Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online before print August 27, 2003, 10.1148/radiol.2291020642
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2291020642v1
229/1/195    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Arac, M.
Right arrow Articles by Isik, S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Arac, M.
Right arrow Articles by Isik, S.
(Radiology 2003;229:195-199.)
© RSNA, 2003

Thoracic Imaging

Lung at Thin-Section CT: Influence of Multiple-Segment Reconstruction on Image Quality1

Mehmet Arac, MD, A. Yusuf Oner, MD, Halil Celik, MD, Sergin Akpek, MD and Sedat Isik, MD

1 From the Department of Radiology, Gazi University School of Medicine, Kat, Besevler, Ankara, Turkey. Received May 29, 2002; revision requested July 30; final revision received January 31, 2003; accepted May 14. Address correspondence to M.A. (e-mail: meharac@yahoo.com).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
PURPOSE: To evaluate multiple-segment reconstruction to reduce cardiac-motion artifacts on thin-section computed tomographic (CT) images in the lung.

MATERIALS AND METHODS: Fifty patients were enrolled in the study. All images were obtained with a scanner capable of 1-second revolution time. Routine lung thin-section CT examination was performed with images reconstructed with bone algorithm. Multiple-segment images reconstructed with lung algorithm were obtained for three levels in the left paracardiac region. Segment images were reconstructed retrospectively with data for 225° rotation rather than the 360° rotation used for a complete scan. To minimize differences resulting from reconstruction algorithms, additional nonsegmented reconstruction was performed with lung algorithm. Three radiologists reviewed each set of images and assigned a quality score. Multiway analysis of variance was performed to compare motion artifact reduction with 225° and 360° reconstructions.

RESULTS: Differences were not significant (P > .05) between scores for images reconstructed with bone or lung algorithms. Differences were significant between scores for reconstructed images obtained with the combination of 360° bone and 225° segment algorithms (P < .001) and for those obtained with the combination of 360° lung and 225° segment algorithms (P < .001).

CONCLUSION: Multiple-segment reconstruction of lung thin-section CT images is an effective technique for reducing cardiac-motion artifacts without increasing patient dose.

© RSNA, 2003

Index terms: Computed tomography (CT), artifact • Computed tomography (CT), image processing, 60.12115, 60.12118 • Lung, CT, 60.12115, 60.12118


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Thin-section computed tomography (CT) has become a useful technique for the detection and characterization of diffuse lung disease. Findings that are nonspecific or not apparent on chest radiographs and conventional CT images are depicted easily with the high spatial resolution used in thin-section CT. Anatomic detail provided on such scans allows refinement in the localization of disease processes to the small airways, airspaces, or alveolar walls (1).

Technical developments in the past decade have helped decrease the scan time for helical CT scanners. With most of the currently available scanners, a gantry revolution can be performed in 1 second or less. Despite the reduced acquisition time, however, the quality of thin-section CT images is often compromised because of cardiac-motion artifacts, especially when small anatomic structures (ie, lung vessels, bronchi) or small abnormal structures are imaged next to the heart. Cardiac-motion artifacts may result in double images of pulmonary vessels and fissures and can mimic the appearance of bronchiectasis (2).

Reduced acquisition time (<1 second) and cardiac gating are two methods that can be used alone or in combination to reduce cardiac-motion artifacts, but they are not widely available. A third technique is the use of multiple-segment reconstruction options, which are available on most currently available scanners. The latter technique is a retrospective reconstruction method that allows segmentation of scan data into time-related images and reconstruction of multiple images from a single scan acquisition. Diagrams in Figure 1 show the relationship between multiple-segment reconstruction and cardiac cycle. The purpose of our prospective study was to evaluate the effect of multiple-segment reconstruction on the reduction of cardiac-motion artifacts on lung thin-section CT images.



View larger version (29K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1a. Diagrams show relationship between cardiac cycle and image reconstruction in a 1-second clockwise rotation. Inner circles represent the scanning period; the starting points (S1-S6) of data acquisition for each segment image are marked. Outer circles represent the two phases of cardiac cycle for heart rate of 83 beats per minute ({approx}220-msec systole and {approx}500-msec diastole per cycle). D1 = first diastole, D2 = second diastole. (a) Best case scenario: Scanning is started simultaneously with cardiac diastole (D1). In this case, one systole occurs during the 1-second scan time. The sixth segment is free of systole; therefore, cardiac-motion artifacts are reduced. (b) Worst case scenario: Scanning is started simultaneously with cardiac systole (SY1). In this case, two systoles (SY1 and SY2) occur during the 1-second scan time. There is no way to reconstruct a segment image that is completely free of systole. Even in this case, however, the second segment contains a smaller part of systole than is depicted on the 1-second scan, Therefore, the resultant image is less affected by cardiac-motion artifacts.

