HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2261010394v1 226/1/231    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kozuka, T. Articles by Kudo, M. Search for Related Content PubMed PubMed Citation Articles by Kozuka, T. Articles by Kudo, M.

Detection of Pulmonary Metastases with Multi–Detector Row CT Scans of 5-mm Nominal Section Thickness: Autopsy Lung Study¹

¹ From the Department of Radiology, Osaka University Medical School, 2-2 Yamadaoka, Suita, Osaka 565-0871, Japan (T.K., T.J., S.H., H. Naito, N.T., M. Koyama, N.M., O.H., H. Nakamura); and GE Yokogawa Medical Systems, Hino, Japan (M. Kudo). Received January 29, 2001; revision requested March 21; final revision received March 21, 2002; accepted April 30. Address correspondence to T.K. (e-mail: koz@radiol.med.osaka-u.ac.jp).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
PURPOSE: To determine the effect of changing pitch and collimation on depiction of pulmonary metastases on scans of 5-mm section thickness obtained with multi–detector row computed tomography (CT) compared with those obtained with single–detector row CT.

MATERIALS AND METHODS: In five autopsy lungs, 1,013 metastatic 0.5–30.0-mm nodules were detected at helical CT with 1-mm collimation and histopathologically diagnosed as metastases. Each nodule was numbered, and its localization was recorded as the standard for subsequent studies. Four types of scans of 5-mm section thickness were obtained with multi–detector row CT and four sets of helical pitch and table speed, respectively, as follows: set A, 3:1 and 7.5 mm per rotation; set B, 6:1 and 15 mm per rotation; set C, 6:1 and 30 mm per rotation; set D, conventional and 5-mm interval. Conventional helical CT scans (set E) were obtained with 5-mm collimation at single–detector row CT. Two independent observers evaluated the five sets of CT scans.

RESULTS: Acquisition times for sets A–D, respectively, were 1.9, 3.8, 7.5, and 1.5 times faster than they were for set E. The mean numbers of detected nodules were 671 (66%) in set A, 661 (65%) in set B, 678 (67%) in set C, 654 (65%) in set D, and 656 (65%) in set E; there was no significant difference in the number of detected nodules among the five sets (P = .997, McNemar test and Bonferroni equation).

CONCLUSION: Regardless of varying pitch or detector collimation, multi– and single–detector row CT scans obtained with 5-mm section thickness have almost the same ability to depict pulmonary metastases and are equivalent.

© RSNA, 2002

Index terms: Computed tomography (CT), helical, 60.12115 • Computed tomography (CT), multi–detector row, 60.12119 • Experimental study • Lung, nodule, 60.332, 60.333 • Lung neoplasms, metastases, 60.33

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Multi–detector row computed tomography (CT) is a promising tool for the evaluation of lung parenchyma because its main advantages are shorter acquisition time than with conventional single–detector row CT and retrospective creation of both thinner and thicker sections from the same raw data (1,2). In fact, by using the fastest parameter (ie, 5-mm collimation, pitch of 6, and 0.8 second per gantry rotation), one helical scan of the whole lung could be acquired during one breath hold within 10 seconds.

The detection of pulmonary nodules is a common and important indication for diagnostic imaging of the lung, and CT has been shown to be the best imaging modality for this detection (3,4). At helical CT, it is possible to scan the entire lung in a single breath hold and, thus, to overcome the problem of misregistration caused by respiratory motion (5–7). Findings in comparative studies with identical section thicknesses and section intervals indicated that helical CT depicts more nodules, particularly in the case of small nodules, than does sequential CT (8,9).

To our knowledge, however, no assessment of the detection of metastatic nodules by using multi–detector row CT has yet been performed. The purpose of this study was to determine the effect of changing pitch and collimation on the detection of pulmonary metastases by using nominal 5-mm section thickness for scans obtained at multi–detector row CT compared with scans obtained at conventional single–detector row CT.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Five autopsy lungs were inflated and fixed (10). Scanning of the lungs was performed with a single–detector row CT system (HiSpeed Advantage; GE Medical Systems, Milwaukee, Wis) and the following parameters: 1-mm collimation, 1-second gantry rotation, 20-cm field of view (FOV), high-spatial-frequency algorithm, 120 kVp, and 200 mA per rotation. Two board-certified chest radiologists (T.K., T.J.) evaluated these CT scans together to identify nodules and to record the size of nodules by consensus. Detected nodules were histopathologically confirmed as metastases on a one-by-one basis. Each lung was cut into serial 1-mm-thick sections, which corresponded to the specimen CT scan sections, by using a microslicer (DTK-3000W; Dosaka EM, Kyoto, Japan). We obtained serial 20-µm-thick microscopic sections from the 1-mm-thick specimens and stained them with hematoxylin-eosin for the areas that corresponded to the nodules depicted by using CT with 1-mm section thickness.

