(Radiology. 2001;218:592-597.)
© RSNA, 2001
US Artifacts: Effects on Out-of-Plane US Images Reconstructed from Three-dimensional Data Sets1
Janet E. Bailey, MD,
Ronald O. Bude, MD and
Theresa Tuthill, PhD
1 From the Department of Radiology, B1D 502, University of Michigan Medical Center, 1500 E Medical Center Dr, Ann Arbor, MI 48109-0030. Received December 3, 1999; revision requested January 21, 2000; revision received May 12; accepted June 1. Address correspondence to R.O.B. (e-mail: ronbude@umich.edu).
 |
ABSTRACT
|
|---|
Three-dimensional volumetric data sets of stacked ultrasonographic (US) scans were obtained in vitro and in vivo to investigate the effect of common US artifacts on cross-sectional images reconstructed out-of-plane to the plane of acquisition of these data sets. The appearance of the artifacts on the reconstructed images was different from that on the source images. Such artifacts have the potential to simulate pathologic abnormalities.
Index terms: Phantoms • Ultrasound (US), artifact, **.129872, **.129892 • Ultrasound (US), experimental studies, **.129872
, **.129892 • Ultrasound (US), physics, **.129872
, **.129892 • Ultrasound (US), three-dimensional, **.129872
, **.129892
|
INTRODUCTION
|
|---|
The use of three-dimensional (3D) volume data sets, obtained with closely spaced, sequentially scanned ultrasonographic (US) sections, to obtain reformatted images has been investigated for the past decade (1–18). Early efforts were hampered by technologic limitations, especially the huge computational load required for image reconstruction. Recent improvements in US and computer technology have now enabled the clinical use of this technique.
One potential use of this technology is the reconstruction of US images out-of-plane to the original plane of insonation (1,2,4,5). Possible advantages include (a) the ability to better visualize anatomic or pathologic characteristics in an optimal plane that cannot be insonated directly; (b) the ability to obtain the entire volume data set during one breath hold, with subsequent reconstruction of planar images, to better evaluate organs subject to respiratory motion (17); (c) the ability to perform the data acquisition portion of an examination very quickly (17); and (d) a reduction of operator dependence, which would result in improved standardization and repeatability of examinations (17).
Artifacts such as distal acoustic shadowing, increased through-transmission, and edge (refractile) artifact are frequently present on US scans. A potential pitfall of planar images reconstructed out of plane to the plane in which the data set was obtained is that the artifacts, but not the structure that caused them, may be cast into the reconstructed plane. This can cause these artifacts to have a different appearance on reconstructed images than on directly scanned images obtained in the same plane as the reconstructed images, and this difference may cause confusion in image interpretation. This is not often the case in conventional two-dimensional US, where both the artifact and its source are usually present on the same image.
We performed this study to investigate the effects of some of these conventional US artifacts on reformatted planar images, both in phantoms constructed solely for this purpose and in vivo.
|
Materials and Methods
|
|---|
Three-dimensional US volumetric data sets, with subsequent reformatted images obtained out of plane to the plane of acquisition, were obtained in two settings. By using a commercially available US unit coupled to a position-encoding device, we scanned two phantoms specifically designed for this experiment. Reformatted images were generated with off-line analysis by using image-processing software. Two patients were scanned with a commercially available US unit, and as part of their clinical studies, the intrinsic image-reformatting capability of the unit was used to generate reformatted images out of plane to the plane of acquisition. A detailed description of the experimental technique in both settings follows.
US Phantom Construction
Two phantoms were constructed. One contained a balloon filled with water to simulate a cyst; the other contained a balloon filled with water with a small stone inside the balloon to simulate a gallbladder containing a gallstone. Both balloons were suspended within an echogenic matrix, the echogenicity of which simulated the speckle of homogeneous organs such as the thyroid or testis, as has been described previously (19). Specifically, unit volumes of the echogenic matrix were prepared by dissolving 20 g of unflavored gelatin and 10 g of psyllium hydrophilic mucilloid fiber (Sugar-Free Metamucil; Procter & Gamble, Cincinnati, Ohio), which was the echogenic scatterer, in 250 mL of boiling water. A layer of this matrix was poured into a container and cooled at approximately 6°C in a refrigerator until the layer gelled. Once the surface of the matrix was firm, the water-filled balloon was placed on the surface, and the remainder of the matrix was poured into the container and allowed to gel.
