This Article Abstract Figures Only Full Text (PDF) Services 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 Google Scholar Google Scholar Articles by ISHIGAKI, S Articles by ISHIGAKI, T Search for Related Content PubMed PubMed Citation Articles by ISHIGAKI, S Articles by ISHIGAKI, T

Three-dimensional CT angiography of the pancreatic artery in 16-channel multislice CT: value of scanning with submillimetre collimation

¹ Department of Radiology, Nagoya University Graduate School of Medicine, ² Department of Technical Radiology, Nagoya University School of Health Sciences, Nagoya, Japan

Correspondence: Satoko Ishigaki, Department of Radiology, Nagoya University Graduate School of Medicine, 65 Tsurumai-cho, Showa-ku, Nagoya 466-8560, Japan. E-mail: satoko{at}med.nagoya-u.ac.jp

	Abstract

Top Abstract Introduction Methods and materials Results Discussion References

 
The aim of this study was to assess the effects of the reconstructed thickness of axial images on image quality of CT pancreatic arteriography with 16-channel multislice CT. In 31 consecutive patients, raw data of arterial-phase scanning with 0.5 mm collimation were reconstructed in the following three ways: 0.5 mm thickness (effective thickness of 0.75 mm) at 0.4 mm intervals in Group 1; 1 mm thickness at 0.5 mm intervals in Group 2; and 2 mm thickness at 1 mm intervals in Group 3. For the visualization of major arteries and small arteries of the pancreatic head, four blinded readers independently performed side-by-side comparison of the CT arteriographic images generated from each axial dataset for the same patient using a three-dimensional volume-rendered technique. In all comparisons using a continuous rating scale, CT arteriographic images generated from thinner axial images were found to be significantly superior (p<0.01). The difference was more pronounced for small arteries. The degree of degradation from Group 1 to Group 2 was markedly smaller than that from Group 1 to Group 3 or from Group 2 to Group 3. For small arteries, paired images were assigned a grade of "almost equivalent" in 73%, 6% and 15% of the comparison between Group 1 and Group 2, Group 1 and Group 3, and Group 2 and Group 3, respectively. We concluded that the image quality of CT pancreatic arteriography, especially for small arteries, can be improved by reconstructing axial images with thinner thickness from the data obtained with submillimetre collimation.

	Introduction

Top Abstract Introduction Methods and materials Results Discussion References

 
Surgical resection is currently the only curative therapy for various pancreatic neoplasms, but, even in expert hands, classical pancreaticoduodenectomy is associated with a mortality rate of about 5% [1, 2]. In order to perform surgical resection of the pancreas safely, it is essential to depict the arterial anatomy of the pancreas clearly. Since the anatomy is quite variable, preoperative assessment has traditionally been performed using conventional angiography, but this procedure suffers from the problem of being invasive. Since the development of helical computed tomography (CT) technology in the 1990s, CT angiography has been used to assess the arterial anatomy of the pancreas [3, 4].

In single-slice CT, it is recommended that the thinnest collimation available be employed with an overlapping reconstruction interval of 50% in order to improve the quality of CT angiographic images [5, 6]. However, in single-slice CT, images of the pancreatobiliary region have conventionally been acquired with collimation of at least 3 mm because of limitations in the breath-hold time. Based on the results of previous studies, thinner collimation is required to depict small pancreatic arteries [3, 4, 7]. Recent advances in helical CT technology, such as multislice detectors and subsecond rotation, have dramatically reduced scanning times and provide unparalleled capabilities for fast data acquisition and narrow collimation. Furthermore, the use of a multislice CT scanner with 16 channels or above has made it possible to scan the pancreas and biliary system even with submillimetre collimation during a single breath-hold of less than 10 s. Previous studies have reported the value of multiplanar reformatted images obtained with 0.5 mm or 1 mm collimation for the evaluation of pancreatic and bile duct anatomy and various duct abnormalities [8–10].

Multislice CT makes it possible to generate images of the optional thickness retrospectively from the same raw data. Reconstructing thicker axial images such as 1 mm images from raw data obtained with submillimetre collimation has the following advantages. First, the reduction in the number of images generated helps to reduce the huge workload on the CT system and on the workstation and the picture-archiving and communication systems. Second, noise in the axial images is reduced, which may help to improve the quality of CT pancreatic arteriographic images, especially in larger patients, although the resolution in the z-axis is slightly degraded. The present study was therefore conducted to assess the effects of the reconstructed thickness of axial images on image quality in CT pancreatic arteriography with 16-channel multislice CT in order to determine whether or not the quality of CT arteriography is improved by reconstructing axial images thinner than 1 mm from the volume data obtained with submillimetre collimation.

