| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Medical Physics |
1 From the Department of Radiology (D.F., G.D.R., P.R.H., C.F.B.) and Radiological Science Library (D.S.P., S.Y.Y., S.N.), Stanford University School of Medicine, 300 Pasteur Dr, S-072 Stanford, CA 94305-5105. From the 1999 RSNA scientific assembly. Received August 2, 1999; revision requested October 5; revision received October 28; accepted November 10. D.F. supported by the Austrian Science Fund, Vienna. G.D.R. supported by National Institutes of Health grant R01 HL 58915. D.S.P. supported by the National Library of Medicine, Bethesda, Md. P.R.H. supported by the Swiss National Science Foundation and the Holderbank Foundation, Zurich, Switzerland. S.N. supported by National Institutes of Health grant R01 CA 72023. Manuscript preparation supported in part by the Ludwig Boltzmann Institute for Radiological Sciences, Vienna, Austria. Address correspondence to G.D.R. (e-mail: grubin@stanford.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: A cylindric (12.7 x 305.0-mm) acrylic rod inclined 45° relative to the z axis was scanned at the isocenter and 100 mm from the isocenter with single detector-row (single-channel) helical CT (beam width, 1–10 mm; pitch, 1.0, 2.0, or 3.0) and multiple detector-row (four-channel) helical CT (detector width, 1.25, 2.5, 3.75, and 5 mm; pitch, 0.75 or 1.5). The SD of radius measurements along the rod (SDr) was used to quantify artifacts in all 72 data sets and to analyze their frequency patterns. Volume-rendered images of the data sets were ranked by six independent and blinded readers; findings were correlated with acquisition parameters and SDr measurements.
RESULTS: SDr was smaller in four- than in single-channel helical CT for any given table increment (TI). In single-channel helical CT, SDr increased linearly with beam width and geometrically with pitch. In four-channel helical CT, SDr measurements were directly proportional to the TI, regardless of the detector width and pitch combination used. Off-center object position on average increased SDr by a factor of 1.6 for single-channel helical CT and by a factor of 2.0 for four-channel helical CT. Subjective rankings of image quality correlated excellently with SDr (Spearman r = 0.94, P < .001).
CONCLUSION: Artifacts are quantitatively and subjectively smaller with four- compared with single-channel helical CT for any given TI.
Index terms: Computed tomography (CT), artifact, **.12113 • Computed tomography (CT), helical technology • Computed tomography (CT), three-dimensional • Test objects
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The technical aspects of artifacts are complex, but their analysis is important to assess their effect on image quality. This is particularly true for new multiple detector-row CT systems, which are a recent innovation whose tradeoffs have not yet been characterized. To optimize image quality, artifacts have to be taken into account when acquisition parameters are selected. Thus, knowledge of their magnitude and relation to the acquisition parameters is of paramount importance to the radiologist and the technician. In this experimental study, we aimed to characterize and compare the effects of acquisition parameters and object position on the magnitude, frequency, and subjective appearance of stair-step artifacts between single and multiple detector-row helical CT.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Single-Channel Helical CT
Standard single detector-row (single-channel) helical CT (see Glossary at end of article) was performed with a 360° gantry rotation time of 1 second. First, the center of the rod was placed in the isocenter of the gantry by using the scan localizing light at a 45° angle relative to the longitudinal or z axis of the scanner (in the coronal plane). The rod was scanned with beam widths (collimation) of 1, 3, 5, 7, and 10 mm and pitch (table translation per 360° gantry rotation per beam width) of 1.0, 2.0, and 3.0 for each of the beam widths. Tube voltage was 120 kVp in all series. Tube current settings are given in Table 1. Field of view of 256 mm, matrix of 512 x 512, and 180° linear interpolation algorithm were used to reconstruct images at intervals corresponding to approximately half the beam width of each series (Table 1). The table travel along the z axis was 18–21 cm for all series, except for the 1-mm collimation 1.0 pitch series in which tube loading did not allow scanning of the entire rod within a single acquisition. Instead, the central portion of the rod was scanned for a 50-mm range along the z axis. Subsequently, the table of the scanner was elevated to position the rod 100 mm off the isocenter of the scanner (in the anterior or y direction), and the sequence of acquisitions was repeated with identical parameter settings.