Gray-Scale US
In vitro US was performed with a LOGIC 700 MR US unit (GE Medical Systems, Milwaukee, Wis). Images were obtained (by J.E.B. and R.O.B.) by using an M12L variable-frequency linear-array transducer operating at 13 MHz, with multiple focal zones at the region of interest.
In Vitro 3D Volume Data Set Acquisition
Three-dimensional volume data acquisitions were obtained by using the following technique, as has been described elsewhere (20). The transducer was fixed to a freehand scan system where the B-mode scan plane was perpendicular to the scanning motion. A framework allowed the transducer to slide along a track but prevented lateral, range, and rotational movement. A computer-controlled position encoder recorded the transducer location and regulated an external trigger so that B-mode scans were acquired at fixed intervals of 100 µm ± 10. Approximately 5 minutes was required per 3D volume acquisition to scan and obtain the data. Each B-mode scan was digitized as 379 x 380-pixel images, with each pixel corresponding to a 100 x 100-µm region. Once the data sets were obtained for both phantoms, off-line coronal images orthogonal to the plane of original image acquisition were obtained by using image-processing software (MATLAB; Mathworks, Natick, Mass). The desired reconstruction section thickness was obtained by averaging in the original axial dimension. Calibration experiments with a homogeneous speckle phantom have previously shown that the section thickness of the transducer, when operated with the selected parameters, is approximately 200 µm. Images were therefore reconstructed from the 3D data set at a section thickness of 200 µm, throughout the entire thickness of the phantoms, to enable accurate comparison with the directly scanned coronal images.
In Vivo US
All in vivo 3D data set acquisitions were obtained with a LOGIC 700 MR US unit (GE Medical Systems) by using its 3D volume acquisition capability. Three-dimensional volume acquisitions were obtained freehand by recording the images while translating the transducer over the region of interest. For the liver, this was accomplished during a single breath hold by using a 2–4-MHz variable-frequency probe. For the dialysis fistula, this was accomplished in approximately 10 seconds by using a probe operating at 8 MHz. Coronal images of the liver and forearm orthogonal to the plane of 3D volume acquisition were then reconstructed by using the software of the US unit.
|
Results
|
|---|
In Vitro Images
The images of the cyst phantom (Fig 1) showed that increased through-transmission and refractile artifacts distal to a cyst can be reconstructed into an image whose plane of reconstruction includes these artifacts. This produces an artifactual appearance on the reconstructed image that would not be present on a scan obtained with direct scanning in the same plane as the plane of reconstruction.
View larger version (125K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1a. US images from the cyst phantom. (a) Representative directly scanned US image obtained by scanning perpendicular to the top edge of the phantom (top of image). A region of increased through-transmission (t), or distal acoustic enhancement, is present internal to the simulated cyst. Refractile shadowing (r) is seen at the lateral edges of the area of increased through-transmission. The thick white line indicates the plane of the reconstructed image in b, and the thin white line denoted by the curved arrow indicates the plane of the reconstructed image in c. (b) Coronal US image of the simulated cyst reconstructed from the 3D data set in the plane indicated by the thick white line in a. (c) Coronal US image reconstructed in the plane indicated by the thin white line in a. The increased through-transmission (t) manifests as an oval area of increased echogenicity produced by the simulated cyst in a, and this oval area is surrounded by the hypoechoic refractile artifact (r). The two regions are completely artifactual and were not present in direct coronal scans obtained in the same plane, which showed only the homogeneous speckle pattern of the phantom substrate. Note that in c, the artifacts mimic the appearance of an echogenic lesion with a hypoechoic halo. This appearance in vivo would simulate an echogenic mass.
|
|
View larger version (179K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1b. US images from the cyst phantom. (a) Representative directly scanned US image obtained by scanning perpendicular to the top edge of the phantom (top of image). A region of increased through-transmission (t), or distal acoustic enhancement, is present internal to the simulated cyst. Refractile shadowing (r) is seen at the lateral edges of the area of increased through-transmission. The thick white line indicates the plane of the reconstructed image in b, and the thin white line denoted by the curved arrow indicates the plane of the reconstructed image in c. (b) Coronal US image of the simulated cyst reconstructed from the 3D data set in the plane indicated by the thick white line in a. (c) Coronal US image reconstructed in the plane indicated by the thin white line in a. The increased through-transmission (t) manifests as an oval area of increased echogenicity produced by the simulated cyst in a, and this oval area is surrounded by the hypoechoic refractile artifact (r). The two regions are completely artifactual and were not present in direct coronal scans obtained in the same plane, which showed only the homogeneous speckle pattern of the phantom substrate. Note that in c, the artifacts mimic the appearance of an echogenic lesion with a hypoechoic halo. This appearance in vivo would simulate an echogenic mass.
|
|
View larger version (175K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1c. US images from the cyst phantom. (a) Representative directly scanned US image obtained by scanning perpendicular to the top edge of the phantom (top of image). A region of increased through-transmission (t), or distal acoustic enhancement, is present internal to the simulated cyst. Refractile shadowing (r) is seen at the lateral edges of the area of increased through-transmission. The thick white line indicates the plane of the reconstructed image in b, and the thin white line denoted by the curved arrow indicates the plane of the reconstructed image in c. (b) Coronal US image of the simulated cyst reconstructed from the 3D data set in the plane indicated by the thick white line in a. (c) Coronal US image reconstructed in the plane indicated by the thin white line in a. The increased through-transmission (t) manifests as an oval area of increased echogenicity produced by the simulated cyst in a, and this oval area is surrounded by the hypoechoic refractile artifact (r). The two regions are completely artifactual and were not present in direct coronal scans obtained in the same plane, which showed only the homogeneous speckle pattern of the phantom substrate. Note that in c, the artifacts mimic the appearance of an echogenic lesion with a hypoechoic halo. This appearance in vivo would simulate an echogenic mass.
|
|
The images of the gallbladder phantom (Fig 2) illustrate several additional features of interest:
View larger version (153K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2a. US images of the gallbladder-with-stone phantom. (a) Representative directly scanned US image obtained by scanning perpendicular to the top edge of the phantom (top of image). The following artifacts are seen: stone shadowing (s), increased through-transmission (t), and refractile artifact (r). The thick white line indicates the plane of the reconstructed image in b, and the thin white line denoted with the curved arrow indicates the plane of the reconstructed image in d. (b) Coronal US image reconstructed in the plane of the thick white line in a. cf = simulated gallbladder fluid, r = refractile artifact, s = stone shadowing, St = the edge of the stone, t = area of through-transmission. (c) Directly scanned coronal US image obtained in the same plane as b (scan was obtained perpendicular to the right edge of the phantom in a by using the same machine settings for technical factors as those used for obtaining the 3D data set). The following differences are noted between this image and that in b: (a) On the direct coronal image, only the near edge of the stone is seen, as typically happens in vivo. On the coronal reconstruction (b), however, the entire stone periphery is seen. (b) In the direct coronal scan, the stone shadow obscures the far edge of the stone and the portions of the simulated gallbladder wall and phantom internal to the stone. In the coronal reconstruction, the shadow is reconstructed into the center of the stone as it shines down from above (the original direction of insonation, perpendicular to the plane of the reconstructed image) and does not obscure any portions of the simulated gallbladder wall or adjacent phantom. The stone shadow projected into the center of the stone gives the impression that the internal contents of the stone are hypoechoic, when, in fact, the hypoechoic appearance is entirely artifactual and is produced by the stone shadow alone. (c) In the coronal reconstruction, the increased through-transmission and the refractile artifact surround the simulated gallbladder fluid like the rings of Saturn because they are cast into the plane of reconstruction from above, just as with the shadowing from the stone. These artifacts project distal to the simulated gallbladder with direct scanning. r = refractile artifact, s = stone shadowing, St = edge of stone. (d) Coronal US image reconstructed in the plane of the thin white line denoted by the curved arrow in a. An echogenic target lesion is simulated in the coronal reconstructed image at this level. In vivo, this might be misinterpreted as a lesion of importance if the source of the artifacts was not appreciated. This appearance, however, is entirely due to artifacts cast into the reconstruction plane. A direct coronal scan through this area (not shown) only demonstrated the homogeneous speckle of the phantom substrate. r = refractile artifact, s = stone shadow, t = area of through-transmission.