Each nodule was numbered, and its location was recorded as the standard for subsequent studies. Before the five lungs were cut, they also were scanned with a multi–detector row CT scanner (LightSpeed QX/i, GE Medical Systems) and four sets of parameters, sets A, B, C, and D. For set A, 2.5-mm detector collimation, 7.5 mm-per-rotation table speed, and 3:1 pitch were used. For set B, 2.5-mm detector collimation, 15 mm-per-rotation table speed, and 6:1 pitch were used. For set C, 5-mm detector collimation, 30 mm-per-rotation table speed, and 6:1 pitch were used. For set D, conventional (ie, step-and-shoot) scanning was performed with 5-mm scanning collimation for four sections. Other parameters (ie, 5-mm section thickness, 5-mm reconstruction interval, 0.8-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation) were the same for the four sets. In addition, these lungs were scanned by using the conventional method with a single–detector row CT scanner (HiSpeed Advantage RP, GE Medical Systems) and the following parameters: 5-mm collimation, 3-mm reconstruction interval, 1-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation (set E).

The radiation dose (weighted CT index, milligray) on the indicator of the multi–detector row CT scanner was recorded. The radiation dose of set E was requested from the manufacturer because the radiation dose was not indicated on our single–detector row CT scanner. Thus, we did not actually measure the dose ourselves with our multi– and single–detector row CT scanners. Images were displayed on hard copy with 4 x 5-image format and a window level of -700 HU and a window width of 1,200 HU, which were appropriate for representation of lung parenchyma. Scanning time for the four sets (A–D) was compared with that for set E. Two board-certified chest radiologists evaluated all five sets to confirm the existence of all pathologically demonstrated nodules and made a standard model for each set of images on the basis of the CT scans with 1-mm section thickness. Another pair of board-certified chest radiologists (H.N., S.H.) who were different from the two who made the standard model and who each had more than 20 years of experience independently evaluated the five data sets. Differences in detection ability with sets A–E were analyzed with the McNemar test, and interobserver agreement was determined with the statistic.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
A total of 1,013 nodules were depicted with a single–detector row CT scanner, and all of them were confirmed histopathologically as metastases. The size of the detected nodules ranged from 0.5 to 30.0 mm (mean, 3.2 mm). Mean acquisition times for sets A–E were 25.5, 12.8, 6.4, 31.9, and 47.8 seconds, respectively. Acquisition times for sets A, B, C, and D were, respectively, 1.9, 3.8, 7.5, and 1.5 times faster than the acquisition time for set E. The radiation doses (weighted CT index, milligray) on the scanner indicators for sets A–E were 20.37, 10.19, 9.01, 13.51, and 14.76, respectively.

There was excellent agreement between the observers ( statistic, 0.81, 0.79, 0.74, 0.78, and 0.71 for sets A–E, respectively) for the detection of metastatic nodules (Table). There were no false-positive nodules detected by the two observers. The mean numbers of detected nodules were 671 (66%) for set A, 661 (65%) for set B, 678 (67%) for set C, 654 (65%) for set D, and 656 (65%) for set E (Figure); there was no significant difference in the number of detected nodules among the five sets (P = .997, McNemar test and Bonferroni equation).

View larger version (82K): [in this window] [in a new window] [Download PPT slide]   Figure a. Transverse CT scans of five inflated and fixed lungs removed at autopsy. Scans in a-d were obtained with multi-detector row CT and 5-mm section thickness, 0.8-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation. There was no statistically significant difference in the numbers of nodules detected between sets A-D and set E. (a) Set A. Scan obtained with 2.5-mm collimation, 7.5 mm-per-rotation table speed, and 3:1 pitch. (b) Set B. Scan obtained with 2.5-mm collimation, 15 mm-per-rotation table speed, and 6:1 pitch. (c) Set C. Scan obtained with 5-mm collimation, 30 mm-per-rotation table speed, and 6:1 pitch. (d) Set D. Scan obtained with conventional 5-mm section thickness and 5-mm interval. (e) Set E. Scan obtained with single-detector row CT and 5-mm collimation, 1-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation.

