Clinical feasibility of noncontrast-enhanced magnetic resonance lymphography of the thoracic duct
Hiroyuki Takahashi
Study objective: The dilatation of the thoracic duct was previously demonstrated in liver cirrhosis by lymphangiography, endoscopic ultrasound, and at autopsy. The evaluation of the morphologic change of the thoracic duct may be important in assessing the altered lymphodymanics in liver cirrhosis. The objectives of this study were to determine which combination of posture and breathing phase during noncontrast-enhanced magnetic resonance lymphography (MRL) provided the clearest images, and to evaluate the morphologic changes in the thoracic duct ill healthy volunteers and patients with liver disease and malignancy.
Design: Prospective study. Setting: Community general hospital.
Design and subjects: Twenty-three healthy volunteers and 113 patients underwent the MRL examination using a three-dimensional, half-Fourier, fast spin echo sequence on a 1.5-T, whole-body magnetic resonance system. The appropriate posture and breathing phase of MRL to obtain the best visualization was first determined by trial on 14 healthy volunteers. Morphologic changes of the thoracic ducts were evaluated in 23 healthy volunteers including the 14 healthy volunteers for the first trial and 113 patients using this appropriate method. The width of the thoracic ducts in both patients and volunteers was measured.
Measurements and results: MRL with respiratory gating in the supine position depicted the thoracic duct well and was the most comfortable for the subjects. In 82 of 113 patients (72.6%), the thoracic ducts were entirely visualized from the diaphragm level to the subclavian region. The remaining 31 patients had ducts that could not be entirely visualized due to sections or short lengths that were obscured. The maximum diameter was 3.74 [+ or -] 0.81 mm in all healthy volunteers, 6.98 [+ or -] 2.77 nun in alcoholic cirrhosis, 4.12 [+ or -] 1.51 mm in nonalcoholic cirrhosis, 3.76 [+ or -] 1.10 mm in malignancy, and 3.60 -+ 0.80 mm in chronic hepatitis (mean [+ or -] SD). The diameter in alcoholic cirrhosis was significantly greater than in other groups (p < 0.01).
Conclusions: Respiratory gating in the supine position is the best MRL method for acquiring the clearest images. This may be a good method of detecting morphologic changes in the thoracic duct. The patients with alcoholic cirrhosis showed a greater thoracic duct diameter than other groups.
Key words: liver cirrhosis: lymphography: MRI: thoracic duct
Abbreviations: FSE = fast spin echo; MIP = maximum intensity projection; MRL = magnetic resonance lymphography; 3D = three dimensional
**********
The lymphatic system is an important circulatory system for the maintenance of” an organism and its immunity. It contributes to a rerun1 of excess liquid and protein from the interstitial tissue to the circulatory system and to protection against bacteria. (1) Until recently, the only methods available to examine the lymphatic system have been either by radiograph lymphography with contrast material or lymphography with radionuclides. Radiograph lymphography is an invasive procedure with drawbacks such as the use of contrast material and a long examination time. (2) Lymphography with radionuclides lacks morphologic detail. More recently, endoscopic ultrasound was developed to observe the thoracic duet, but this method is semi-invasive and is unable to evaluate the entire anatomy. (3) A new method of visualizing the vessel by using magnetic resonance lymphography (MRL) with three-dimensional (3D), short echo-spacing, half-Fourier, fast spin echo (FSE) sequences was reported by Hayashi and Miyazaki. (4) The thoracic duet, which is the widest lymph duet in the human body, can be visualized with this technique. However, to our knowledge, neither a study examining the appropriate MRL method nor a study dealing with the feasibility of its clinical usefulness have been reported. In the present study, we set out to determine which combination of posture and breathing phase during MRL of the thoracic duct provide the clearest images, and evaluated the morphologic changes in the thoracic duet in healthy volunteers and patients with liver disease and malignancy.
SUBJECTS AND METHODS
study Populations
Twenty three healthy volunteers (18 men and 5 women; mean age, 35.7 years; age range, 26 to 48 years) and 113 patients (81 men and 32 women; mean age, 65.1 years; age range, 34 to 85 years) with liver disease and malignancy (Table 1) underwent “MILL during the period between June 1999 and September 2002. Informed consent was obtained from each subject. The MRL for the patients was performed at the same time as the magnetic resonance examination of the liver and other organs. The total acquisition time for MRL was approximately 5 min, and the average study time for MRL was i0 rain.