|
|
View larger version (179K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2b. US images of the gallbladder-with-stone phantom. (a) Representative directly scanned US image obtained by scanning perpendicular to the top edge of the phantom (top of image). The following artifacts are seen: stone shadowing (s), increased through-transmission (t), and refractile artifact (r). The thick white line indicates the plane of the reconstructed image in b, and the thin white line denoted with the curved arrow indicates the plane of the reconstructed image in d. (b) Coronal US image reconstructed in the plane of the thick white line in a. cf = simulated gallbladder fluid, r = refractile artifact, s = stone shadowing, St = the edge of the stone, t = area of through-transmission. (c) Directly scanned coronal US image obtained in the same plane as b (scan was obtained perpendicular to the right edge of the phantom in a by using the same machine settings for technical factors as those used for obtaining the 3D data set). The following differences are noted between this image and that in b: (a) On the direct coronal image, only the near edge of the stone is seen, as typically happens in vivo. On the coronal reconstruction (b), however, the entire stone periphery is seen. (b) In the direct coronal scan, the stone shadow obscures the far edge of the stone and the portions of the simulated gallbladder wall and phantom internal to the stone. In the coronal reconstruction, the shadow is reconstructed into the center of the stone as it shines down from above (the original direction of insonation, perpendicular to the plane of the reconstructed image) and does not obscure any portions of the simulated gallbladder wall or adjacent phantom. The stone shadow projected into the center of the stone gives the impression that the internal contents of the stone are hypoechoic, when, in fact, the hypoechoic appearance is entirely artifactual and is produced by the stone shadow alone. (c) In the coronal reconstruction, the increased through-transmission and the refractile artifact surround the simulated gallbladder fluid like the rings of Saturn because they are cast into the plane of reconstruction from above, just as with the shadowing from the stone. These artifacts project distal to the simulated gallbladder with direct scanning. r = refractile artifact, s = stone shadowing, St = edge of stone. (d) Coronal US image reconstructed in the plane of the thin white line denoted by the curved arrow in a. An echogenic target lesion is simulated in the coronal reconstructed image at this level. In vivo, this might be misinterpreted as a lesion of importance if the source of the artifacts was not appreciated. This appearance, however, is entirely due to artifacts cast into the reconstruction plane. A direct coronal scan through this area (not shown) only demonstrated the homogeneous speckle of the phantom substrate. r = refractile artifact, s = stone shadow, t = area of through-transmission.
|
|
View larger version (157K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2c. US images of the gallbladder-with-stone phantom. (a) Representative directly scanned US image obtained by scanning perpendicular to the top edge of the phantom (top of image). The following artifacts are seen: stone shadowing (s), increased through-transmission (t), and refractile artifact (r). The thick white line indicates the plane of the reconstructed image in b, and the thin white line denoted with the curved arrow indicates the plane of the reconstructed image in d. (b) Coronal US image reconstructed in the plane of the thick white line in a. cf = simulated gallbladder fluid, r = refractile artifact, s = stone shadowing, St = the edge of the stone, t = area of through-transmission. (c) Directly scanned coronal US image obtained in the same plane as b (scan was obtained perpendicular to the right edge of the phantom in a by using the same machine settings for technical factors as those used for obtaining the 3D data set). The following differences are noted between this image and that in b: (a) On the direct coronal image, only the near edge of the stone is seen, as typically happens in vivo. On the coronal reconstruction (b), however, the entire stone periphery is seen. (b) In the direct coronal scan, the stone shadow obscures the far edge of the stone and the portions of the simulated gallbladder wall and phantom internal to the stone. In the coronal reconstruction, the shadow is reconstructed into the center of the stone as it shines down from above (the original direction of insonation, perpendicular to the plane of the reconstructed image) and does not obscure any portions of the simulated gallbladder wall or adjacent phantom. The stone shadow projected into the center of the stone gives the impression that the internal contents of the stone are hypoechoic, when, in fact, the hypoechoic appearance is entirely artifactual and is produced by the stone shadow alone. (c) In the coronal reconstruction, the increased through-transmission and the refractile artifact surround the simulated gallbladder fluid like the rings of Saturn because they are cast into the plane of reconstruction from above, just as with the shadowing from the stone. These artifacts project distal to the simulated gallbladder with direct scanning. r = refractile artifact, s = stone shadowing, St = edge of stone. (d) Coronal US image reconstructed in the plane of the thin white line denoted by the curved arrow in a. An echogenic target lesion is simulated in the coronal reconstructed image at this level. In vivo, this might be misinterpreted as a lesion of importance if the source of the artifacts was not appreciated. This appearance, however, is entirely due to artifacts cast into the reconstruction plane. A direct coronal scan through this area (not shown) only demonstrated the homogeneous speckle of the phantom substrate. r = refractile artifact, s = stone shadow, t = area of through-transmission.
|
|
View larger version (183K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2d. US images of the gallbladder-with-stone phantom. (a) Representative directly scanned US image obtained by scanning perpendicular to the top edge of the phantom (top of image). The following artifacts are seen: stone shadowing (s), increased through-transmission (t), and refractile artifact (r). The thick white line indicates the plane of the reconstructed image in b, and the thin white line denoted with the curved arrow indicates the plane of the reconstructed image in d. (b) Coronal US image reconstructed in the plane of the thick white line in a. cf = simulated gallbladder fluid, r = refractile artifact, s = stone shadowing, St = the edge of the stone, t = area of through-transmission. (c) Directly scanned coronal US image obtained in the same plane as b (scan was obtained perpendicular to the right edge of the phantom in a by using the same machine settings for technical factors as those used for obtaining the 3D data set). The following differences are noted between this image and that in b: (a) On the direct coronal image, only the near edge of the stone is seen, as typically happens in vivo. On the coronal reconstruction (b), however, the entire stone periphery is seen. (b) In the direct coronal scan, the stone shadow obscures the far edge of the stone and the portions of the simulated gallbladder wall and phantom internal to the stone. In the coronal reconstruction, the shadow is reconstructed into the center of the stone as it shines down from above (the original direction of insonation, perpendicular to the plane of the reconstructed image) and does not obscure any portions of the simulated gallbladder wall or adjacent phantom. The stone shadow projected into the center of the stone gives the impression that the internal contents of the stone are hypoechoic, when, in fact, the hypoechoic appearance is entirely artifactual and is produced by the stone shadow alone. (c) In the coronal reconstruction, the increased through-transmission and the refractile artifact surround the simulated gallbladder fluid like the rings of Saturn because they are cast into the plane of reconstruction from above, just as with the shadowing from the stone. These artifacts project distal to the simulated gallbladder with direct scanning. r = refractile artifact, s = stone shadowing, St = edge of stone. (d) Coronal US image reconstructed in the plane of the thin white line denoted by the curved arrow in a. An echogenic target lesion is simulated in the coronal reconstructed image at this level. In vivo, this might be misinterpreted as a lesion of importance if the source of the artifacts was not appreciated. This appearance, however, is entirely due to artifacts cast into the reconstruction plane. A direct coronal scan through this area (not shown) only demonstrated the homogeneous speckle of the phantom substrate. r = refractile artifact, s = stone shadow, t = area of through-transmission.
|
|
1. When the reconstruction plane includes the stone (Fig 2b), the appearance of the stone is markedly different on the reconstructed image than it is on the directly scanned image (Fig 2c). On the directly scanned image, only the leading edge of the stone is seen. On the coronal reconstruction, the entire periphery of the stone, as imaged orthogonally to the plane of reconstruction, is seen. Also, on the directly scanned image, the shadow of the stone is projected distal to the stone. On the coronal reconstruction, however, the shadowing produced by the leading edge of the stone is reconstructed into the center of the stone because the shadow "shines down from above" in the 3D data set and projects into the reconstructed image. This creates the false impression that the stone is hypoechoic internally.
2. As in the cyst phantom (Fig 1), the increased through-transmission and refractile artifact produced by the simulated gallbladder shine down from above and manifest as hyperechoic and hypoechoic circular bands, respectively, that surround the simulated gallbladder (Fig 2b). These bands surround the fluid in the simulated gallbladder because the level of reconstruction was near the bottom of the simulated gallbladder, where its cross-sectional diameter was smaller than the maximum value. This allowed the refractile shadowing arising from the edge of the portion of the simulated gallbladder with the greatest diameter, as well as the increased through-transmission, to shine down from above and project outside the simulated gallbladder. This appearance was not seen on a direct coronal scan (Fig 2c).
3. On an image reconstructed in a plane distal to the simulated gallbladder (Fig 2d), not only are the area of increased through-transmission and the refractile edge artifact projected into the image (Fig 2d), as in Figure 1c, but the stone shadow is also projected into the image and manifests as a hypoechoic region within the hyperechoic area of increased through-transmission (Fig 2d). This complex appearance simulates that of an echogenic target lesion.
In Vivo Images
The in vivo images show that the principles demonstrated in vitro apply in vivo as well. The reconstructed image of the soft tissues internal to a dialysis fistula (Fig 3c) shows the artifactual appearance of the refractile artifact cast into the deep soft tissues. The reconstructed image of liver parenchyma deep to a benign cyst (Fig 4c) shows the artifactual appearance produced by distal acoustic enhancement, with the artifact simulating an echogenic lesion within the liver.
View larger version (162K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a. In vivo US images of a forearm dialysis fistula. (a) Representative directly scanned US image from the 3D data set shows a transverse cross-section of the fistula (f), with scanning performed perpendicular to the top of the image. A refractile artifact (r) is seen to arise from the edges of the fistula. The white line at top indicates the plane of the reconstructed image in b, and the white line denoted with the curved arrow indicates the plane of the reconstructed image in c. (b) Coronal US image of the fistula (arrowheads), reconstructed as if scanned perpendicular to the right side of the image in a at the level of the white line, demonstrates the curved course of the fistula through the forearm. (c) Coronal US image reconstructed at the level of the white line denoted with the curved arrow in a, but deeper into the soft tissues than in b, so that the reconstructed image includes only the refractile artifacts (r) (the two concentric hypoechoic rings) and does not contain any portion of the fistula. These concentric hypoechoic rings would not be present on a directly scanned coronal image (not obtained in this case because of the difficulties imposed on direct scanning by the geometry of the forearm).
|
|
View larger version (108K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b. In vivo US images of a forearm dialysis fistula. (a) Representative directly scanned US image from the 3D data set shows a transverse cross-section of the fistula (f), with scanning performed perpendicular to the top of the image. A refractile artifact (r) is seen to arise from the edges of the fistula. The white line at top indicates the plane of the reconstructed image in b, and the white line denoted with the curved arrow indicates the plane of the reconstructed image in c. (b) Coronal US image of the fistula (arrowheads), reconstructed as if scanned perpendicular to the right side of the image in a at the level of the white line, demonstrates the curved course of the fistula through the forearm. (c) Coronal US image reconstructed at the level of the white line denoted with the curved arrow in a, but deeper into the soft tissues than in b, so that the reconstructed image includes only the refractile artifacts (r) (the two concentric hypoechoic rings) and does not contain any portion of the fistula. These concentric hypoechoic rings would not be present on a directly scanned coronal image (not obtained in this case because of the difficulties imposed on direct scanning by the geometry of the forearm).
|
|
View larger version (120K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3c. In vivo US images of a forearm dialysis fistula. (a) Representative directly scanned US image from the 3D data set shows a transverse cross-section of the fistula (f), with scanning performed perpendicular to the top of the image. A refractile artifact (r) is seen to arise from the edges of the fistula. The white line at top indicates the plane of the reconstructed image in b, and the white line denoted with the curved arrow indicates the plane of the reconstructed image in c. (b) Coronal US image of the fistula (arrowheads), reconstructed as if scanned perpendicular to the right side of the image in a at the level of the white line, demonstrates the curved course of the fistula through the forearm. (c) Coronal US image reconstructed at the level of the white line denoted with the curved arrow in a, but deeper into the soft tissues than in b, so that the reconstructed image includes only the refractile artifacts (r) (the two concentric hypoechoic rings) and does not contain any portion of the fistula. These concentric hypoechoic rings would not be present on a directly scanned coronal image (not obtained in this case because of the difficulties imposed on direct scanning by the geometry of the forearm).
|
|
View larger version (138K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4a. In vivo US images of a liver cyst. (a) Directly scanned US image of a liver cyst (arrow) shows a well-defined band of distal acoustic enhancement (arrowheads) deep to the cyst. (b) US image of the midportion of the liver cyst (arrows), reconstructed from the 3D data set in the coronal plane, orthogonal to the plane in which a was acquired. (c) US image of liver parenchyma deep to the cyst, reconstructed from the 3D data set in a coronal plane parallel to that in b. Note the rounded area of increased echogenicity (arrows) compared to adjacent liver parenchyma that is due to inclusion of distal acoustic enhancement artifact in the reconstructed image. This artifact has the potential to simulate an echogenic liver mass such as a hemangioma or echogenic metastasis.
|
|
View larger version (149K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4b. In vivo US images of a liver cyst. (a) Directly scanned US image of a liver cyst (arrow) shows a well-defined band of distal acoustic enhancement (arrowheads) deep to the cyst. (b) US image of the midportion of the liver cyst (arrows), reconstructed from the 3D data set in the coronal plane, orthogonal to the plane in which a was acquired. (c) US image of liver parenchyma deep to the cyst, reconstructed from the 3D data set in a coronal plane parallel to that in b. Note the rounded area of increased echogenicity (arrows) compared to adjacent liver parenchyma that is due to inclusion of distal acoustic enhancement artifact in the reconstructed image. This artifact has the potential to simulate an echogenic liver mass such as a hemangioma or echogenic metastasis.