	Methods and materials

Top Abstract Introduction Methods and materials Results Discussion References

 
Patients
At our institution, three-phase contrast-enhanced CT examinations are performed using a multislice CT scanner in patients who are referred for the further evaluation of known or suspected diseases of the pancreas and biliary system, and who are judged to be candidates for surgical resection or intervention. The study group comprised 31 consecutive patients (19 men and 12 women; age range 29–80 years, mean age 64.5 years) who had undergone three-phase contrast-enhanced CT examination between 9 July and 18 September 2003. CT acquisition in the present study was performed in accordance with the established clinical standards of our institution, and each patient gave informed consent to undergo CT examination after the purpose, methods and risk were fully explained. Our institutional review board approved this study.

CT acquisition
In all patients, three-phase contrast-enhanced CT (consisting of the arterial, pancreatic and portal venous phases) was performed using a 16-channel multislice CT scanner (Aquilion, Toshiba Medical Systems, Tokyo, Japan). The scan parameters employed are summarized in Table 1. Non-ionic contrast material with an iodine concentration of 300 mg ml^–1 was injected at 0.08 ml kg^–1 body weight s^–1 (up to an upper limit of 5 ml s^–1) over 30 s, and a 5% dextrose flush was administered at a fixed rate of 5 ml s^–1 over 6 s immediately after the end of contrast material injection. Injection of both the contrast material and the 5% dextrose flush was performed using two automatic power injectors (Auto enhance A50 and Auto injector 1205, Nemotokyorindo, Tokyo, Japan) via a 20-gauge intravenous catheter placed in an antecubital vein.

Image processing and interpretation
The raw data of arterial-phase scanning in each patient were reconstructed in the following three ways with a field of view of 200–220 mm: 0.5 mm thickness (effective thickness of 0.75 mm) at 0.4 mm intervals in Group 1; 1 mm thickness at 0.5 mm intervals in Group 2; and 2 mm thickness at 1 mm intervals in Group 3. Specifically, an overlapping reconstruction interval of about 50% for an effective slice thickness was employed in all groups. In order to evaluate the image noise in each group, one radiologist (with 6 years of experience in abdominal image interpretation) measured the standard deviation (SD) of each CT number at a workstation (Alatoview, Toshiba) using a circular ROI 128 pixels in size placed in the aorta at the level of the origin of the coeliac artery. In this measurement, the circular ROI was placed at the same position in the aorta for the axial images of three types generated for the same patient.

These axial images were transferred to a workstation (ZIO M900, Ziosoft, Tokyo, Japan). The same radiologist as mentioned above generated CT arteriographic images using a three-dimensional volume-rendered technique from each axial dataset. In each patient, these images were optimized to visualize the arteries around the pancreas most clearly by interactive real-time control of two rendering parameters (low- and high-voxel values) on the angiographic opacity control curve, with the same parameters used in the same patient. The CT arteriographic images generated from the different axial datasets for the same patient were paired off as follows: Group 1 and Group 2, Group 1 and Group 3, and Group 2 and Group 3. For the interpretation of each paired set, images showing frontal and right anterior oblique views were printed out on a reading sheet (Figure 1). In this pairing, the left–right distribution of the images in the same group was equalized.

View larger version (76K): [in this window] [in a new window]   Figure 1. Example of reading sheet. The CT arteriographic images on the right and left sides are generated from axial images with different reconstruction parameters for the same patient. In this case, the axial datasets on the right and left sides are 0.5 mm thickness (effective thickness of 0.75 mm) at 0.4 mm intervals and 2 mm thickness at 1 mm intervals, respectively. The images in the upper and lower panels are frontal and right anterior oblique views, respectively. Each observer independently performs direct side-by-side comparison of these images and reports the difference in image quality between the right and left sides by marking his or her evaluation score with a pencil on a line with a length of 10 cm. Here, the rating scale is determined based on the following criteria: from –50 to –30, the image on the left side is superior; from –30 to –10, the image on the left side is slightly superior; from –10 to +10, the images on both sides are almost equivalent; from +10 to +30, the image on the right side is slightly superior; and from +30 to +50, the image on the right side is superior.