MRI
In the first part of this study, the appropriate posture and breathing phase to obtain the best visualization of the thoracic duct was determined by trial on 14 healthy volunteers. They were examined with four different postures and breathing methods: (1) intermittent breath-holding after expiration in the supine position, (2) intermittent breath-holding after inspiration in the supine position, (3) intermittent breath-holding after expiration in the prone position, and (4) respiratory gating in the supine position. An intermittent breath-holding was applied fur each slice encoding. Each slice encoding required a breath-holding of approximately 1 s, and normal breathing was permitted during the remainder of the repetition time. This was repeated for the number of slice encodings in the 3D acquisition, After acquisition, maximum intensity projection (MIP) processing was performed. Respiratory gating parameters were adjusted for each patient individually, including respiratory cycle length, gate delay, and gate width. Prospective respiratory gating was used using the sensor that detected the motion of the body surface around the chest wall. The actual single shot acquisition time was 880 ms. A 1.5-T magnetic resonance imager (VISART EX; Toshiba; Tokyo, Japan) with a quadrature detection phased-array coil was used with a 3D, half-Fourier, FSE sequence. Coronal in-plane acquisition was performed with a resolution of 1.1 x 1.1 ram. The sequences were as follows: repetition time in milliseconds/echo times in milliseconds, 3,000 to 6,000/500 (respiratory gating), 7,000/500 (intermittent breath holding); echo spacing, 12.5 ms; matrix size, 320 x 320; 30 slice partitions with a 2,0-mm slice thickness; and field of view, 360 x 360 mm.
Image Analysis
For the observation of the MIP images, the thoracic duct was divided into three parts: (1) upper part, from the subclavian region to the level of left main bronchus; (2) middle part, proximal half of the distance from the left main bronchus to the diaphragm; and (3) lower part, distal half of the duct. The thoracic ducts observed with the four methods were compared on the basis of the degree of visualization of the three parts. The degree of visualization was as follows: poor (the anatomic part was not or partially visible), good (the anatomic part was mostly visible), and excellent (the entire anatomic part was clearly’ visible). The widest portion of the thoracic duets was measured on doe MIP images after magnifying on the display, and the lateral views, slight left anterior and right anterior oblique paracoronal views, and original images were used for reference, and then statistically analyzed. At the end of the MRL examination, each volunteer was asked to answer which of the four methods was the most comfortable. All images were evaluated by two radiologists, and a final decision was reached by consensus. MRL readers were blinded to each patient’s diagnosis. After we decided on the most appropriate technique for our 14 healthy volunteers, the other 9 healthy volunteers and all patients were examined using this same technique.
As the second part of this study, morphologic change of the thoracic duet was evaluated in 23 healthy volunteers including the 14 healthy volunteers used in the first trial and 113 patients using the appropriate method. We assessed the visualization of the main thracic duet for each part, the existence of tortuosity, the visualization of drainage into the subclavian region, and the cisterna chyli. We classified the patients into four groups: alcoholic liver cirrhosis (group 1; n = 12), nonalcoholic liver cirrhosis (group 2; n = 45), malignancy (group 3; n = 43), and chronic hepatitis (group 4; n – 13). The diagnosis of malignancy included gastric cancer (n = 24), colon cancer (n = 4), lung cancer (n = 3), gallbladder and common bile duct cancer (n = 3), pancreas cancer (n = 3), renal cancer (n = 2), esophageal cancer (n 1), gastric malignant lymphoma (n = 1), prostatic cancer (n = 1), and breast cancer (n 1). The widest portion of the thoracic duets in both patients and volunteers was measured, and then statistically analyzed. Serum albumin and sodium are factors that affect hepatic lymph production, and an increased lymph production causes expansion of the thoracic duct. We examined the relation between the diameter of the thoracic duct and the serum albumin and sodium in 10 patients with alcoholic liver cirrhosis, 24 patients with nonalcoholic liver cirrhosis, 14 patients with malignancy, and 6 patients with chronic hepatitis. These tests were performed within 1 week of their MRL examinations.
Statistical Analysis
Values are expressed as means [+ or -] SD. Differences in the diameter were evaluated using the Kruskal-Wallis test. Directed paired comparisons of individual groups were conducted using the Mann-Whitney U test. Regression analysis was used to evaluate the correlation between the diameter of tim tlmraeic duct and the serum albumin and sodium levels. The software used was StatView (Version .5.0; SAS Institute; Gray, NC); p < 0.05 was considered statistically significant.