 


View larger version (30K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1b. Diagrams show relationship between cardiac cycle and image reconstruction in a 1-second clockwise rotation. Inner circles represent the scanning period; the starting points (S1-S6) of data acquisition for each segment image are marked. Outer circles represent the two phases of cardiac cycle for heart rate of 83 beats per minute ({approx}220-msec systole and {approx}500-msec diastole per cycle). D1 = first diastole, D2 = second diastole. (a) Best case scenario: Scanning is started simultaneously with cardiac diastole (D1). In this case, one systole occurs during the 1-second scan time. The sixth segment is free of systole; therefore, cardiac-motion artifacts are reduced. (b) Worst case scenario: Scanning is started simultaneously with cardiac systole (SY1). In this case, two systoles (SY1 and SY2) occur during the 1-second scan time. There is no way to reconstruct a segment image that is completely free of systole. Even in this case, however, the second segment contains a smaller part of systole than is depicted on the 1-second scan, Therefore, the resultant image is less affected by cardiac-motion artifacts.

 

   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Between December 2001 and May 2002, 50 patients (34 men and 16 women; age range, 29–80 years; mean age, 59.35 years) referred to our institution to undergo thin-section CT examination of the lung were enrolled in our study. Inclusion criteria were a clinical need for imaging of the left paracardiac lung parenchyma and a good breath-hold capability. All of the referred patients met these criteria. A summary of indications is provided in Table 1. Institutional review board approval and informed consent from each patient were obtained for this study. Heart rate was recorded for each patient just before the CT examination. Average heart rate was 83 beats per minute (range, 60–116 beats per minute).


View this table:
[in this window]
[in a new window]

  		TABLE 1. Indications for Thin-Section CT of Lung

 
Image Acquisition and Technical Parameters
All images were obtained during full inspiratory breath hold with a spiral CT scanner (CT/i Hispeed; GE Medical Systems, Milwaukee, Wis) capable of 1-second revolution time without electrocardiographic gating. Routine thin-section CT was performed, with scans acquired every 10 mm and reconstructed with bone algorithm. Section thickness was 1 mm, and acquisition parameters were 120 kV and 200 mA. Field of view was 25 cm, and each hemithorax was evaluated separately. For the purpose of this study, after the examination was completed, multiple-segment reconstructed images from three sections in the left paracardiac area were obtained retrospectively. The first section represented the aortic root plane; the second section, the four-chamber plane; and the third section, the anatomic plane where the inferior vena cava opened to the right atrium.

Segment images were reconstructed with data from only a 225° rotation, rather than the 360° rotation used for a complete 1-second scan. Therefore, the reconstructed images contained data from 0.6 second of the 1-second scan time. For each level, six-segment reconstructed images were obtained. For the same level, each segment image was reconstructed with a delay of approximately 166 msec. The use of other reconstruction options, such as half scanning, was also considered. Because of the fan beam angle in this option, the same 225° data were used to reconstruct a single image compared with the six images obtained with multiple-segment reconstruction. This decrease in the total number of reconstructed images could dramatically alter the capability of reducing cardiac-motion artifacts. Therefore, no other technique was used in this study.

Of the six images obtained with multiple-segment reconstruction, the one with the lowest degree of cardiac-motion artifact was chosen. The criteria for choosing one segment image from among the other segment images for the same level was the clarity of the lung parenchyma, fissure, cardiac contours, small vessels, and bronchi. For this purpose, all six-segment images were reviewed by an experienced radiologist (S.A.), and the image with the lowest degree of motion artifact was selected. This radiologist did not participate in image scoring.

Neither high–spatial-frequency bone nor sharp reconstruction algorithms were available with segment reconstruction. Segment images were available with different algorithms, such as the lung kernel. To minimize differences as a result of the reconstruction algorithms and to allow a better comparison, an additional nonsegmented (360°-rotation) reconstruction was obtained with lung algorithm. As a result, sets of images were available for each level: Each set consisted of one 360°-rotation thin-section CT image reconstructed with bone algorithm, a second 360°-rotation thin-section CT image reconstructed with lung algorithm, and a third 225°-rotation segment image chosen from the six-segment images reconstructed with lung algorithm. All sets of images for each level were acquired in random order with lung windows (level, -600 HU; width, 1,500 HU). The time for multiple-segment reconstruction and image acquisition was measured.