Statistical analysis
To determine the comparison evaluation scores among the three types of reconstruction methods, analysis of variance (ANOVA) for a two repeated-measures design was performed, with the scores obtained using the above-mentioned method as the dependent variable and the paired reconstruction methods as the within-subjects factor [11]. Here, if the null hypothesis in Mauchly's test of sphericity was rejected, adjustment of the degrees of freedom of the F-test statistics were performed using the Greenhouse–Geisser epsilon, the Huynh–Feldt epsilon and the lower-bound epsilon. SSPS 12 software (SPSS, Chicago, IL) was used for all statistical analyses, with a level of significance of 5% adopted for the statistical test, and, to compensate for the multiple comparisons problem, a level of significance of 1% adopted for ANOVA.

	Results

Top Abstract Introduction Methods and materials Results Discussion References

 
The results for the SD values of the CT numbers are summarized in Figure 2. The mean SD values of the CT numbers in the axial images in Group 1, Group 2 and Group 3 were 19.0, 16.9 and 13.2, respectively. In comparison with the axial images in Group 1, the SD values were reduced by approximately 7% and 28% in Group 2 and Group 3, respectively.

View larger version (12K): [in this window] [in a new window]   Figure 2. Graph of SD values of CT numbers. In comparison with axial images with a 0.5 mm thickness (effective thickness of 0.75 mm), the SD values are reduced by approximately 7% and 28% for axial images with a 1 mm thickness and a 2 mm thickness, respectively.

View larger version (18K): [in this window] [in a new window]   Figure 3. Comparison of CT pancreatic arteriographic images using a continuous rating scale: (a) major visceral arteries; (b) small pancreatic arteries. In all comparisons, CT pancreatic arteriographic images generated from thinner axial images are significantly superior. The difference between 0.5 mm thickness at 0.4 mm intervals and 1 mm thickness at 0.5 mm intervals is smaller than that between 0.5 mm thickness at 0.4 mm intervals and 2 mm thickness at 1 mm intervals and that between 1 mm thickness at 0.5 mm intervals and 2 mm thickness at 1 mm intervals. The most severe degradation is observed in small pancreatic arteries in CT arteriographic images generated from images with a 2 mm thickness at 1 mm intervals.

View larger version (96K): [in this window] [in a new window]   Figure 4. CT pancreatic arteriography in a 67-year-old woman with bile duct carcinoma. (a) Axial images with a 0.5 mm thickness at 0.4 mm intervals; (b) axial images with a 1 mm thickness at 0.5 mm intervals; (c) axial images with a 2 mm thickness at 1 mm intervals. The major visceral arteries are depicted in CT pancreatic arteriographic images generated from all axial datasets, but the sharpness of the vessel margins falls gradually from 0.5 mm thickness at 0.4 mm intervals (a) through 1 mm thickness at 0.5 mm intervals (b) to 2 mm thickness at 1 mm intervals (c). A similar stepwise degradation is more clearly observed in the depiction of the anterior pancreaticoduodenal arcade (arrows). Thick arrow (a) indicates endoscopic cholangiodrainage tube.

View larger version (77K): [in this window] [in a new window]   Figure 5. CT pancreatic arteriography in a 72-year-old man with cancer of the pancreatic head. (a) Axial images with a 0.5 mm thickness at 0.4 mm intervals; (b) axial images with a 1 mm thickness at 0.5 mm intervals; (c) axial images with a 2 mm thickness at 1 mm intervals. Arrows and curved arrow indicate posterior pancreaticoduodenal arcade and anterior pancreaticoduodenal arcade, respectively. The depiction of these small pancreatic arteries is severely degraded in CT arteriographic images generated from axial images with a 2 mm thickness at 1 mm intervals (c), while it is not easy to perceive any difference between 0.5 mm thickness at 0.4 mm intervals (a) and 1 mm thickness at 0.5 mm intervals (b). Thick arrow (a) indicates transhepatic cholangiodrainage tube.