RESULTS
Figure 1 shows the MRL of the thoracic duct of a healthy volunteer in the four different postures and breathing techniques. Figure 2 shows the mean maximum diameter of the thoracic duct in the four methods. Table 2 summarizes the degree of visualization in each case. The mean maximum diameter of the thoracic duct was 3.46 [+ or -] 0.56 mm in respiratory gating, 3.99 [+ or -] 0.82 nun in intermittent breath-holding after expiration in the supine position, 3.57 [+ or -] 0.77 mm in intermittent breath-holding after inspiration in the supine position, and 3.77 [+ or -] 0.60 mm in intermittent breath-holding after expiration in the prone position. The diameter of the thoracic duet in intermittent breath-holding after expiration in the supine position was significantly greater than in respiratory gating and intermittent breath-holding after inspiration in the supine position (p < 0.05). In the upper and lower parts of the thoracic duet, there was little difference in the visualization among the four methods. The middle part, however, was most clearly visualized when using the respiratory gating in the supine position. All but one volunteer reported respiratory gating was the most comfortable. The remaining volunteer reported both respiratory gating and the intermittent breath-holding method in the supine position were comfortable. We thus regarded respiratory gating in the supine position as the appropriate method for MRL of the thoracic duct.
[FIGURE 1 OMITTED]
The examination with respiratory gating in the supine position was employed on the remaining nine healthy volunteers and 113 patients. The maximum diameter of the duct ranged from 2.62 to 5.57 mm (mean, 3.74 [+ or -] 0.81 mm) in all 23 volunteers. In 82 of 113 patients (72.6%), the thoracic ducts were entirely visualized in all three parts. The remaining 31 patients 1,ad ducts that could not be entirely visualized due to sections or short lengths that were obscured. In 19 of these patients, some part of the duet was obscured due to heart compressions. In 12 of 31 patients, the thoracic duct either above or below the heart could not be seen in its entirety possibly as a result of pleural change, compression by organs, osteophytes, hiatal hernias, etc. The thoracic ducts were divided into two separate channels in 10 eases (8.85%). Sixty-five eases (57.5%) showed adequate drainage into the subclavian region. Sixty-two patients (54.9%) demonstrated a cisterna chyli or the duets origin at the upper lumbar level. Figure 3 shows the statistical results of healthy volunteers and the four patient groups. The maximum diameters of the thoracic duet ranged from 1.44 to 10.34 mm (mean, 4.23 [+ or -] 1.76 ram) in all patients; from 3.26 to 10.34 mm (mean, 6.98 [+ or -] 2.77 mm) in group 1; from 1.44 to 9.19 mm (mean, 4.12 [+ or -] 1.51 mm) in group 2; from 2.30 to 7.47 mm (mean, 3.76 [+ or -] 1.10 mm) in group 3; and from 2.53 to 5.17 mm (mean, 3.60 [+ or -] 0.80 ram) in group 4. In 7 of 12 patients with alcoholic liver cirrhosis, in 4 of 45 patients with nonalcoholic liver cirrhosis, and in 2 of 43 patients with malignancy, the thoracic duct diameters were > 6 ram. None of the patients with chronic hepatitis nor the healthy volunteers had such wide thoracic ducts. The difference in maximum diameter among the four patient groups and healthy volunteers was significant (p < 0.01). The diameter of group 1 was significantly wider than that of healthy volunteers, and of groups 2, 3, and 4 (p 0.1). Figure 4 shows the MRL of the thoracic duct in a 54-year-old man with alcoholic liver cirrhosis, and Figure 5 is from a 66-year-old man with nonalcoholic liver cirrhosis. Tortuous ducts were observed in 22 of 57 patients (38.6%) with liver cirrhosis. The mean serum albumin level was 3.51 [+ or -] 0.63 g/dL (mean interval, 1.90 days) in 10 patients of group 1, 3.23 [+ or -] 0.43 g/dL (mean interval, 2.53 days) in 24 patients of group 2, 3.58 [+ or -] 0.52 g/dL (mean interval, 2.21 days) in 14 patients of group 3, and 3.82 [+ or -] 0.65 g/dL (mean interval, 0.67 days) in 6 patients of group 4. The difference among the four groups was not significant (p > 0.1). The mean serum sodium levels of 10 patients of group 1, 24 patients of group 2, 14 patients of group 3, and 6 patients of group 4 were 137.90 [+ or -] 1.91 mEq/L, 137.00 [+ or -] 4.85 mEq/L, 138.29 [+ or -] 3.93 mEq/L, and 140.00 [+ or -] 4.78 mEq/L, respectively. The difference in serum sodium level among the four groups was not significant (p > 0.1). There was no correlation between the diameter of the thoracic duet and the serum albumin level or serum sodium level (p > 0.05).