Image Analysis
Three radiologists (M.A., A.Y.O., H.C.), who were blinded to the reconstruction technique, independently reviewed each set of images and assigned a quality score to each. Before they scored the images, the readers observed two scans that were obtained in patients who were not included in the study, and they assigned a score together to ensure consistency in use of the grading scheme.

Quality scores were assessed on a four-point scale from 1 (no motion artifact) to 4 (excessive motion artifacts). Clarity of the lung parenchyma, fissure, cardiac contours, small vessels, and bronchi formed the basis for the judgement of motion. Because only patients with good ability to hold their breath were included in the study, motion artifacts were considered to be of cardiac origin.

Statistical Analysis
Scores attributed by each reader for different reconstruction techniques were averaged and expressed as the mean plus or minus SD. Statistical analysis was performed (SPSS, version 10.0; SPSS, Chicago, Ill). Multiway analysis of variance was performed between reconstruction techniques to evaluate reduction in motion artifacts, and {kappa} scores were calculated to assess interreader agreement. A P value of less than .05 was considered to indicate a statistically significant difference.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Time to complete multiple-segment reconstruction and image acquisition was 2 minutes. Mean motion artifact scores assigned by each radiologist for each reconstruction technique are shown in Table 2. Differences in scores between readers were not significant (P > .05). Differences between quality scores of images reconstructed with 360° combined bone and lung algorithm were not significant (P > .05). The {kappa} scores ranged between 0.32 and 0.56 (P < .05), which reflects a moderate to good interobserver agreement. A set of representative CT images for each technique is presented in Figures 2 and 3.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Mean Quality Scores for Reconstruction Algorithms

 


View larger version (53K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. In a 38-year-old female patient with known bronchiectasis, images were obtained at anatomic plane where inferior vena cava opens to right atrium. Top: On transverse 360° thin-section CT scan reconstructed with bone algorithm, blurred cardiac margin, double-line artifacts of fissure (curved arrow) and vessels, and distortion of bronchi (straight arrow) mimic appearance of bronchiectasis. Bottom: On segment image reconstructed with lung algorithm, cardiac border is delineated clearly and outline of fissure is sharp, but star artifacts as a result of cardiac pulsatility are not totally eliminated.

 


View larger version (90K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. Images in a 37-year-old male patient with cystic bronchiectasis. Imaging level is cardiac "four-chamber" plane. Left: On transverse 360° thin-section CT image reconstructed with bone algorithm, outlines of cystic bronchiectasis are relatively sharp in peripheral lung parenchyma but are severely distorted in paracardiac region (arrow). Right: On segment image reconstructed with lung algorithm, outlines of cystic bronchiectatic changes are delineated clearly in paracardiac region, as well as in peripheral lung parenchyma. Note that cardiac margin is sharper on the right image than on the left.

 
Scores assigned for multiple-segment reconstructed images were significantly lower than those for 360° thin-section CT images reconstructed with combined bone and lung algorithm (P < .001); this result reflects higher image quality obtained with segment reconstruction (Fig 4). Figure 5 shows a set of thin-section CT images acquired at the same left paracardiac level and reconstructed with the three algorithms. There was no difference in cardiac-motion artifacts between 360° thin-section CT images reconstructed with combined bone and lung algorithm. On corresponding multiple-segment reconstructed images, however, artifacts caused by cardiac pulsatility were largely eliminated. When each of the six-segment reconstructed images for the same level were evaluated, each image depicted a different degree of cardiac-motion artifacts (Fig 6). Figures 2, 3, and 5 show a comparison of segment and nonsegment reconstructed images obtained at the same paracardiac level in three patients. We noted that the multiple-segment reconstruction technique was successful in the reduction of, although not the total elimination of, cardiac-motion artifacts.



View larger version (29K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4. Bar graph shows mean image quality scores assigned by different observers to thin-section CT images reconstructed with the three algorithms. Black bars = 360° thin-section CT images reconstructed with bone algorithm. Gray bars = 360° thin-section CT images reconstructed with lung algorithm. White bars = multiple-segment reconstructed images.

 


View larger version (56K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5. Images in a 35-year-old female patient suspected of having bronchiectasis but without radiologic findings in the left paracardiac region. Imaging plane is cardiac four-chamber level. Left: A 360° image reconstructed with bone algorithm. Middle: A 360° image reconstructed with lung algorithm. Right: Segment image reconstructed with lung algorithm. Note clear delineation of cardiac border (black arrow) on right image compared with that on left and middle images. Double-image artifacts due to distorted pulmonary fissure and vessels in left and middle images were eliminated (white arrow) with segment reconstruction.