View larger version (81K): [in this window] [in a new window] [Download PPT slide]   Figure b. Transverse CT scans of five inflated and fixed lungs removed at autopsy. Scans in a-d were obtained with multi-detector row CT and 5-mm section thickness, 0.8-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation. There was no statistically significant difference in the numbers of nodules detected between sets A-D and set E. (a) Set A. Scan obtained with 2.5-mm collimation, 7.5 mm-per-rotation table speed, and 3:1 pitch. (b) Set B. Scan obtained with 2.5-mm collimation, 15 mm-per-rotation table speed, and 6:1 pitch. (c) Set C. Scan obtained with 5-mm collimation, 30 mm-per-rotation table speed, and 6:1 pitch. (d) Set D. Scan obtained with conventional 5-mm section thickness and 5-mm interval. (e) Set E. Scan obtained with single-detector row CT and 5-mm collimation, 1-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation.

View larger version (86K): [in this window] [in a new window] [Download PPT slide]   Figure c. Transverse CT scans of five inflated and fixed lungs removed at autopsy. Scans in a-d were obtained with multi-detector row CT and 5-mm section thickness, 0.8-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation. There was no statistically significant difference in the numbers of nodules detected between sets A-D and set E. (a) Set A. Scan obtained with 2.5-mm collimation, 7.5 mm-per-rotation table speed, and 3:1 pitch. (b) Set B. Scan obtained with 2.5-mm collimation, 15 mm-per-rotation table speed, and 6:1 pitch. (c) Set C. Scan obtained with 5-mm collimation, 30 mm-per-rotation table speed, and 6:1 pitch. (d) Set D. Scan obtained with conventional 5-mm section thickness and 5-mm interval. (e) Set E. Scan obtained with single-detector row CT and 5-mm collimation, 1-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation.

View larger version (82K): [in this window] [in a new window] [Download PPT slide]   Figure d. Transverse CT scans of five inflated and fixed lungs removed at autopsy. Scans in a-d were obtained with multi-detector row CT and 5-mm section thickness, 0.8-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation. There was no statistically significant difference in the numbers of nodules detected between sets A-D and set E. (a) Set A. Scan obtained with 2.5-mm collimation, 7.5 mm-per-rotation table speed, and 3:1 pitch. (b) Set B. Scan obtained with 2.5-mm collimation, 15 mm-per-rotation table speed, and 6:1 pitch. (c) Set C. Scan obtained with 5-mm collimation, 30 mm-per-rotation table speed, and 6:1 pitch. (d) Set D. Scan obtained with conventional 5-mm section thickness and 5-mm interval. (e) Set E. Scan obtained with single-detector row CT and 5-mm collimation, 1-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation.

View larger version (82K): [in this window] [in a new window] [Download PPT slide]   Figure e. Transverse CT scans of five inflated and fixed lungs removed at autopsy. Scans in a-d were obtained with multi-detector row CT and 5-mm section thickness, 0.8-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation. There was no statistically significant difference in the numbers of nodules detected between sets A-D and set E. (a) Set A. Scan obtained with 2.5-mm collimation, 7.5 mm-per-rotation table speed, and 3:1 pitch. (b) Set B. Scan obtained with 2.5-mm collimation, 15 mm-per-rotation table speed, and 6:1 pitch. (c) Set C. Scan obtained with 5-mm collimation, 30 mm-per-rotation table speed, and 6:1 pitch. (d) Set D. Scan obtained with conventional 5-mm section thickness and 5-mm interval. (e) Set E. Scan obtained with single-detector row CT and 5-mm collimation, 1-second gantry rotation, 20-cm FOV, standard algorithm, 120 kVp, and 200 mA per rotation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

An increase in the pitch causes widening of the section-sensitivity profile, and this widening of the section sensitivity profile will lead to obscurity of object that is due to the effect of greater partial volume (11). In previous studies (1), it was mentioned that the helical pitch should be limited to no more than 1.5 for the detection of pulmonary metastatic nodules at single–detector row CT because poor visibility of small nodules was caused by reduced lesion contrast when a larger pitch was used. Overlapping image reconstruction improves the ability to detect pulmonary nodules, especially when the size of the nodules is smaller than the section thickness of the helical CT (2,12,13).

Furthermore, thinner section thickness should improve detection of pulmonary metastatic nodules because of a reduction in the partial-volume effect. In daily clinical practice, however, a 7- to 10-mm section thickness usually is used for single–detector row CT, but multi–detector row CT can be used with a thinner section thickness and faster scanning time. In fact, with multi–detector row CT, a whole-lung scan with 5-mm collimation can be obtained within 10 seconds during one breath hold. In the study presented here, we evaluated the depiction of metastatic nodules on different sets of scans of 5-mm section thickness obtained with multi–detector row CT and compared the depiction with that at single–detector row CT. Our study results proved that the detection of pulmonary metastatic nodules by using multi–detector row CT with 5-mm section thickness is virtually the same regardless of the factors used.