|
|
A two-point linear interpolation algorithm, which corresponds to 180° linear interpolation as used in single-channel helical CT (11), was used to reconstruct the primary series. In primary series, the nominal section thickness equaled the detector width. In addition, secondary series were obtained with use of a new variable-thickness interpolation algorithm. The z-filtering algorithm allowed interpolation of more than two points in the z direction for each projection angle, forming a thicker, but supposedly less noisy, composite section (11). This approach was used to generate secondary series with a nominal section thickness that was 1.33, 1.5, or 2 times greater than the detector width. For all four-channel helical CT series, the reconstruction interval was set at half the nominal section thickness of the reconstructed images (Table 2). The reconstruction field of view was 256 mm and matrix was 512 x 512, resulting in transverse pixel dimensions of 0.5 x 0.5 mm. The scanning length along the z axis was 18–21 cm for all four-channel helical CT acquisitions, which comprised 25–30 cm of the inclined rod.
Quantitative Image Analysis
All 72 data sets were networked to a workstation (O2; Silicon Graphics, Mountain View, Calif). To measure the magnitude of stair-step artifacts, we quantified the distortion of the surface profile of the rod in the coronal plane in each data set as illustrated schematically in Figure 1. First, a straight digital centerline path was computed through the rod (12). The radial distance from the centerline path to the surface of the rod was then automatically determined with use of a threshold of 0 HU on trilinearly interpolated image data to define the rod surface. In each data set, the radial distance was computed at 360 equidistant intervals along the centerline path over a 180-mm range in the longitudinal (z) direction. Thus, the sampling intervals between radial distance measurements were approximately 0.71 mm (0.5mm x 
2) along the path, to match exactly 0.5 mm relative to the table translation (z) direction. For the 1-mm collimation, 1.0 pitch single-channel helical CT series, the sampling rate was three times greater to collect 300 radial distance samples in the shorter (50 mm in z) rod segments. The radial distance measurements of each series were plotted against their longitudinal (z) positions to derive what we refer to as longitudinal profiles of the rod radius.
|
Frequency analysis.—We used a fast Fourier transform of the longitudinal profiles to identify and quantify simultaneously occurring and overlapping artifact frequencies in each data set. For each data set, we subtracted 256 radius measurements from the respective mean radius and computed a 512-point fast Fourier transform, resulting in 256 positive frequencies. As the sampling interval relative to z was 0.5 mm, the corresponding wavelengths in the table translation direction ranged from (512 x 0.5 mm)/1 to (512 x 0.5 mm)/256, which equals from 256 to 1 mm along the z axis. For the 1-mm collimation, 1.0 pitch single-channel helical CT series, the wavelengths were from 256/3 to 1/3 mm in the z direction. The quantitative measures of artifacts (ie, SDr and the spectral magnitudes of contributing frequency components) were correlated with the beam width (single-channel helical CT), detector width (four-channel helical CT), nominal section thickness (four-channel helical CT), pitch, and TI settings for both isocenter and off-center series with use of linear regression analysis. Pearson correlation coefficients were calculated, and a P value of .05 was used as the threshold for significant differences.
Subjective Image Analysis
To determine if the measures of artifact magnitude corresponded to visual perception of surface irregularities, all 72 data sets were 3D rendered (VOXELVIEW 2.5.4; Vital Images, Minneapolis, Minn) with parameters commonly used for 3D display of CT angiographic data. The same parameters were used for all rendered images. Frontal views were recorded from each data set. Each image was digitally rotated into a horizontal orientation and edited to remove all background information, such as the walls of the container holding the phantom (PHOTOSHOP 5.0; Adobe Systems, San Jose, Calif). Rendered images of the 1-mm collimation, 1.0 pitch single-channel helical CT series were electronically multiplied in the direction of the rod axis to give the same length of the phantom as in all the other series.
The images were coded with random numbers and then printed in color on 45 x 110-mm cards with use of a photographic-quality dye-sublimation printer (NP-1660; Codonics, Middleburg Heights, Ohio). All 72 cards were given to each of six independent readers (four radiologists [D.F., G.D.R., P.R.H., C.F.B.] and two nonmedical imaging scientists [D.S.P., S.N.]) in random order together with the original plastic rod. The readers, blinded to the scanner model and the acquisition technique, were asked to rank the cards according to how well the images on the cards represented the rod. The median rank and the range of the ranks assigned by the six readers were derived for each CT series. The medians of the subjective ranks were then correlated with the quantitative measurements (Spearman rank correlation). A P value of .05 was considered the threshold for significant differences.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
Four-Channel Helical CT
Detector width (primary four-channel helical CT series).—Analogous to beam width in single-channel helical CT, the relationships between the artifact magnitude and the detector widths of four-channel helical CT were linear. Figure 5 demonstrates that y intercepts and slopes (SDr divided by detector width [in millimeters]) increased with both pitch and off-center object position: 0.75 pitch, center y = 0.0 + 0.02x, off-center y = 0.03 + 0.04x; 1.5 pitch, center y = 0.01 + 0.04x, off-center y = 0.06 + 0.09x; P < .01 for all correlations.
|
|
TI.—For a given object position, artifact magnitude (SDr) increased linearly with TI (Fig 7). This relationship was independent of pitch, as shown by the similar slopes of 0.75 and 1.5 pitch series when obtained either at the isocenter or off center, respectively (Table 5). Furthermore, as previously noted, secondary series had only slightly greater SDr compared with their primary series counterparts. Thus for a given object position, the TI is the most useful predictor of artifact magnitude in four-channel helical CT, as it is independent from pitch, detector width, or z-filter width.
|
|
Artifact Magnitude with Single- versus Four-Channel Helical CT
From a practical standpoint, TI is the most suitable scale with which to compare the artifact magnitude between single- and four-channel helical CT. As 3.0 pitch is not used clinically for single-channel helical CT, these series were not considered for this analysis. To facilitate the comparison, the remaining 1.0 and 2.0 pitch single-channel helical CT series were pooled and regrouped into isocenter and off-center series. With 0.75 and 1.5 pitch, primary series in four-channel helical CT were pooled analogously into isocenter and off-center groups. When artifact magnitude was plotted against TI for single- and four-channel helical CT together (Fig 8), SDr was invariably smaller with the latter than with the former for any given TI and object position. The slopes of the regression lines were steeper for both isocenter and off-center single- compared with four-channel helical CT (Table 6). Figure 8, however, also illustrates the effect of off-center object position on the artifact magnitude, which is more pronounced in four- than in single-channel helical CT.
|
|
|
|
|
The frequency patterns observed in the 0.75 pitch series were more complex (Fig 11, part B). The lower frequency components were once again determined by the frequency of one artifact per TI in all series. The higher frequency components, however, occurred at variable periods. In the primary 0.75 pitch series, the highest frequency components appeared at the frequency of six artifacts per TI (except in the 3.75-mm TI acquisition, as a frequency of 0.625 mm-1 was not detectable with our analysis). In addition, we noted peaks of similar amplitude at the frequency of three artifacts per TI in all primary series (including the 3.75-mm TI acquisition). At the isocenter, both of these high-frequency components prevailed over the low-frequency component. Off center, however, the components of frequency of six artifacts per TI decreased and of three artifacts per TI increased slightly, but they were by far outweighed by the increase in the low-frequency component of one artifact per TI. In the secondary 0.75 pitch four-channel helical CT series, we never observed a frequency component of six artifacts per TI. This is probably due to the wider reconstruction interval (which was chosen according to the nominal section thickness rather than the detector width), which filtered out any frequency components higher than the section-reconstruction frequencies (similar to a low-pass filter). In all off-center secondary series, however, the same slight increase in the frequency of three artifacts per TI and the substantial increase in the frequency of one artifact per TI were observed.
For each four-channel helical CT series, the magnitudes of two characteristic frequency components were recorded for further analysis. The frequency of one artifact per TI was used as the low-frequency contributor in all data sets. For the high-frequency contributor, the frequency components of either six artifacts per TI (in the 1.5 pitch series) or three artifacts per TI (in 0.75 pitch series) were used.
High- versus Low-Frequency Artifact Contributors in Single- and Four-Channel Helical CT
Figure 12 illustrates the magnitude of artifacts when divided into the amplitudes of two characteristic contributing frequencies for both CT systems. Figure 12, part a shows that the amplitudes of the high-frequency components increased with TI but were clearly independent of the object position relative to the isocenter, as indicated by the parallel course of the regression lines for corresponding isocenter and off-center groups. The lower frequency components (Fig 12, part b) also increased with TI, but this increase was substantially different between isocenter and off-center series for each scanner. Figure 12, parts a and b, shows that the most prominent artifacts in four-channel helical CT derived from the low-frequency undulating component arising from off-center objects. In an attempt to model the cumulative effects of the two main contributing frequencies, in Figure 12, part c, the vector sums of high- and low-frequency artifact amplitudes [(lower frequency amplitude2 + higher frequency amplitude2)] are plotted. We used this approach rather than simply adding the amplitudes to average out the effect of phase coherence. Note that the plot closely resembles that obtained from measuring the artifact magnitude by means of SDr (Fig 8). This similarity is corroborated by the excellent correlation (Fig 13) found between the magnitude of artifacts measured with SDr (in millimeters) and the cumulative amplitudes of the contributing frequencies (Pearson r = 0.97, P < .001).
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Our results suggest that the height of the stair steps arising from the inclined rod in four-channel helical CT is also related to the distance between samples along the table translation direction. A somewhat unexpected finding was that when off center, superimposed on the stair steps, the rod surface was grossly distorted in an undulating fashion that was synchronized with the TI. Both artifact phenomena were directly proportional to the TI, but the magnitude of the lower frequency undulating component was greater than the high-frequency stair-step component and, particularly off center, dominated the visual appearance of the rod. In general, the magnitude of artifactual surface distortion of the rod phantom in four-channel helical CT was primarily a function of the TI, secondarily affected by the position of the rod relative to the isocenter, and only marginally affected by the pitch, detector collimation, and secondary z filtering.
From a practical point of view, the major result from this study is that the magnitude of artifacts—whether measured as the surface distortion of the rod (SDr), modeled as amplitudes of contributing sine waves, or ranked subjectively—is always smaller in four- than in single-channel helical CT. Specifically, when the size of stair steps is compared between single- and four-channel helical CT, single-channel helical CT leads to three times larger stair steps arising from inclined objects than does four-channel helical CT for any given TI as long as the object is located at or close to the isocenter. When off center, however, the advantage of the diminished size of stair steps seen with four-channel helical CT is reduced by newly introduced undulating low-frequency artifactual surface distortions. This unequivocal advantage of four-channel helical CT is furthermore amplified by the fact that, for a given TI, the effective section thickness (FWHM) of reconstructed transverse sections is smaller in four-channel helical CT (z filtering always allows reconstruction of thicker sections, if considered necessary). This is illustrated in Figure 15, in which the magnitude of artifacts is plotted against TI and each data point is annotated with horizontal whiskers that indicate the size of the FWHM of the respective series, measured by using an aluminum phantom (13).
|
Therefore, for high-quality CT angiographic applications, use of a TI of 15 mm per 360° gantry revolution, which corresponds to a table translation speed of 18.75 mm/sec with 0.8-second gantry rotation, allows a substantial improvement over single-channel helical CT. In most individuals, this would allow inclusion of the entire thoracic and abdominal aorta from the supraaortal branches to the iliac bifurcation in the scanning range (14) and may allow CT angiography of the lower extremity runoff vessels (15).
The physics behind artifact phenomena observed in single-channel helical CT has been the focus of prior investigations. Wang and Vannier (8), with use of a longitudinally oriented cone phantom, showed that two symmetric stair steps in the transverse plane (on opposing surfaces of the cone) can be observed to rotate with a period equal to the TI in the reconstructed volume, resulting in interlaced spiraling stair steps at periods of half the TI. Wilting and Timmer (9) also showed that the pattern of cross-sectional distortion of an inclined rod on consecutive transverse images repeats with a period equal to half the TI along z. In all cases, the images were reconstructed with use of the 180° linear interpolation algorithm. The stair-step artifacts observed with single-channel helical CT in this study, however, appear at two distinct frequencies. The higher frequency artifact occurs at two stair-step artifacts per TI), which is in accord with the reported frequency of stair-step artifacts (8,9). This frequency corresponds to the distance between the samples in the z direction, which is 0.5 TI for images reconstructed with the 180° linear interpolation algorithm. Our quantitative analysis showed that the magnitude of this high-frequency component was independent of the object position relative to the isocenter (Fig 12, part a). The lower frequency artifact component we observed in this study was characterized by a frequency of one stair-step artifact per TI, and its frequency was clearly dependent on the object location relative to the isocenter of the scanner (Fig 12, part b). The exact source of this low-frequency artifact component still needs to be identified. Longitudinal aliasing associated with the sampling geometry of helical CT might play a role. Yen et al (6) have shown that longitudinal aliasing is a spatially varying phenomenon that is minimal at the isocenter but worsens with distance from the isocenter, similar to the low-frequency artifact component observed in this study.
Analogous to the artifact components seen with single-channel helical CT, we observed distinct artifact components in four-channel helical CT which, when summed, account for the complex appearance of stair-step artifacts. The particular geometry of four simultaneously acquired data sets (interlaced with their derived four 180°-opposed complementary data sets) determines the distance between samples in z to equal one per sixth of the TI with both 0.75 and 1.5 pitches. The details of the reconstruction algorithm for four-channel helical CT with use of real and complementary data sets are given elsewhere (11). Analogous to single-channel helical CT, we may thus expect the highest frequency artifact component to appear at six stair-step artifacts per TI. This was observed on all four-channel helical CT scans except the secondary series, on which such high frequencies were not detectable due to the use of a reconstruction interval that was half the nominal section thickness. Again, the magnitude of the high-frequency artifact component was independent of the object-to-isocenter distance (Fig 12, part a). The low-frequency undulating artifact phenomena observed in four-channel helical CT in this study, however, are clearly dependent on the object-to-isocenter distance (Fig 12, part b) and their frequency is TI—and thus view angle (gantry angle)—dependent. It is worth reiterating that on the basis of both quantitative and qualitative assessments, the low-frequency artifacts are both more relevant to structural visualization and overall magnitude of the surface distortions than are the higher frequency stair steps. The exact source of this phenomenon cannot be fully explained on the basis of our data or accessible nonproprietary information about the reconstruction technique. Consequently, we cannot estimate the effect of the reconstruction process itself on the nature of the observed artifacts, which therefore awaits further theoretic and experimental evaluation.
A potential limitation of phantom studies is that quantitatively obtained results of image quality cannot be translated directly into the visual perception of human readers. In this study, we tried to control for this aspect by rendering the image data in 3D to parallel the quantitative analysis with subjective readings. Despite minor disagreement between observers, whether sharp and grainy or smooth surfaces are more representative of the rod, subjective rankings correlated excellently with quantitative measurements of artifacts, thus corroborating the validity of our results. Results of both quantitative and qualitative analyses agreed that the small high-frequency stair steps and ripples are much less important than the gross undulating surface distortions observed in four-channel helical CT.
Opposed to clinical CT angiography, our phantom measurements were obtained from a single "vessel" with only one diameter, at one predefined tilt angle relative to the z axis (45°), and with only one vessel-to-background contrast. Although the measurements of artifacts would almost certainly yield different values with use of varying angulation, size, and object-to-background contrast, this should not affect the general proportions of the relationships between artifact magnitude and acquisition parameters, nor should it alter the observed frequencies of the artifact components. Further, objects smaller than the TI will not be deformed by the low-frequency artifacts in four-channel helical CT but rather displaced. For larger, curved surfaces, such as the air-insufflated colon in CT colonography, low-frequency undulating artifacts may not have the same importance as those seen in obliquely oriented vessels. On the other hand, small stair steps might become more important with the higher object-to-background contrast and greater magnification of surface detail used with most immersion visualization techniques.
All series supposedly obtained with the rod at the isocenter necessarily include off-center effects, as the inclined rod crossed the isocenter rather than being aligned with it. This implies that we could not have obtained pure isocenter data, but on the other hand, this allows observation of the continuous increase in magnitude of some artifact phenomena along the rod with increasing distance from the isocenter (Figs 9–11). This effect displays the off-center effects on a continuous scale, as opposed to our otherwise two-point data represented by the isocenter and 100-mm off-isocenter design. From the line graphs and rendered images, it is obvious that off-center artifacts start to evolve immediately off center, and the figures suggest a linear increase with distance from the isocenter. Thus on the basis of our observations, we theorize that the low-frequency artifact component will not occur at isocenter, where real and complementary projections are identical. As the distance from isocenter increases, the disparity between real and complementary projections increases and thus the low-frequency artifacts achieve increasingly greater magnitude with distance from the isocenter. In distinction, however, the high-frequency artifact component is constant across the field of view, independent of distance from isocenter.
We conclude from this study that the inherent advantages of four-channel helical CT over single-channel helical CT with regard to speed and longitudinal resolution capabilities are not deteriorated by the cost of larger artifacts. In fact, four-channel helical CT consistently leads to fewer stair-step–like surface distortions of longitudinally inclined objects than does single-channel helical CT at any given TI, or four-channel helical CT incurs artifacts comparable to those with single-channel helical CT at approximately three times greater TI. The weakest part of image quality in four-channel helical CT in this study was its sensitivity to artifacts that appear at off-center locations.