 


View larger version (116K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6. (A-F) Six segment reconstructions obtained in the same patient and at the same level as in Figure 5. Note that not all images are free of cardiac-motion artifacts. A is the best image in terms of reduction of cardiac-motion artifacts, whereas C and D are the worst images, with blurred cardiac margins and distorted pulmonary vessels.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Image quality of CT scans has improved since their initial introduction. Shortened scan times have been a major factor in this improvement. In the 1980s, scan times were 2–5 seconds. Scan time of 1 second has become available routinely with spiral imaging (3,4). Currently, subsecond imaging is possible with faster spiral and multisection scanners.

Thin-section CT is now accepted as the method of choice for evaluation of lung parenchyma. Temporal resolution has improved with widespread use of spiral scanners capable of 1-second scan time (5). Despite the reduced acquisition time, however, the quality of thin-section CT images is often blurred by cardiac-motion artifacts when small structures next to the heart are imaged. These artifacts may result in double images of small bronchi, pulmonary vessels, and fissures, as well as doubling or blurring of the cardiac border, low-attenuating areas adjacent to the vessel, star artifacts, dark spots, and distortion of normal structures. These motion artifacts degrade image quality and can sometimes mimic bronchiectasis (2).

Corrective reconstruction, postprocessing, tube alignment, fast scanning, and electrocardiographic gating are methods used to reduce motion artifacts (6–9). Ritchie et al (4) stated that scan times of 0.6 second or electron-beam CT with a scan time of 100 msec are not sufficient to eliminate all motion artifacts. Complete elimination of motion artifacts necessitates an acquisition time of less than 19.1 msec (4). This scan time is far from the capability of currently available CT scanners, even the newest multisection machines.

In the present study, multiple-segment reconstructed images were obtained with data from a 225° rotation instead of the 360° rotation for a complete 1-second scan. In this way, temporally overlapping images were obtained at 166-msec intervals, with each image containing data from the 0.6-second period. The success of motion-artifact reduction increased when cardiac systole occupied a small part in the data used for image reconstruction. Cardiac systole typically occupies approximately 30% of the cardiac cycle (10). With an average heart rate of 83 beats per minute, the systolic phase occupies approximately 220 msec. In the best case scenario in the present study, at least one of the multiple-segment reconstructed images was free of the systolic phase when data collection began just after cardiac systole was completed. On the other hand, in the worst case scenario in the current study, images obtained with a 1-second scan time were affected by cardiac-motion artifacts when data collection began simultaneously with systole. Even in the worst case, however, one of the six-segment reconstructed images contained a smaller part of systole than was depicted on a 1-second scan, which contained two complete systoles. This explains the success in reduction of cardiac-motion artifact on segment reconstructed images.

Bone or other high–spatial-frequency algorithms were not available with segment reconstruction. To prevent faulty evaluation of motion artifacts as a result of different image reconstruction algorithms, nonsegmented (360°) images reconstructed with lung algorithm were also included in the comparison for each level. Differences in terms of motion artifacts between thin-section CT images reconstructed with bone algorithm and nonsegmented (360°) images reconstructed with lung algorithm were not significant. When segment images reconstructed with lung algorithm were compared with images obtained with each of the other two techniques, a statistically significant reduction in motion artifacts was found (P < .001).

An important issue with multiple-segment reconstructed images is noise. Increased noise can cause diagnostic problems in the detection of ground-glass areas. Nevertheless, findings in several studies show that with increased noise, there is only minimal compromise in diagnostic information, which is not likely to affect image quality in the lung (11). Additionally, with the multiple-segment reconstruction technique for thin-section CT scans used in the current study, images obtained with bone algorithm and those obtained with multiple-segment technique are obtained simultaneously. If there is any diagnostic ambiguity, the 360° thin-section CT images can be used.

An advantage of segment reconstruction is that no additional scanning is required, and there is no increase in patient dose. The whole procedure is completed without any increase in cost or scan time. The only cost is the additional time for postprocessing. A trained technician can complete all the postprocessing and acquisition of the multiple-segment reconstructed images in 2 minutes. With these benefits, segment reconstructed images can be used successfully to reduce cardiac-motion artifacts in thin-section CT images of the lung. For every thin-section CT image, six segment reconstructions are provided. This can result in an increased number of images stored for each patient. Therefore, dynamic evaluation of thin-section CT images and selection of those that contain motion artifacts suitable for multiple-segment reconstruction is advisable for clinics with a high workload.