	Discussion

Top Abstract Introduction Methods and materials Results Discussion References

The capabilities of CT angiography performed with a multislice CT scanner in depicting the visceral arterial anatomy have been widely discussed [12–15]. However, in all of these previous studies, CT angiographic images were generated from axial images with an effective thickness of more than 1 mm. Previous studies on CT coronary angiography have reported the value of the high spatial resolution scan protocol with submillimetre collimation in 16-channel multislice CT, but it was recommended that, for image reconstruction, sections not thinner than 1 mm should usually be reconstructed in order to achieve a good balance between temporal resolution and contrast resolution [18–20]. To our knowledge, there have been no studies that have assessed the value of scanning with submillimetre collimation in the generation of CT angiography for the visceral artery.

The primary goal of the present study was to determine whether or not reducing the effective thickness of axial images to less than 1 mm improves the quality of CT pancreatic arteriography. Therefore, in order to assess the effects of the thickness of axial images on image quality of CT pancreatic arteriography precisely, the present study was conducted in the following manner. To avoid any possible effects due to differences between individual subjects, images with different thicknesses were reconstructed from the same raw data as follows: 0.5 mm thickness (effective thickness of 0.75 mm) at 0.4 mm intervals; 1 mm thickness at 0.5 mm intervals; and 2 mm thickness at 1 mm intervals. In the generation of three-dimensional volume-rendered images, the parameter settings were fixed in the same patient. Four blinded readers independently performed direct side-by-side comparison for the CT arteriographic images generated from the different axial datasets for the same patient and were asked to evaluate which images were superior by assessment using a continuous rating scale.

The results of the present study proved that, in comparison among the reconstruction methods of 0.5 mm thickness at 0.4 mm intervals, 1 mm thickness at 0.5 mm intervals, and 2 mm thickness at 1 mm intervals using a continuous rating scale, CT arteriographic images generated from thinner axial images were found to be significantly superior for the visualization of both the major visceral arteries and, the small pancreatic arteries. However, regarding the major visceral arteries, the difference was small, and, in the assessment using a five-point scoring system, even the images generated from axial images with a 2 mm thickness at 1 mm intervals are assigned a grade of "almost equivalent" in 87 (70%) of 124 cases compared with those with a 0.5 mm thickness at 0.4 mm intervals. The results of the previous studies have also demonstrated that CT angiography performed with a multislice CT scanner permits the major visceral arteries to be identified in all cases [12–15]. These results suggest that, when CT arteriographic images are required only to provide information concerning the anatomy of the major visceral arteries, it is satisfactory to reconstruct axial images with a 2 mm thickness at 1 mm intervals.

On the other hand, compared with axial images with a 0.5 mm thickness at 0.4 mm intervals or a 1 mm thickness at 0.5 mm intervals, the use of images with a 2 mm thickness at 1 mm intervals was associated with the serious degradation in the quality of CT arteriographic images for small pancreatic arteries. Shioyama et al [12] have reported that pancreatic arteries less than 1.5 mm in diameter are infrequently visualized when CT angiographic images obtained with 2.5 mm collimation at 1.25 mm intervals are assessed in manual cine mode. Lee et al [14] have reported that CT angiography with images generated from axial datasets with 1.25 mm collimation at 0.62 mm intervals using volume rendering and maximum intensity projection techniques is limited in its ability to depict small hepatic arteries. These results indicate that, to depict small pancreatic arteries with greater certainty, it is essential to generate CT angiography from axial images with an effective thickness of 1 mm or less achieved by scanning with submillimetre collimation.

In the present study, although noise in the axial images with a 0.5 mm thickness at 0.4 mm intervals was increased by approximately 7% in comparison with those with a 1 mm thickness at 0.5 mm intervals, CT pancreatic arteriographic images generated from the former were judged to be significantly superior in the assessment using a continuous-rating scale. However, the difference in the quality of CT arteriographic images was small, and, in the assessment using a five-point scoring system, images were assigned a grade of "almost equivalent" in 109 (88%) and 91 (73%) of 124 cases for the major visceral arteries and for small pancreatic arteries, respectively. These results imply that it is not easy to perceive any difference between these two types of images in a substantial number of cases without direct side-by-side comparison as performed in the present study, and further research is needed to determine whether or not it is more useful in actual clinical practice to reconstruct axial images thinner than 1 mm for the generation of CT pancreatic arteriography.