Glossary
Beam width.—In single-channel helical CT, nominal width of x-ray fan beam, constrained by collimation at the source.
Detector width.—In four-channel helical CT, the width of each of four detector rows used to acquire four simultaneous data sets (channels). Each row is connected (switched) with one to four detector elements each with 1.25-mm diameter in the z direction, resulting in possible detector widths of 1.25, 2.5, 3.75, and 5 mm.
Four-channel helical CT.—CT system equipped with 16 detector rows in the z direction that is capable of simultaneously acquiring data (sections) from four channels, each of which combines one to four detector elements in the z direction.
FWHM.—FWHM of the section-sensitivity profile of transverse sections (measured by using a 15-µm-thick aluminum disk [13]).
Nominal section thickness.—In four-channel helical CT, with use of two-point (180°) linear interpolation, the nominal section thickness is identical to the detector width. Those acquisitions are referred to as "primary series." The use of z filtering (or variable thickness filtering) allows more than two points to be used for section interpolation. This allows reconstruction of sections with a 1.33, 1.5, or 2 times greater nominal section thickness than is possible with two-point interpolated images. Those acquisitions are referred to as "secondary series."
Pitch.—TI divided by nominal beam width for single-channel helical CT, and TI divided by four times the detector width for four-channel helical CT.
Single-channel helical CT.—Standard single detector-row helical CT.
TI.—Table translation distance (in millimeters) in one 360° gantry rotation.
|
FOOTNOTES |
|---|
Abbreviations: FWHM = full width at half maximum, SDr = SD of radius measurements along the rod, TI = table increment, 3D = three-dimensional
Author contributions: Guarantors of integrity of entire study, D.F., G.D.R., S.N.; study concepts, D.F., G.D.R., S.N.; study design, D.F., G.D.R., D.S.P., S.N.; definition of intellectual content, D.F., G.D.R., S.Y.Y., S.N.; literature research, D.F., S.Y.Y.; experimental studies, D.F., G.D.R., D.S.P.; data acquisition, D.F., G.D.R.; data analysis, D.F., G.D.R., D.S.P., P.R.H., C.F.B., S.N.; statistical analysis, D.F.; manuscript preparation, D.F., D.S.P., S.Y.Y.; manuscript editing, G.D.R., S.Y.Y., S.N.; manuscript review, D.F., G.D.R., P.R.H., C.F.B., S.N.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E J CHOI, Y-W OH, S Y HAM, K Y LEE, and E-Y KANG Comparison between coronal reformatted images and direct coronal CT images of the swine lung specimen: assessment of image quality with 64-detector row CT Br. J. Radiol., June 1, 2008; 81(966): 463 - 467. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. D. Dodd, C. A. Souza, and N. L. Muller High-resolution MDCT of pulmonary septic embolism: evaluation of the feeding vessel sign. Am. J. Roentgenol., September 1, 2006; 187(3): 623 - 629. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. D. Dodd, C. A. Souza, and N. L. Muller Conventional high-resolution CT versus helical high-resolution MDCT in the detection of bronchiectasis. Am. J. Roentgenol., August 1, 2006; 187(2): 414 - 420. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. R. Krestan, H. Noske, V. Vasilevska, M. Weber, G. Schuel |
) and the surface of the rod, with use of an attenuation threshold of 0 HU. SDr in the coronal plane served as the measure of artifactual surface distortion of the rod.
) and 2.0 (
) pitch series obtained with beam widths of 1, 3 , 5, and 7 mm are shown. For four-channel helical CT, the 0.75 pitch (
, HS) primary series and the 22.5-mm TI, 5-mm nominal section thickness secondary series are shown. The whiskers represent the width of the FWHM of the various sections in millimeters. Note that apart from invariably smaller artifact magnitude at any given TI (as indicated by the oblique line separating the two CT systems), the FWHM of the four-channel helical CT primary series is consistently smaller than that for single-channel helical CT. The horizontal dotted lines at isocenter of 0.16-mm and off center of 0.27-mm SDr indicates the artifact magnitude of the 3-mm beam width, 2.0 pitch single-channel helical CT series. Note that at the isocenter, comparable artifact magnitude and similar FWHM is maintained with four-channel helical CT at a TI of 22.5 mm. Off center, however, four-channel helical CT reaches the artifact level corresponding to 3-mm beam width, 2.0 pitch single-channel helical CT as early as at TI of approximately 15 mm.