Determination of the helical pitch depends on the CT system used for multi–detector row CT. With the multi–detector row CT system, two kinds of helical pitch, pitch of 3 and 6, are used (14). Helical pitch 3, called the high-quality mode, is used when it is important to obtain a detection effectiveness comparable to that produced by using single–detector CT systems with a pitch of 1:1 to 1.5:1. The table speed of the high-quality mode is 1.5–3.0 times faster than that of single–detector row CT systems with a 1:1 to 1.5:1 pitch, and the tube current is about 50%-60% of that with single–detector row CT systems. Helical pitch 6, called the high-speed mode, is used when it is important to attain table speeds that are three to six times faster than those required for 1:1 pitch helical single–detector row CT scans, but the detection effectiveness is comparable to that obtained with a 1.5:1 to 2:1 pitch with a single–detector row CT system. The acquisition time is twice as fast as that for the high-quality mode (14).

In our study, detection effectiveness for parameter set A was relatively high, and a thin-section CT scan of 2.5-mm section thickness could be constructed retrospectively. However, scanning time for one helical scan of the whole lung was more than 30 seconds, so acquisition of data within one breath hold was impossible. Therefore, this method was considered to be suitable for scanning focal or localized lesions. With parameter set B, one helical scan of the whole lung was possible within one breath hold, and thin-section CT with a 2.5-mm section thickness could be constructed retrospectively from the same raw data. With this method, it is possible to perform both screening and a detailed examination at the same time. For parameter set C, scans with 5-mm section thickness could be acquired the fastest, the breath-hold time was the shortest, and the radiation dose was the lowest. However, thin-section CT could not be reconstructed from this raw data set. Consequently, this method was limited to or more useful as a screening tool.

Hu et al (15) reported that the full width at half maximum of a CT scan obtained with a 6:1 pitch and a table speed of 30 mm per rotation was greater than that of a CT scan obtained with a 1:1 pitch. Theoretically, widening of the section-sensitivity profile leads to obscurity of object due to greater partial-volume effect. The values for full width at half maximum of the section-sensitivity profile for these five techniques of sets A, B, C, D, and E are 5.00, 5.00, 6.25, 5.00, and 5.00 mm, respectively, according to the manufacturer of the scanners. Although there was no significant difference in the number of detected nodules with the five sets, more nodules were depicted on CT scans obtained with a 30 mm-per-rotation table speed (set C) than on CT scans obtained with a 1:1 pitch (set E). This discordance was considered to be due to observational error.

This study has certain limitations. First, the detection rates of nodules obtained in our study were relatively low, and the mean rate was 60% of 1,013 nodules. We did not use the overlap reconstruction method, so small nodules were not detected because of the partial-volume effect (2,12,13). However, since retrospective overlap reconstruction is feasible for each series that we investigated, improvement in the detectability of small pulmonary nodules can be expected.

Second, we could not examine the effect of motion artifacts caused by cardiac pulsation or respiration because we used postmortem lungs for our study. Motion artifact is thought to affect detection.

Third, we could not assess the effect of chest wall and mediastinal structures because we used postmortem lungs. The presence of high-attenuating structures relatively far from isocenter will result in a greater degree of helical artifact, which may affect lung representation and would vary substantially with the different protocols used.

Fourth, there are now single–detector row CT scanners with subsecond gantry rotation. However, in this study, we used a single–detector row CT scanner with 1-second gantry rotation as a control because we did not have a single–detector row CT scanner with subsecond gantry rotation.

Last, this study included only a small number of cases, so further large-scale studies will be necessary. However, we believe our results are valid because this study includes a fairly large number of metastatic nodules, and reproducibility of this method is satisfactory, as demonstrated by the statistic for the results of both observers.

In conclusion, CT scans of 5-mm section thickness obtained with multi–detector row CT and various parameters have almost the same ability to depict pulmonary metastases regardless of varying pitch or detector collimation and are equivalent to scans obtained with conventional single–detector row CT.

Practical application: The scans of 5-mm section thickness obtained with multi–detector row CT and a table speed of 30 mm per rotation with 6:1 pitch are considered adequate for depicting pulmonary metastases and are equivalent to scans obtained with conventional single–detector row CT, on the basis of results of our autopsy lung study.