|
|
View larger version (147K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4c. In vivo US images of a liver cyst. (a) Directly scanned US image of a liver cyst (arrow) shows a well-defined band of distal acoustic enhancement (arrowheads) deep to the cyst. (b) US image of the midportion of the liver cyst (arrows), reconstructed from the 3D data set in the coronal plane, orthogonal to the plane in which a was acquired. (c) US image of liver parenchyma deep to the cyst, reconstructed from the 3D data set in a coronal plane parallel to that in b. Note the rounded area of increased echogenicity (arrows) compared to adjacent liver parenchyma that is due to inclusion of distal acoustic enhancement artifact in the reconstructed image. This artifact has the potential to simulate an echogenic liver mass such as a hemangioma or echogenic metastasis.
|
|
|
Discussion
|
|---|
Reconstruction of cross-sectional images from a 3D data set offers several potential advantages, including visualization of structures in planes that cannot be directly insonated (1,2,4,5), faster data acquisition in one breath hold with subsequent image reconstruction, and a reduction in operator dependence of the examination (17).
In conventional US, the artifact and its cause are often present on the same image, which allows the artifact and its source to be visually linked (Figs 1a, 2a, 2c, 3a, 4a). Such artifacts only rarely cause difficulty in image interpretation and, in fact, can be helpful for determining the true nature of a lesion. Examples of useful artifacts are acoustic enhancement distal to a benign cyst or shadowing distal to a calculus.
On reformatted cross-sectional images, however, the plane of reconstruction may include an artifact but not its cause. The results of our study show that the appearance of conventional US artifacts such as distal acoustic shadowing, refractile shadowing, and distal acoustic enhancement (increased through-transmission) on images reconstructed out of plane to the original plane of data acquisition can be different from that on directly scanned images (Figs 1c, 2b, 2d, 3c, 4c). This is because the relationship between the artifacts and the structures that generated them differs on reconstructed images and directly scanned images. Therefore, if only a single reconstructed image is reviewed, the source of the artifact may not be appreciated, and the artifact may mimic an abnormal finding. In our study, we demonstrated that such abnormal findings include, but may not be limited to, echogenic "pseudomasses" (Figs 1c, 4c) or target lesions (Fig 2d). Thus, it may be necessary in some instances to either (a) display reference scans from the original data set alongside the reconstructed images to enable linkage of a reconstructed artifact and its cause or (b) review or reconstruct the entire 3D image set to avoid confusing these artifacts with pathologic lesions.
If we had investigated only artifact-induced pseudolesions on reconstructed images of phantoms, it is possible that the phantoms would be dissimilar enough from living tissue that such pseudolesions might not be conspicuous in vivo. Our in vivo results (Figs 3, 4) suggest otherwise, and they indicate that US artifacts can manifest with a conspicuous appearance in reformatted in vivo images as well.
Other types of artifacts, such as grating or side lobe artifacts, ring down artifact, and mirror image artifact, were not considered in this study. The principle demonstrated in this study, however, applies to any artifact that projects away from its source and, therefore, should apply to these artifacts as well.
In conclusion, because of the presence of commonly encountered US artifacts, the appearance of cross-sectional US images reconstructed from a volumetric data set in a different plane than the plane of acquisition may differ from that of images obtained with direct scanning in the same plane as the reconstruction plane. These artifacts may mimic pathologic lesions on reconstructed images, which may lead to interpretation errors.
|
FOOTNOTES
|
|---|
2**. indicates multiple body systems. 
Abbreviation: 3D = three-dimensional
Author contributions: Guarantors of integrity of entire study, J.E.B., R.O.B.; study concepts and study design, R.O.B.; definition of intellectual content, R.O.B.; literature research, R.O.B.; clinical studies, J.E.B.; experimental studies, J.E.B., R.O.B., T.T.; data acquisition, J.E.B., R.O.B., T.T.; data analysis, J.E.B., R.O.B., T.T.; manuscript preparation, J.E.B., R.O.B.; manuscript editing, J.E.B., R.O.B., T.T.; manuscript review, J.E.B., R.O.B., T.T.; manuscript final version approval, R.O.B.
|
REFERENCES
|
|---|
-
Halliwell M, Key H, Jenkins D, Jackson PC, Wells PNT. New scans from old: digital reformatting of ultrasound images. Br J Radiol 1989; 62:824-829.[Abstract]
-
Pretorius DH, Nelson TR, Jaffe JS. Three-dimensional sonographic analysis based on color flow Doppler and gray scale image data: a preliminary report. J Ultrasound Med 1992; 11:225-232.[Abstract]
-
Pretorius DH, Nelson TR. Prenatal visualization of cranial sutures and fontanelles with three-dimensional ultrasonography. J Ultrasound Med 1994; 13:871-876.[Abstract]
-
Hamper UM, Trapanotto V, Sheth S, Dejong MR, Caskey CI. Three-dimensional US: preliminary clinical experience. Radiology 1994; 191:397-401.[Abstract/Free Full Text]
-
Steiner H, Staudach A, Spitzer D, Schaffer H. Three-dimensional ultrasound in obstetrics and gynaecology: technique, possibilities and limitations. Hum Reprod 1994; 9:1773-1778.[Abstract/Free Full Text]
-
Pretorius DH, Nelson TR. Fetal face visualization using three-dimensional ultrasonography. J Ultrasound Med 1995; 14:349-356.[Abstract]
-
Downey DB, Fenster A. Vascular imaging with a three-dimensional power Doppler system. AJR Am J Roentgenol 1995; 165:665-668.[Free Full Text]
-
Pretorius DH, House M, Nelson TR, Hollenbach KA. Evaluation of normal and abnormal lips in fetuses: comparison between three- and two-dimensional sonography. AJR Am J Roentgenol 1995; 165:1233-1237.[Abstract/Free Full Text]
-
Bonilla-Musoles F, Raga F, Osborne NG, Blanes J. Use of three-dimensional ultrasonography for the study of normal and pathologic morphology of the human embryo and fetus: preliminary report. J Ultrasound Med 1995; 14:757-765.[Abstract]
-
Riccabona M, Nelson TR, Pretorius DH, Davidson TE. Distance and volume measurement using three-dimensional ultrasonography. J Ultrasound Med 1995; 14:881-886.[Abstract]
-
Elliot TL, Downey DB, Tong S, McLean CA, Fenster A. Accuracy of prostate volume measurements in vitro using three-dimensional ultrasound. Acad Radiol 1996; 3:401-406.[Medline]
-
Capineri L, Masotti L, Rocchi S, Andreuccetti F, Cerofolini M, Tondini A. Nearly real-time visualization of arbitrary two-dimensional sections from three-dimensional acquisition. Ultrasound Med Biol 1996; 22:319-328.[Medline]
-
Riccabona M, Johnson D, Pretorius DH, Nelson TR. Three-dimensional ultrasound: display modalities in the fetal spine and thorax. Eur J Radiol 1996; 22:141-145.[Medline]
-
Chan L, Lin WM, Uerpairojkit B, Hartman D, Reece EA, Helm W. Evaluation of adnexal masses using three-dimensional ultrasonographic technology: preliminary report. J Ultrasound Med 1997; 16:349-354.[Abstract]
-
Sivan E, Chan L, Uerpairojkit B, Chu GP, Reece EA. Growth of the fetal forehead and normative dimensions developed by three-dimensional ultrasonographic technology. J Ultrasound Med 1997; 16:401-405.[Abstract]
-
Wu MH, Hsu CC, Huang KE. Detection of congenital mullerian duct anomalies using three-dimensional ultrasound. J Clin Ultrasound 1997; 25:487-492.[Medline]
-
Rankin RN, Fenster A, Downey DB, Munk PL, Levin MF, Vellet AD. Three-dimensional sonographic reconstruction: techniques and diagnostic applications. AJR Am J Roentgenol 1993; 161:695-702.[Abstract/Free Full Text]
-
Tuthill TA, Krucker JF, Fowlkes JB, Carson PL. Automated three-dimensional US frame positioning computed from elevational speckle decorrelation. Radiology 1998; 209:575-582.[Abstract/Free Full Text]
-
Bude RO, Adler RS. An easily made, low-cost, tissue-like ultrasound phantom material. J Clin Ultrasound 1995; 23:271-273.[Medline]
-
Fenn RC, Fowlkes JB, Moskalik A, et al. A hand controlled, 3-D ultrasound guide and measurement system. In: Lees S, Ferrari LA, eds. Acoustical imaging. New York, NY: Plenum, 1997; 237-242.
This article has been cited by other articles:
|
|
|
|
 
B. R. Benacerraf
Three-dimensional Fetal Sonography: Use and Misuse
J. Ultrasound Med.,
October 1, 2002;
21(10):
1063 - 1067.
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