A major limitation of the present study is that we did not assess the diagnostic value of CT pancreatic arteriographic images generated from different axial datasets in actual clinical use. However, for practical purposes, it is problematic to perform this assessment properly because the quality required in CT pancreatic arteriography differs from case to case according to the clinical objective in utilizing the images. For example, CT angiography is required to provide detailed information concerning the anatomy of small pancreatic arteries when limited surgery is planned based on a thorough understanding of the anatomy of the pancreas [21–23]. In such cases, it is preferable to employ axial images with the thinnest thickness available from the results of this study. In addition, further improvement in the quality of CT arteriographic images would be required by the development of new surgical procedures such as laparoscopy-assisted resection [23–25]. Further research is needed in order to determine the suitable application of reconstructing axial images with submillimetre thickness in consideration of these clinical situations.

Furthermore, a recently developed 64-channel multislice CT scanner makes it easier to perform scanning with submillimetre collimation. The number of images generated can be reduced by 20% using performing reconstruction with a 1 mm thickness at 0.5 mm intervals rather than a 0.5 mm thickness at 0.4 mm intervals. This is a great benefit in a 16-channel multislice CT scanner used in the present study because it takes approximately 3 min for this scanner to reconstruct 400 axial images for arterial-phase scanning alone. However, 64-channel multislice CT with faster reconstruction speed helps to overcome this restriction. Therefore, when determining the parameters for image acquisition and reconstruction in the generation of CT angiography for the small visceral artery, the results of this study that the thinner the effective slice thickness, even in axial images with a 1 mm thickness or less, the better the quality in CT arteriographic images should be noted.

In conclusion, image quality of CT pancreatic arteriography can be improved by reconstructing axial images with thinner thickness from the volume data obtained with submillimetre collimation using a 16-channel multislice CT scanner, and this improvement is useful for the depiction of small pancreatic arteries.

Received for publication November 9, 2006. Revision received March 30, 2007. Accepted for publication April 16, 2007.

	References

Top Abstract Introduction Methods and materials Results Discussion References

 