	FOOTNOTES

 
Abbreviation: FOV = field of view

Author contributions: Guarantor of integrity of entire study, T.K.; study concepts and design, T.K., T.J.; literature research, T.K., M. Koyama; clinical studies, T.K., N.M.; experimental studies, T.K., O.H.; data acquisition, T.K., T.J., O.H., M. Kudo; data analysis/interpretation, T.K., H. Naito, S.H.; statistical analysis, T.K., T.J.; manuscript preparation, T.K., N.T.; manuscript definition of intellectual content, T.K., T.J.; manuscript editing and revision/review, S.H.; manuscript final version approval, H. Nakamura.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

1. Rydberg J, Buckwalter KA, Caldemeyer KS, et al. Multisection CT: scanning techniques and clinical applications. RadioGraphics 2000; 20:1787-1806.[Abstract/Free Full Text]
2. Klingenbeck-Regn K, Schaller S, Flohr T, Ohnesorge B, Kopp AF, Baum U. Subsecond multi-slice computed tomography: basics and applications. Eur J Radiol 1999; 31:110-124.[CrossRef][Medline]
3. Muhm JR, Brown LR, Crowe JK, Sheedy PF, Hattery PR, Stephens DH. Comparison of whole lung tomography and computed tomography for detecting pulmonary nodules. AJR Am J Roentgenol 1978; 131:981-984.[Abstract]
4. Chang AE, Schaner EG, Conkle DM, Flye MW, Doppman JL, Rosenberg SA. Evaluation of computed tomography in the detection of pulmonary metastases: a prospective study. Cancer 1979; 43:913-916.[CrossRef][Medline]
5. Krudy AG, Doppman JL, Herdt JR. Failure to detect a 1.5 centimeter lung nodule by chest computed tomography. J Comput Assist Tomogr 1982; 6:1178-1180.[Medline]
6. Vock P, Soucek M, Daepp M, Kalender WA. Lung: spiral volumetric CT with single-breath-hold technique. Radiology 1990; 176:864-867.[Abstract/Free Full Text]
7. Costello P, Anderson W, Blume D. Pulmonary nodule: evaluation with spiral volumetric CT. Radiology 1991; 179:875-876.[Abstract/Free Full Text]
8. Friese SA, Rieber A, Fleiter T, Brambs HJ, Claussen CD. Pulmonary nodules in spiral volumetric and single slice computed tomography. Eur J Radiol 1994; 18:48-51.[CrossRef][Medline]
9. Remy-Jardin M, Remy J, Giraud F, Marquette CH. Pulmonary nodules: detection with thick-section spiral CT versus conventional CT. Radiology 1993; 187:513-520.[Abstract/Free Full Text]
10. Markardian B, Daily ET. Preparation of inflated lung specimens. In: Heitzman ER, eds. The lung: radiologic-pathologic correlations. 2nd ed. St Louis, Mo: Mosby, 1984; 4-12.
11. Polacin A, Kalender WA, Marchal G. Evaluation of section sensitivity profiles and image noise in spiral CT. Radiology 1992; 185:29-35.[Abstract/Free Full Text]
12. Diederich S, Lentschig MG, Winter F, Roos N, Bongartz G. Detection of pulmonary nodules with overlapping vs non-overlapping image reconstruction at spiral CT. Eur Radiol 1999; 9:281-286.[CrossRef][Medline]
13. Diederich S, Semik M, Lentschig MG, et al. Helical CT of pulmonary nodules in patients with extrathoracic malignancy: CT-surgical correlation. AJR Am J Roentgenol 1999; 172:353-360.[Abstract/Free Full Text]
14. McCollough CH, Zink FE. Performance evaluation of a multi-slice CT system. Med Phys 1999; 26:2223-2230.[CrossRef][Medline]
15. Hu H, He HD, Foley WD, Fox SH. Four multidetector-row helical CT: image quality and volume coverage speed. Radiology 2000; 215:55-62.[Abstract/Free Full Text]

This article has been cited by other articles:

				A. M. Parsons, E. K. Ennis, B. C. Yankaskas, L. A. Parker Jr, W. B. Hyslop, and F. C. Detterbeck Helical Computed Tomography Inaccuracy in the Detection of Pulmonary Metastases: Can It Be Improved? Ann. Thorac. Surg., December 1, 2007; 84(6): 1830 - 1836. [Abstract] [Full Text] [PDF]

				P. Peloschek, J. Sailer, M. Weber, C. J. Herold, M. Prokop, and C. Schaefer-Prokop Pulmonary Nodules: Sensitivity of Maximum Intensity Projection versus That of Volume Rendering of 3D Multidetector CT Data Radiology, May 1, 2007; 243(2): 561 - 569. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2261010394v1 226/1/231    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kozuka, T. Articles by Kudo, M. Search for Related Content PubMed PubMed Citation Articles by Kozuka, T. Articles by Kudo, M.