatin M, Rudnicki P, Grodzki M, Burger C. A JAVA environment for medical image data analysis: initial application for brain PET quantitation. Med Inform (Lond) 1998; 23:207-214.
- Sugawara Y, Braun DK, Kison PV, Russo JE, Zasadny KR, Wahl RL. Rapid detection of human infections with fluorine-18 fluorodeoxyglucose and positron emission tomography: preliminary results. Eur J Nucl Med 1998; 25:1238-1243.[CrossRef][Medline]
- Kalicke T, Schmitz A, Risse JH, et al. Fluorine-18 fluorodeoxyglucose PET in infectious bone diseases: results of histologically confirmed cases. Eur J Nucl Med 2000; 27:524-528.[CrossRef][Medline]
- Robiller FC, Stumpe KD, Kossmann T, Weisshaupt D, Bruder E, von Schulthess GK. Chronic osteomyelitis of the femur: value of PET imaging. Eur Radiol 2000; 10:855-858.[CrossRef][Medline]
- Guhlmann A, Brecht-Krauss D, Suger G, et al. Chronic osteomyelitis: detection with FDG PET and correlation with histopathologic findings. Radiology 1998; 206:749-754.[Abstract/Free Full Text]
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]
|
|
|
|
|
|
|
D. S. Boss, R. V. Olmos, M. Sinaasappel, J. H. Beijnen, and J. H. M. Schellens
Application of PET/CT in the Development of Novel Anticancer Drugs
Oncologist,
January 1, 2008;
13(1):
25 - 38.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
W. V. Vogel, J. A. van Dalen, B. Wiering, H. Huisman, F. H.M. Corstens, T. J.M. Ruers, and W. J.G. Oyen
Evaluation of Image Registration in PET/CT of the Liver and Recommendations for Optimized Imaging
J. Nucl. Med.,
June 1, 2007;
48(6):
910 - 919.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
M. A. Blake, A. Singh, B. N. Setty, J. Slattery, M. Kalra, M. M. Maher, D. V. Sahani, A. J. Fischman, and P. R. Mueller
Pearls and Pitfalls in Interpretation of Abdominal and Pelvic PET-CT
RadioGraphics,
September 1, 2006;
26(5):
1335 - 1353.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
P. H. Kuo and D. W. Cheng
Artifactual Spinal Metastases Imaged by PET/CT: A Case Report
J. Nucl. Med. Technol.,
December 1, 2005;
33(4):
230 - 231.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
M. Schwaiger, S. Ziegler, and S. G. Nekolla
PET/CT: Challenge for Nuclear Cardiology
J. Nucl. Med.,
October 1, 2005;
46(10):
1664 - 1678.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
W. Sureshbabu and O. Mawlawi
PET/CT Imaging Artifacts
J. Nucl. Med. Technol.,
September 1, 2005;
33(3):
156 - 161.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
N. Subhas, P. V. Patel, H. K. Pannu, H. A. Jacene, E. K. Fishman, and R. L. Wahl
Imaging of Pelvic Malignancies with In-Line FDG PET-CT: Case Examples and Common Pitfalls of FDG PET
RadioGraphics,
July 1, 2005;
25(4):
1031 - 1043.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
C. Love, S. E. Marwin, M. B. Tomas, E. S. Krauss, G. G. Tronco, K. K. Bhargava, K. J. Nichols, and C. J. Palestro
Diagnosing Infection in the Failed Joint Replacement: A Comparison of Coincidence Detection 18F-FDG and 111In-Labeled Leukocyte/99mTc-Sulfur Colloid Marrow Imaging
J. Nucl. Med.,
November 1, 2004;
45(11):
1864 - 1871.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
K. D. M. Stumpe, H. P. Notzli, M. Zanetti, E. M. Kamel, T. F. Hany, G. W. Gorres, G. K. von Schulthess, and J. Hodler
FDG PET for Differentiation of Infection and Aseptic Loosening in Total Hip Replacements: Comparison with Conventional Radiography and Three-Phase Bone Scintigraphy
Radiology,
May 1, 2004;
231(2):
333 - 341.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
B. S. Halpern, M. Dahlbom, C. Waldherr, C. S. Yap, C. Schiepers, D. H. Silverman, O. Ratib, and J. Czernin
Cardiac Pacemakers and Central Venous Lines Can Induce Focal Artifacts on CT-Corrected PET Images
J. Nucl. Med.,
February 1, 2004;
45(2):
290 - 293.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
W. V. Vogel, W. J.G. Oyen, J. O. Barentsz, J. H.A.M. Kaanders, and F. H.M. Corstens
PET/CT: Panacea, Redundancy, or Something in Between?
J. Nucl. Med.,
January 1, 2004;
45(90010):
15S - 24.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
O. Ratib
PET/CT Image Navigation and Communication
J. Nucl. Med.,
January 1, 2004;
45(90010):
46S - 55.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
G. Goerres, D. Schmid, and G. Eyrich
Do hardware artefacts influence the performance of head and neck PET scans in patients with oral cavity squamous cell cancer?
Dentomaxillofac. Radiol.,
November 1, 2003;
32(6):
365 - 371.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
E. Dizendorf, T. F. Hany, A. Buck, G. K. von Schulthess, and C. Burger
Cause and Magnitude of the Error Induced by Oral CT Contrast Agent in CT-Based Attenuation Correction of PET Emission Studies
J. Nucl. Med.,
May 1, 2003;
44(5):
732 - 738.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