1. Conlon KC, Klimstra DS, Brennan MF. Long-term survival after curative resection for pancreatic ductal adenocarcinoma: clinicopathologic analysis of 5-year survivors. Ann Surg 1996;223:273–9.[CrossRef][Medline]
2. Yeo CJ, Cameron JL, Sohn TA, Lillemore KD, Pitt HA, Talamini MA, et al. Six hundred and fifty consecutive pancreaticoduodenectomies in the 1990s: Pathology, complications, outcomes. Ann Surg 1997;226:248–60.[CrossRef][Medline]
3. Blomley MJK, Albrecht T, Williamson RCN, Allison DJ. Three-dimensional spiral CT angiography in pancreatic surgical planning using non-tailored protocols: comparison with conventional angiography. Br J Radiol 1998;71:268–75.[Abstract]
4. Hong KC, Freeny PC. Pancreaticodudenal arcades and dorsal pancreatic artery: Comparison of CT angiography with three-dimensional volume rendering, maximum intensity projection, and shaded-surface display. AJR Am J Roentgenol 1999;172:925–31.[Abstract/Free Full Text]
5. Brink JA, Lim JT, Wang G, Heiken JP, Deyoe LA, Vannier MW. Technical optimization of spiral CT for depiction of renal artery stenosis: in vitro analysis. Radiology 1995;194:157–63.[Abstract/Free Full Text]
6. Kuszyk BS, Heath DG, Johnson PT, Eng J, Fishman EK. CT angiography with volume rendering for quantifying vascular stenoses: in vitro validation of accuracy. AJR Am J Roentgenol 1999;173:449–55.[Abstract/Free Full Text]
7. Chong M, Freeny PC, Schmiedl UP. Pancreatic arterial anatomy: depiction with dual-phase helical CT. Radiology 1998;208:537–42.[Abstract/Free Full Text]
8. Itoh S, Ikeda M, Ota T, Satake H, Takai K, Ishigaki T. Assessment of the pancreatic and intrapancreatic bile ducts using 0.5 mm collimation and multiplanar reformatted images in multislice CT. Eur Radiol 2003;13:277–85.[Medline]
9. Itoh S, Fukushima H, Takada A, Suzuki K, Satake H, Ishigaki T. Assessment of anomalous pancreaticobiliary ductal junction with high-resolution multiplanar reformatted images in MDCT. AJR Am J Roentgenol 2006;187:668–75.[Abstract/Free Full Text]
10. Fukushima H, Itoh S, Takada A, Mori Y, Suzuki K, Sawaki A, et al. Diagnosis value of curved multiplanar reformatted images in multislice CT for the detection of resectable pancreatic ductal adenocarcinoma. Eur Radiol 2006;16:1709–18.[CrossRef][Medline]
11. Tabachnick BG, Fidell LS. Computer-assisted research design and analysis. Boston, UK: Allyn and Bacon, 2001.
12. Shioyama Y, Kimura M, Horihata K, Masuda M, Hagihira T, Okumura T, et al. Peripancreatic arteries in thin-section multislice helical CT. Abdom Imaging 2001;26:234–42.[CrossRef][Medline]
13. Takahashi S, Murakami T, Takamura M, Kim T, Hori M, Narumi Y, et al. Multi-detector row helical CT angiography of hepatic vessels: Depiction with dual-arterial phase acquisition during single breath hold. Radiology 2002;222:81–8.[Abstract/Free Full Text]
14. Lee SS, Kim TK, Byun JH, Ha HK, Kim PN, Kim AY, et al. Hepatic arteries in potential donors for living related liver transplantation: evaluation with multi-detector row CT angiography. Radiology 2003;227:391–9.[Abstract/Free Full Text]
15. Matsuki M, Tanikake M, Kani H, Tatsugami F, Kanazawa S, Kanamoto T, et al. Dual-phase 3D CT angiography during a single breath-hold using 16-MDCT: Assessment of vascular anatomy before laparoscopic gastrectomy. AJR Am J Roentgenol 2006;186:1079–85.[Abstract/Free Full Text]
16. Itoh S, Suzuki K, Iwano S, Satake H, Ota T, Ikeda M, et al. Three-phase CT examination of the pancreatobiliary region using multislice CT with 1 mm collimation. Rad Med 2005;23:283–91.
17. Goshima S, Kanematsu M, Kondo H, Yokoyama R, Miyoshi T, Kato H, et al. Optimal scan delay for contrast-enhanced multi-detector row CT. Radiology 2006;241:167–74.[Abstract/Free Full Text]
18. Kopp AF, Kuttner A, Heuschmid M, Schroder S, Ohnesorge B, Claussen CD. Multidetector-row CT cardiac imaging with 4 and 16 slices for coronary CTA and imaging of atherosclerotic plaques. Eur Radiol 2002;12:S17–S24.[CrossRef][Medline]
19. Schoepf UJ, Becker CR, Ohnesorge BM, Yucel EK. CT of coronary artery disease. Radiology 2004;232:18–37.[Abstract/Free Full Text]
20. Kopp AF, Kuttner A, Trabold T, Heuschmid M, Schroder S, Claussen CD. Multislice CT in cardiac and coronary angiography. Br J Radiol 2004;77:S87–S97.[Abstract/Free Full Text]
21. Takada T, Yasuda H, Uchiyama K, Hasegawa H, Iwagaki T, Yamakawa Y. A proposed new pancreatic classification system according to segments: operative procedure for a medial pancreatic segmentectomy. J Hep Bil Pancr Surg 1994;1:322–5.[CrossRef]
22. Kimura W, Nagai H. Study of surgical anatomy for duodenum-preserving resection of the head of the pancreas. Ann Surg 1995;221:359–63.[Medline]
23. Orsenigo E, Baccari P, Bissolotti G, Staudacher C. Laparoscopic central pancreatectomy. Am J Surg 2006;191:549–52.[CrossRef][Medline]
24. Jaroszewski DE, Schlinkert RT, Thompson GB, Schlinkert DK. Laparoscopic localization and resection of insulinomas. Arch Surg 2004;139:270–4.[Abstract/Free Full Text]
25. Uranues S, Alimoglu O, Todoric B, Toprak N, Auer T, Rondon L, et al. Laparoscopic resection of the pancreatic tail with splenic preservation. Am J Surg 2006;192:257–61.[CrossRef][Medline]

This Article Abstract Figures Only Full Text (PDF) Services 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 Google Scholar Google Scholar Articles by ISHIGAKI, S Articles by ISHIGAKI, T Search for Related Content PubMed PubMed Citation Articles by ISHIGAKI, S Articles by ISHIGAKI, T