[FIGURES 4&5 OMITTED]
DISCUSSION
The lymphatic system is an independent network for the circulation of fluid throughout the body, and it contributes to the return of excess liquid and protein from the interstitial tissue to the circulation system. (1) The thoracic duct consists of a relatively small vessel that shows slower flow than that of the cardiovascular system. Since MRL is magnetic resonance hydrography using a long repetition time and long effective, echo time, this technique is particularly suitable for the delineation of the thoracic duct. In the present study, we evaluated the appropriate method of MRL for the thoracic duet and its clinical application to disease.
Hayashi and Mivazaki (1) reported that the thoraic duct could be visualized in six healthy volunteers at MRL with a short spacing, 3D, half-Fourier, FSE sequence and intermittent breath-hold. However, no study examining the appropriate posture and breathing type of MRL obtaining best visualization has been reported. Since in this study we did not compare MRL with another “gold standard,” it cannot be said that this is a superior method of assessing the thoracic duct; however, we found that respiratory gating in the supine position was the best method for MRL visualization of the thoracic duet while taking into consideration the physical comfort of the patient. Pomerantz et al (5) reported that the eisterna chili and left supraclavieular lymph nodes could be observed in 35 of 98 patients (35.7%) and in 56 of 115 patients (48.7%) in radiograph lymphangiography with the use of a contrast agent, respectively. In the present study, we could image the cisterna chyli and/or their origin at the upper lumbar level in 62 of 113 cases (54.9%) and the entry of the duet into the subclavian region in 63 cases (55.8%). Divided thoracic channels were observed in 10 of 113 patients (8.9%), and divided channel was seen ill one of 23 volunteers (4.4%). Rosenberger and Abrams (6) reported that multiple thoracic channels were observed in 30 of 390 subjects (7.7%). The frequency of visualizing the cisterna chyli, the entry into the subclavian region, and the divided channel in the present study, were similar to those reported in radiograph lymphagiography.
The maximum diameters ranged from 2.62 to 5.57 mm in healthy volunteers and 1.44 to 10.34 mm in patients. Pomerantz et al (5) reported the diameter at the end of the thoracic duet varied between 0.5 mm and IS iron in 11,5 eases in radiograph lymphangiography, and other studies (6-8) documented the diameter ranging from 1 to 8 ram. The diameter of the thoracic duct in the present study concurred with these previous values, thus supporting the premise that the method in the present study appeared to measure the true diameter of the thoracic duct.
Thoracic duct dilatation has been demonstrated in portal hypertension and liver cirrhosis by lymphangiography, endoscopic ultrasound and at autopsy. (3, 7, 9, 10) It was reported that enlargement of the thoracic duct in liver cirrhosis is due to an increased production of liver cirrhosis is due to an increased production of liver lymph. Thoracic duct pressure is related to the amount of duct dilatation, and cannulation of the thoracic duct with decompression produces changes similar to a portosystemic shunt. (10) The production of liver lymph depends on the resultant gradient of hydrostatic and oncotic pressure differences between blood and tissues. (11, 12) Portal pressure and serum sodium are factors involved in hydrostatic pressure, and serum albumin is a factor of oncotic pressure. According to Witte et al, (12, 13) intrahepatic portal pressure becomes elevated and the transmural oncotic gradient in the liver increases in patients with liver cirrhosis despite hypoproteinemia. In this study, there was no correlation between the diameter of the thoracic duct and the serum albumin or sodium level. However, the meaning of this result is unclear since this measurement was performed only on a small number of patients without control of hydration. Portal hypertension, which results from the obstruction of hepatic venules, is the most significant driving force responsible for the outpouring of hepatic lymph in liver cirrhosis. (7, 10, 12, 13) Vidins et at (14) reported that patients with alcoholic liver disease showed a marked reduction in sinusoidal areas of their livers relative to those of patients with nonalcoholic liver disease, and that portal pressure increases inversely with the sinusoidal area when that area decreases below 20% of normal. In the present study, significant thoracic duet dilatation was only recognized in patients with alcoholic liver cirrhosis, and no correlation was observed between the diameter of the thoracic duct and serum albumins/sodium level. This may result from the difference in the hepatic architecture between alcoholic and nonalcoholic liver cirrhosis. However, we have no direct evidence to support this in the present study since no measurement of portal pressure was performed.
Though overlapping of thoracic duct sizes among patients groups occurred, all of the ducts of hepatitis patients and healthy volunteers fell < 6 mm. We calculated the sensitivity, specificity, positive predictive value, and negative, predictive value for each thoracic duet diameter to diagnose the presence of alcoholic liver cirrhosis (Table 3). A cut-off point of 8 mm seems to be the best diagnostic guide for the detection of alcoholic liver cirrhosis. When the criterion of 8 mm was used, sensitivity, specificity, positive predictive value, and negative predictive value were 50.0%, 98.0%, 75.0%, and 94.3%. respectively.
In conclusion, noncontrast-enhanced MRL is a safe and noninvasive method for imaging the thoracic duct. Respiratory gating in the supine position seems to be the appropriate method for MRL to obtain the best visualization. MRL may be a good method of detecting morphologic changes in the thoracic duct. MRL may be useful for a preoperative evaluation to avoid thoracic duct injury, diagnosis of chylothorax, and evaluation of lymph production.
Table 1–Characteristics of Patients *
Disease Cases Men Women Age, yr
Liver cirrhosis 57 43 14 62.3 (43-45)
Alcoholic 12 12 0 56.3 (44-69)
Nonalcoholic 45 31 14 64.9 (43-85)
Chronic hepatitis 13 8 5 57.9 (34-85)
Malignancy 43 30 13 69.9 (39-85)
Total 113 81 34 65.1 (34-85)
* Data are presented as No. or mean (range).
Table 2–Degree of Visualization of Each Anatomic Part of Thoracic
Duct According to Method
MRL Method *
Degree of
Visualization A B C D
Upper
Excellent 12 11 8 10
Good 2 2 3 2
Poor 0 1 3 2
Middle
Excellent 6 3 2 3
Good 6 7 3 5
Poor 2 4 9 6
Lower
Excellent 14 13 11 11
Good 0 0 2 1
Poor 0 1 1 2
* A = respiratory gating in the supine position; B = intermittent
breath-holding after expiration in the supine position;
C = intermittent breath-holding method after inspiration in the
supine position; and D = intermittent breath-holding after expira-
tion in the prone position.
Table 3–Sensitivity, Specificity, Positive Predictive
Value, and Negative Predictive Value for Thoracic
Duct Diameter in Diagnosing the Presence of Alcoholic
Liver Cirrhosis
Thoracic Duct Sensitivity, Specificity
Diameter, mm % %
5.0 66.7 83.2
6.0 58.3 93.1
7.0 50.0 97.0
8.0 50.0 98.0
9.0 33.3 99.0
Positive Negative
Thoracic Duct Predictive Predictive
Diameter, mm Value, % Value, %
5.0 32.0 95.5
6.0 50.0 94.9
7.0 66.7 94.2
8.0 75.0 94.3
9.0 80.0 91.7
Respiratory gating 3.46 [+ or -] 0.56
After expiration 3.99 [+ or -] 0.82
After inspiration 3.57 [+ or -] 0.77
In the prone position 3.77 [+ or -] 0.60
FIGURE 2. Bar chart of the maximum diameters of the thoracic
duet in the four different posture and breathing techniques. The
diameter in intermittent breath-holding after expiration in the
supine position was significantly greater than in respiratory gating
and intermittent breath-holding alter inspiration in the supine
position (p < 0.05). Respiratory gating = respiratory gating
in the supine position. After expiration = intermittent breath-holding
after expiration in the supine position. After inspiration =
intermittent breath-holding after inspiration in the supine position.
In the prone position = intermittent breath-holding after expiration in
the prone position.
Volunteers 3.74 [+ or -] 0.81
Group 1 6.98 [+ or -] 2.77
Group 2 4.12 [+ or -] 1.51
Group 3 3.76 [+ or -] 1.10
Group 4 3.6 [+ or -] 0.80
FIGURE 3. Bar chart of the maximum diameters of the thoracic
duct in healthy volunteers and four patient groups. The difference
in the maximum diameter among healthy volunteers and
four groups was significant (p < 0.01). The diameter of group 1
was significantly wider than those of healthy volunteers, group 2,
group 3, and group 4 (p < 0.0l). Volunteers = all healthy volunteers
(n = 23); group 1 = alcoholic liver cirrhosis (n = 12); group
2 = nonalcoholic liver cirrhosis (n = 45); group 3 = malignancy
(n = 43); group 4 = chronic hepatitis (n = 13).
Acknowledgment: We thank Kouji Nagata magnetic resonance engineer, for assistance.
REFERENCES
(1) Gyton AC, Hall JE. Textbook of medical physiology, 9th ed. Pennsylvania, PA: W. B. Saunders Company [Japanese edition: Tokyo, Japan; Igaku-shoin Ltd.], 1999:197-201
(2) Kinmonth JB, Taylor GW, Harper RK. Lymphangiography: a technique for its clinical use in the lower limb. BMJ 1955; 1:940-942
(3) Parasher VK, Meroni E, Malesci A, et al. Observation of thoracic duct morphology in portal hypertension by endoscopic ultrasound. Gastrointest Endosc 1998:48:588-592
(4) Hayashi S. Miyazaki M. Thoracic duct; visualization at non-enhanced MR lymphography-initial experience. Radiology 1999; 212:598-600
(5) Pomerantz M, Herdt JRL, Ruckoff SD, et al. Evaluation of functional anatomy of thoracic duct by lymphangiography. J Thorac Cardiovasc Surg 1963; 46:568-575
(6) Rosenberger A, Abrams HL, Radiology of the thoracic duct. AJR Am J Roentgenol 1971; 111:807-820
(7) Shieber W, Lymphangiographic demonstration of thoracic duct dilatation in portal cirrhosis. Surgery 1965; 57:522-524
(8) Nusbaum M, Baum S, Hedges RC, et al. Roentgenographic and direct visualization of thoracic duet. Arch Surg 1964; 88:105-113
(9) Dumont AE, Mulholland JH. Flow rate and composition of thoracic duct lymph in patient with cirrhosis. N Engl J Med 1960; 263:471-474
(10) Dumont AE, Mulholland JH. Alteration in thoracic duct lymph flow in hepatic cirrhosis: significance in portal hypertension. Ann Surg 1962; 156:668-677
(11) Witte MH, Witte CL, Dumont AF. Estimated net transcapillary water and protein flux in the liver and intestines of patients with portal hypertension from hepatic cirrhosis. Gastroenterology 1981; 80:265-272
(12) Witte CL, Witte MH, Dumont AE. The portal triad in hepatic cirrhosis. Surg Gynecol Obstet 1978; 146:965-974
(13) Witte MH, Dumont AE, Cole WR, et al. Lymph circulation in hepatic cirrhosis: effect of portacaval shunt. Ann Intern Med 1969; 70:303-310
(14) Vidins EI, Britton RS, Medline A, et al. Sinusoidal caliber in alcoholic and nonalcoholic liver disease: diagnostic and pathogenic implications, Hepatology 1985; 5:408-414
* From the Department of Radiology (Dr Takahashi), Japan Seamen’s Relief Association Moji Hospital, Kitakyushu-shi; Department of Internal Medicine (Drs. Kuboyama and Abe), Fukuoka Prefecture Asakura Hospital Hepato-Gastroenterology Center Fukuoka-ken Department of Radiology (Drs. Aoki and Nakata), University of Occupational and Environmental Health School of Medicine. Kitakyushu-shi; and MR Engineering Department (Dr. Miyazaki). Medical System R&D Center, Toshiba Corporation. Tochiji-ken, Japan.
This work was performed at Fukuoka Prefecture Asakura Hospital Hepato-Gastroenterology Center.
Manuscript received November 5, 2002; revision accepted July 22, 2003.
Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (e-mail: permissions@chestnet.org).
Correspondence to: Hiroyuki Takahashi, MD, Department of Radiology, Japan Seamen’s Relief Association Moji Hospital, 1-3-3 1 Kiyotaki Moji-ku, Kitakyushu-shi, 801-8550 Japan; e-mail: cbn15560@pop16.odn.ne.jp
COPYRIGHT 2003 American College of Chest Physicians
COPYRIGHT 2004 Gale Group