Findings in this study demonstrate that multiple-segment reconstructions of thin-section CT images show decreased cardiac-motion artifacts in lung parenchyma. This option is more widely available than is electrocardiographic gating. It can be used with selected thin-section CT images containing cardiac-motion artifacts without necessitating additional cost, patient dose, or scan time.


   FOOTNOTES
 
Author contributions: Guarantor of integrity of entire study, M.A.; study concepts, M.A., H.C.; study design, M.A., A.Y.O., H.C.; literature research, A.Y.O., H.C.; clinical studies, M.A., A.Y.O., H.C.; data acquisition, A.Y.O., H.C.; data analysis/interpretation, S.A., M.A., A.Y.O., H.C.; statistical analysis, A.Y.O., H.C.; manuscript preparation, M.A., A.Y.O.; manuscript definition of intellectual content and editing, M.A., A.Y.O., S.A.; manuscript revision/review and final version approval, M.A., S.I.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Gutierrez FR, Woodard PK, Fleishman MJ, Semenkovich JW, Anderson DC. Thorax: techniques and normal anatomy. In: Lee JKT, Sagel SS, Stanley RJ, Heiken JP, eds. Computed body tomography with MRI correlation. 3rd ed. Philadelphia, Pa: Lippincott-Raven, 1998; 183-260.
  2. Tarver RD, Conces DJ, Godwin JD. Motion artifacts on CT simulate bronchiectasis. AJR Am J Roentgenol 1988; 151:1117-1119.[Free Full Text]
  3. Posniak HV, Olson MC, Demos TC, Pierce KL, Kalbhen CL. CT of the chest and abdomen in patients on mechanical pulmonary ventilators: quality of images made at 0.6 vs 1.0 sec.. AJR Am J Roentgenol 1994; 163:1073-1077.[Abstract/Free Full Text]
  4. Ritchie CJ, Godwin JD, Crawford CR, et al. Minimum scan speeds for suppression of motion artifacts in CT. Radiology 1992; 185:37-42.[Abstract/Free Full Text]
  5. Rubin GD, Leung AN, Robertson VJ, Stark P. Thoracic spiral CT: influence of subsecond gantry rotation on image quality. Radiology 1998; 208:771-776.[Abstract/Free Full Text]
  6. Montaudon M, Berger P, Blachere H, De Boucaud L, Latrabe V, Laurent F. Thin section CT of the lung: influence of 0.5-s gantry rotation and ECG triggering on image quality. Eur Radiol 2001; 9:1681-1687.
  7. Helenon O, Chanin DS, Laval-Jeantet M, Frija J. Artifacts on lung CT scans: removal with Fourier filtration. Radiology 1989; 171:572-574.[Abstract/Free Full Text]
  8. Goldberg HI, Gould RG, Feuerstein IM, Sigeti JS, Lipton MJ. Evaluation of ultrafast CT scanning of the adult abdomen. Invest Radiol 1989; 24:537-543.[CrossRef][Medline]
  9. Moore SC, Judy PF, Garnic J. Prospectively gated cardiac computed tomography. Med Phys 1983; 10:846-855.[CrossRef][Medline]
  10. Remy-Jardin M, Baghaie F, Bonnel F, Masson P, Duhamel A, Remy J. Thoracic helical CT: influence of subsecond scan time and thin collimation on evaluation of peripheral pulmonary arteries. Eur Radiol 2000; 10:1297-1303.[CrossRef][Medline]
  11. Lee KS, Primack SL, Staples CA, et al. Chronic infiltrative lung disease: comparison of diagnostic accuracies of radiography and low and conventional dose thin section CT. Radiology 1994; 191:669-673.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Am. J. Roentgenol.Home page
F. Liu, C. Cuevas, A. A. Moss, O. Kolokythas, T. J. Dubinsky, and P. E. Kinahan
Gas Bubble Motion Artifact in MDCT
Am. J. Roentgenol., February 1, 2008; 190(2): 294 - 299.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Roentgenol.Home page
H. I. Ha, H. W. Goo, J. B. Seo, J.-W. Song, and J. S. Lee
Effects of high-resolution CT of the lung using partial versus full reconstruction on motion artifacts and image noise.
Am. J. Roentgenol., September 1, 2006; 187(3): 618 - 622.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2291020642v1
229/1/195    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Arac, M.
Right arrow Articles by Isik, S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Arac, M.
Right arrow Articles by Isik, S.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE