J Appl Physiol 105: 958-963, 2008.
First published July 3, 2008; doi:10.1152/japplphysiol.01204.2007
8750-7587/08 $8.00
MRI analysis of structural changes in skeletal muscles and surrounding tissues following long-term walking exercise with training equipment
Ryusuke Nakai,1
Takashi Azuma,1
Mai Sudo,2
Shin-ichi Urayama,3
Osamu Takizawa,4 and
Sadami Tsutsumi5
1Department of Medical Simulation Engineering, Research Center for Nano Medical Engineering, Institute for Frontier Medical Sciences, Kyoto University, Kyoto; 2Human Science Research Center, Wacoal, Kyoto; 3Human Brain Research Center, Kyoto University Graduate School of Medicine, Kyoto; 4Medical Solutions Marketing Division, Siemens Asahi Medical Technologies, Tokyo; and 5Nihon University School of Dentistry, Tokyo, Japan
Submitted 12 November 2007
; accepted in final form 26 June 2008
 |
ABSTRACT
|
|---|
Muscular recovery after exercise is an important topic in sports medicine, and accurate and quantitative measurements of changes in muscle are required to assess muscular recovery. In the present study, we report a new analytical method to measure muscular changes quantitatively. The technique consists of three independent methods: image processing of two-dimensional MR images, morphological analysis using three-dimensional MR images, and diffusion tensor MRI. Using this method, we investigated changes in the quadriceps and biceps femoris and gluteus maximus muscles and surrounding tissues before and after 1 mo of exercise wearing training equipment. The subjects were 21 healthy adult female volunteers, 14 of whom wore training equipment and 7 who wore normal equipment. The percentage of adipose tissue in muscle after exercise in subjects who wore training equipment was on average 4.4% (P < 0.001) lower than that before exercise, and the peak point of the dorsal hip after exercise with use of the equipment was on average 10.8 mm higher than that before exercise. Further, the fractional anisotropy of water diffusion in muscles increased by an average of 0.039 (P < 0.001) after exercise with use of training equipment. In contrast, there was no significant difference before and after exercise in subjects who wore normal equipment. These results show that walking exercise while wearing training equipment thickens and tightens the muscular fiber tissues. This noninvasive measurement approach may allow quantitation of the athletic ability of the muscles, which is not measured conventionally, and is an effective method for analyzing skeletal muscles.
magnetic resonance imaging; image analysis
MUSCLES are broadly classified into skeletal muscle, smooth muscle, and myocardium. There are more than 400 skeletal muscles of different sizes, and these comprise 
40% of body weight. Skeletal muscles are referred to as voluntary muscles, since they are innervated by motor neurons and cannot contract without nerve stimulation. These muscles are attached to the skeleton and control the joints by contraction; therefore, they play a very important role in joint motion (4, 7, 17).
Skeletal muscle and joint motion are related to muscle strength, which, in principle, increases proportionally with the cross-sectional area of muscle fibers. Muscular strength is determined by the muscle fiber cross-sectional area, the number of muscle fibers, and the muscle fiber type. Muscular strength increases because of an increase in muscle mass caused by an increase in the muscle fiber cross-sectional area. Muscle mass and shape both change with use frequency; for example, muscle fibers are thickened by training with appropriate external pressure, whereas muscle mass is decreased with reduced usage and muscles are thinned (2, 3, 5, 12). Many studies have examined the physiological mechanism of muscle hypertrophy and atrophy, and these processes have gradually been elucidated at the genetic level, including establishment of the role of growth factors and hormones (1, 9, 11, 18). Such molecular analysis is important, but simple noninvasive quantitative measurement of changes in muscle is required to diagnose the clinical effects of exercise and the progress of diseases.
Conventionally, measurement of muscular volume has mainly been performed for assessment of muscular change. However, volume measurement does not allow evaluation of the cause of the change, and therefore we have developed a new analytical method to measure muscular changes quantitatively. The technique includes three independent methods: image processing of two-dimensional MR images (M1), morphological analysis using three-dimensional MR images (M2), and diffusion tensor MRI (M3), which permits imaging of the microstructure. In the present study, changes in stressed quadriceps and biceps femoris and gluteus maximus muscles and surrounding tissues were measured using this method.
|
MATERIALS AND METHODS
|
|---|
The subjects were 21 healthy adult female volunteers who gave written informed consent after receiving a full explanation of the study. We have received approval of this experiment from our Review Committee. Fourteen subjects wore training equipment designed to exert appropriate stress on the quadriceps and biceps femoris and gluteus maximus muscles (Bottom Clothes, Hip Walker, Wacoal), and 7 subjects wore normal equipment. All subjects took walking exercise of more than 10,000 steps/day in normal life. The number of steps was measured using a pedometer. The subjects underwent MRI (Magnetom Sonata 1.5T, Siemens; maximum amplitude 40 mT/m, slew rate 200 mT·m–1·ms–1) with coils (phased array coils) in a prone position before and after 1 mo of exercise wearing training or normal equipment.
M1: gray scale analysis.
MR images were taken in the sagittal plane using a Turbo Spin Echo sequence [repetition time (TR), 910 ms; echo time (TE), 10 ms; flip angle, 150°; bandwidth, 205 Hz/pixel; slice thickness, 3.5 mm; matrix size, 288 x 512; field of view (FOV), 202.5 x 360 mm; pixel size, 0.7x0.7 mm; average, 3; echo train length, 3; scan time, 2 min 14 s; 25 slices; slice distance factor, 10%] and fat-suppressed Turbo Spin Echo sequence (TR, 1,410 ms; TE, 9.9 ms; flip angle, 150°; bandwidth, 205 Hz/pixel; slice thickness, 3.5 mm; matrix size, 288 x 512; FOV, 202.5 x 360 mm; pixel size, 0.7 x 0.7 mm; average, 3; echo train length, 3; scan time, 4 min 34 s; 25 slices; slice distance factor, 10%). The right femur condyle was defined as the center of the imaging region. Linear hyperintensity objects (adipose tissue) were extracted from the two types of acquired images, using the following image processing procedures (21). First, MR images (discrete images) were smoothed using a Gaussian filter to produce continuous images. Discrete images are expressed as f(xi,yi), and continuous images as 
(x,y) in the following equation. Parameter 
is 1.5.
 | (1) |
G(s) is shown by the following equation.
 | (2) |
In the next step, linear hyperintensity object (adipose tissue) images were extracted from processed continuous images using a method to extract pixels showing hyperintensity in muscle. The two-dimensional image f(x,y) is shown as a Taylor expansion near the point (x,y) by Eq. 3, where 
f(x,y) is the first derivative of f(x,y), and H(x,y) is the Hessian matrix.
 | (3) |
In addition, f(x,y) and (dx,dy) can be shown by Eq. 4. The eigenvalue and eigenvector of the Hessian matrix are 
i and vi, respectively, and ki and 
i are coefficients (i = 1,2).
 | (4) |
Because the Hessian matrix is a symmetric matrix, the eigenvector is orthogonal, and therefore Eq. 3 can be transformed into Eq. 5.
 | (5) |
Curve Ci() becomes a parabola obtained by substituting = 0 for C(1,2) and is given by Eq. 6 with respect to the curve in the section where C(1,2) is cut along the vi.
 | (6) |
Curvature 
i in the direction of the vi for point (x,y) on curved surface f(x,y) is shown by Eq. 7.
 | (7) |
The linear hyperintensity area was treated as a curved surface and images were extracted on the basis of this curvature. The number of pixels showing hyperintensity, which corresponds to adipose tissue in muscle, was counted in extracted images to determine the percent change in hyperintense pixels in muscles before and after the exercise period.
M2: morphological analysis.
MR images were taken in the sagittal plane using a three-dimensional (3D) spoiled gradient-echo sequence (3D Volumetric Interpolated Breath-hold Examination, 3D-VIBE) (TR, 3.31 ms; TE, 1.3 ms; flip angle, 12°; bandwidth, 490 Hz/pixel; slice thickness, 2.5 mm; matrix size, 160 x 256; FOV, 187.5 x 300 mm; pixel size, 1.17 x 1.17 mm; average, 1; scan time, 19 s; 144 slices). The pelvis was defined as the center of the imaging region. The subcutaneous fat layer was removed from the image of each slice, and a 3D surface model was produced from the extracted data using a marching cubes algorithm (15, 16). On processed surface models generated from images taken before and after use of the equipment, landmarks (origin of right adductor longus, origin of left adductor longus, origin of right sartorius, and origin of left sartorius) on the anterior surface of the body were selected and registered by 3D Helmert transformation using the least-squares method.
The peak point of the dorsal hip in the posterior direction was determined from the 3D surface model before and after the exercise period, using the maximum coordinate of the 3D surface model in the posterior direction. The distance in the superior direction between two peak points was calculated by subtracting the coordinates of the superior direction of the peak point after the exercise period from the coordinates of the superior direction of the peak point before exercise. The distances were compared between training and normal equipment.
M3: diffusivity analysis.
MR images were taken in the sagittal plane using a SE-EPI diffusion-weighted sequence (TR, 4,100 ms; TE, 66 ms; flip angle, 90°; bandwidth, 1,345 Hz/pixel; slice thickness, 4.0 mm; matrix size, 64 x 128; FOV, 185 x 370 mm; pixel size, 2.89 x 2.89 mm; average, 3; scan time, 2 min 44 s; 20 slices; b value, 400 s/mm2; motion probing gradient, 6 directions). The right femur was defined as the center of the imaging region. Using an application we developed, fractional anisotropy (FA) was estimated from the image data using diffusion tensor analysis. Mean FA values measured in a 7 x 14-pixel region of interest (ROI) in each muscle were compared before and after the exercise period.
 | (8) |
FA defined by the above equation expresses the anisotropic ratio of water diffusion for each voxel; i.e., FA is 0.0 when water diffuses isotropically and 1.0 when water diffuses in one direction only. 1, 2, and 3 are eigenvalues of the diffusion tensor (6, 8, 13, 14).
|
RESULTS
|
|---|
M1: gray-scale analysis.
Images acquired using a Turbo Spin Echo sequence and a fat-suppressed Turbo Spin Echo sequence at the same site are shown in Figs. 1 and 2, respectively. In the image in Fig. 1, hyperintensity areas (indicated by arrows) were detected between the muscular tissues, whereas in the image in Fig. 2 the pixel value of the hyperintensity area between the muscular tissues in Fig. 1 was reduced. The reduced signals in the fat-suppressed image suggest that this area was adipose tissue.
View larger version (114K):
[in this window]
[in a new window]
|
Fig. 1. Images around the pelvis in the sagittal plane, acquired using a Turbo Spin Echo sequence. The area with hyperintensity pixels in muscle is indicated by arrows.
|
|
View larger version (114K):
[in this window]
[in a new window]
|
Fig. 2. Images at the same site as that in Fig. 1, acquired using a fat-suppressed Turbo Spin Echo sequence. The area with hyperintensity pixels shown in Fig. 1 is no longer apparent in the muscle. The disappearance of the hyperintense region in the fat-suppressed sequence suggests that this region was adipose tissue in muscle.
|
|
Images processed by extraction of linear hyperintensity objects (adipose tissue) from the image acquired using a Turbo Spin Echo sequence are shown in Fig. 3. In the processed image, the number of pixels extracted as adipose tissue in muscle was counted and divided by the number of muscular pixels to calculate the percentage of extracted pixels (those corresponding to adipose tissue in muscle) in each slice. The mean of these percentages from before to after the exercise period with training equipment is shown in Table 1, and those for use of the normal equipment are shown in Table 2. In subjects who wore training equipment, the percentage of adipose tissue after exercise (14.7 ± 2.9%) was on average 4.3% lower than that before exercise (19.0 ± 3.9%), and the difference was significant (P < 0.001). In contrast, there was no significant difference (before exercise 16.8 ± 7.3%; after exercise 16.8 ± 6.9%) in these data in subjects who exercised while wearing normal equipment. Individually, in subjects who wore training equipment, 12 subjects showed a significant difference at P < 0.01, one subject showed a significant difference at P < 0.05, and the one remaining subject showed a difference at P < 0.1. In six subjects who wore normal equipment, there was no significant difference. Adipose tissues were increasing in one subject (P < 0.1).
View larger version (35K):
[in this window]
[in a new window]
|
Fig. 3. Image in B processed by extraction of adipose tissue (linear hyperintensity objects) from image in A, acquired using a Turbo Spin Echo sequence. The percent change in the number of extracted pixels (corresponding to adipose tissue) was calculated.
|
|
View this table:
[in this window]
[in a new window]
|
Table 1. Percentage of pixels extracted as adipose tissue in muscle in subjects who exercised while wearing training equipment
|
|
View this table:
[in this window]
[in a new window]
|
Table 2. Percentage of pixels extracted as adipose tissue in muscle in subjects who exercised while wearing normal equipment
|
|
M2: morphological analysis.
3D surface data of the dorsal hip were compared before (Fig. 4A) and after (Fig. 4B) exercise using training equipment. The peak point is shown in Fig. 4. A comparative analysis image is presented in Fig. 5, and data for the peak point of the dorsal hip are shown for subjects who exercised with training and normal equipment in Table 3. A positive value indicates that the peak point of the dorsal hip after exercise was higher in the superior direction compared with that before exercise.
View larger version (88K):
[in this window]
[in a new window]
|
Fig. 4. Three-dimensional (3D) surface model around the pelvis produced using a marching cubes algorithm based on extracted data after image processing for elimination of subcutaneous fat. The maximum peak point of muscle in the posterior direction was measured in this model. A: 3D surface data of the dorsal hip before exercise. B: 3D surface data of the dorsal hip after exercise using training equipment. Asterisks indicate the position of a peak point.
|
|
View larger version (76K):
[in this window]
[in a new window]
|
Fig. 5. Comparative 3D surface data of the dorsal hip before and after exercise while wearing training equipment. The color reflects the distance in the anterior-posterior direction between the 2 data sets. If the point after exercise is positive in the posterior direction, the point is indicated as + (green-red in the color bar). In contrast, if the point is negative, it is indicated as – (green-blue in the color bar). In the bottom part, muscles after exercise while wearing training equipment are more negative in the posterior direction, with the opposite relationship in the top part. This result suggests that muscular shape was changed, and hip muscles were raised by exercise using the training equipment.
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. Comparative data for mean distance in the superior direction between peak points of the dorsal hip before and after exercise in case of using training and normal equipment
|
|
The peak point of the dorsal hip was an average of 10.8 mm higher after exercise with training equipment, but only an average of 1.2 mm higher in subjects who exercised with normal equipment, with these data showing a significant difference (P < 0.001). The results indicate that the peak point shifted in the superior direction, showing that the muscle shape was changed; i.e., the hips were held higher.
M3: diffusivity analysis.
FA map around the pelvis in the sagittal plane is shown in Fig. 6. Comparative data for mean FA in muscles before and after exercise for subjects who wore training equipment are shown in Table 4. FA after exercise with training equipment (0.431 ± 0.037) was on average 0.039 higher than that before exercise (0.392 ± 0.042), and these data showed a significant difference (P < 0.001). Individually, in subjects who wore training equipment, 10 subjects revealed a significant difference at P < 0.01, three subjects showed significance at P < 0.05, and the one remaining subject showed a difference at P < 0.1.
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 6. Fractional anisotropy (FA) map around the pelvis in the sagittal plane, acquired using DTI. For an intensity of 255, FA is 1.0; for an intensity of 0, FA is 0.0. Rectangular region of interest example is shown.
|
|
Comparative data for mean FA in muscles before and after exercise in subjects who wore normal equipment are shown in Table 5. There was no significant difference (before exercise: 0.391 ± 0.045, after exercise: 0.380 ± 0.046). In detail, there was no significant difference in six subjects, and in one subject FA decreased and showed a difference at P < 0.1.
This suggests that the anisotropic orientation of water diffusion in muscles increased after exercise with training equipment; i.e., increased FA indicates increased orientation of muscular fiber (Fig. 6).
|
DISCUSSION
|
|---|
The results (M2) shown in Table 3 suggest that the muscular outer shape of the hip itself was changed and the maximum peak point was shifted in the superior direction by an average of 10.8 mm. This muscle change induced lifting of the hips, including subcutaneous fat. Furthermore, adipose tissue between muscles (muscular fiber bundles) decreased after exercise while wearing training equipment, as indicated in Figs. 1 and 2 and Table 1 (M1) because excess tissues between muscles, including adipose tissue, were reduced by exercise with training equipment and the muscles were denser and tightened. Diffusion tensor MRI indicated that FA also increased after use of training equipment [Table 4 (M3)]. The anisotropy of water diffusion increased due to changes in muscle tissue and tightening of muscular fibers.
In the comparison between training and normal equipment, the distance (M2) between the peak points of the dorsal hip showed a significant difference (P < 0.001). Furthermore, in subjects who wore training equipment, there was a significant difference (P < 0.001) (M1 and M3). In contrast, there were no significant differences in subjects who wore normal equipment. It was suggested by these results (M1–M3) that training equipment is effective in that muscles were denser and tightened.
About the analytical method in the present study, average P value of M3 was higher than M1. Although this cause is the difference in tissue structure, and since there are a noise and distortion in a diffusion-weighted imaging of MRI, accuracy decrease has influenced the cause. Also in the technique of M2, accuracy may decrease according to the error about the creation of the 3D surface model. Therefore, in these methods, M1 may be the best in accuracy. However, since these three methods are investigating a different factor from the adipose tissue in muscles, outer shape, and internal muscular structure, each method is significant to the evaluation of the muscle state and quality. In addition, the overall evaluation by the results of three methods (M1–M3) is also important. The conclusions are strengthened by showing a significant difference in all the methods. In this research, the significant difference was obtained in all three methods. This reinforces the result that muscles were denser and tightened by exercise with training equipment and that training equipment was very effective.
In former papers(10, 22), measurement method of intramuscular adipose tissue was mainly the technique of measuring using an intensity threshold value. The method that uses an intensity threshold value is influenced by the intensity of surrounding pixels or the sensitivity of coils. In M1 in the present study, the above-mentioned problem is solved and adipose tissue with less quantity can also be detected. Analysis like M3 using diffusion tensor MRI is performed frequently these days (19). The control of the distortion of diffusion tensor MRI is difficult. In this research, distortion is suppressed using twice-refocused spin echo. Recently, research of mfMRI has been done frequently (20). If muscle functional MRI were taken in into these analysis methods in future studies, more information will be able to be acquired.
In the present study, training by exertion of stress on muscles was conducted and changes in muscle shape and inner tissues were examined using MRI. Walking exercise at more than 10,000 steps/day while wearing training equipment thickened and tightened the muscular fiber tissues. This occurred because the equipment exerted more stress on the biceps femoris and gluteus maximus muscles, providing a significant effect of exercise. The muscular training effect was examined using three techniques, which showed that both the muscular outer shape (M2) and the internal muscular state could be investigated based on the percentage of adipose tissue in muscles (M1) and FA (M3). This noninvasive approach may allow quantitation of the athletic ability (quality) of the muscles, which is not measured conventionally, and is an effective method for analyzing skeletal muscles.
|
FOOTNOTES
|
|---|
Address for reprint requests and other correspondence: R. Nakai, 53 Kawahara-cho Syogoin, Sakyo-ku, Kyoto 606–8507, Japan (e-mail: rnakai{at}frontier.kyoto-u.ac.jp)
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES
|
|---|
- Baar K, Nader G, Bodine S. Resistance exercise, muscle loading/unloading and the control of muscle mass. Essays Biochem 42: 61–74, 2006.[CrossRef][Web of Science][Medline]
- Baewer DV, Hoffman M, Romatowski JG, Bain JL, Fitts RH, Riley DA. Passive stretch inhibits central corelike lesion formation in the soleus muscles of hindlimb-suspended unloaded rats. J Appl Physiol 97: 930–934, 2004.[Abstract/Free Full Text]
- Bousquet-Santos K, Vaisman M, Barreto ND, Cruz-Filho RA, Salvador BA, Frontera WR, Nobrega AC. Resistance training improves muscle function and body composition in patients with hyperthyroidism. Arch Phys Med Rehabil 87: 1123–1130, 2006.[CrossRef][Web of Science][Medline]
- Bowers AL, Castro MD. The mechanics behind the image: foot and ankle pathology associated with gastrocnemius contracture. Semin Musculoskelet Radiol 11: 83–90, 2007.[CrossRef][Web of Science][Medline]
- Browning RC, Modica JR, Kram R, Goswami A. The effects of adding mass to the legs on the energetics and biomechanics of walking. Med Sci Sports Exerc 39: 515–525, 2007.[CrossRef][Web of Science][Medline]
- Clark CA, Werring DJ, Miller DH. Diffusion imaging of the spinal cord in vivo: estimation of the principal diffusivities and application to multiple sclerosis. Magn Reson Med 43: 133–138, 2000.[CrossRef][Web of Science][Medline]
- Ferreira GN, Teixeira-Salmela LF, Guimaraes CQ. Gains in flexibility related to measures of muscular performance: impact of flexibility on muscular performance. Clin J Sport Med 17: 276–281, 2007.[CrossRef][Web of Science][Medline]
- Finsterbusch J, Frahm J. Diffusion tensor mapping of the human brain using single-shot line scan imaging. J Magn Reson Imaging 12: 388–394, 2000.[CrossRef][Web of Science][Medline]
- Frost RA, Lang CH. Protein kinase B/Akt: a nexus of growth factor and cytokine signaling in determining muscle mass. J Appl Physiol 103: 378–387, 2007.[Abstract/Free Full Text]
- Gallagher D, Kuznia P, Heshka S, Albu J, Heymsfield SB, Goodpaster B, Visser M, Harris TB. Adipose tissue in muscle: a novel depot similar in size to visceral adipose tissue. Am J Clin Nutr 81: 903–910, 2005.[Abstract/Free Full Text]
- Goldspink G. Loss of muscle strength during aging studied at the gene level. Rejuvenation Res 10: 397–405, 2007.[CrossRef][Web of Science][Medline]
- Haddad F, Adams GR. Aging-sensitive cellular and molecular mechanisms associated with skeletal muscle hypertrophy. J Appl Physiol 100: 1188–1203, 2006.[Abstract/Free Full Text]
- Ino T, Nakai R, Azuma T, Yamamoto T, Tsutsumi S, Fukuyama H. Somatotopy of corticospinal tract in the internal capsule shown by functional MRI and diffusion tensor images. Neuroreport 18: 665–668, 2007.[Medline]
- Le Bihan D, Mangin JF, Poupon C, Clark CA, Pappata S, Molko N, Chabriat H. Diffusion tensor imaging: concepts and applications. J Magn Reson Imaging 13: 534–546, 2001.[CrossRef][Web of Science][Medline]
- Lorensen WE, Cline HE. Marching cubes: a high resolution 3D surface construction algorithm. Computer Graphics 21: 163–169, 1987.[CrossRef]
- Montani C, Scateni R, Scopigno R. A modified look-up table for implicit disambiguation of marching cubes. Visual Computer 10: 353–355, 1994.[CrossRef]
- Roberts TJ, Scales JA. Adjusting muscle function to demand: joint work during acceleration in wild turkeys. J Exp Biol 207: 4165–4174, 2004.[Abstract/Free Full Text]
- Schakman O, Gilson H, de Coninck V, Lause P, Verniers J, Havaux X, Ketelslegers JM, Thissen JP. Insulin-like growth factor-I gene transfer by electroporation prevents skeletal muscle atrophy in glucocorticoid-treated rats. Endocrinology 146: 1789–1797, 2005.[Abstract/Free Full Text]
- Schmid P, Lunkenheimer PP, Redmann K, Rothaus K, Jiang X, Cryer CW, Jaermann T, Niederer P, Boesiger P, Anderson RH. Statistical analysis of the angle of intrusion of porcine ventricular myocytes from epicardium to endocardium using diffusion tensor magnetic resonance imaging. Anat Rec (Hoboken) 290: 1413–1423, 2007.[CrossRef][Medline]
- Tawara N, Nitta O, Itoh A. [Visualizing muscle activation by fusing fast MR images]. Nippon Hoshasen Gijutsu Gakkai Zasshi 63: 1355–1361, 2007.[Medline]
- Urayama S, Hiroki M, Hanakawa T, Fukuyama H, Abe M, Sugimoto N. Detection of linear hyperintensity objects on high-resolution T2 enhanced image: quantification for severity of hypertension. IEICE Technical Report 103: 17–22, 2003.
- Wu EX, Tang H, Tong C, Heymsfield SB, Vasselli JR. In vivo MRI quantification of individual muscle and organ volumes for assessment of anabolic steroid growth effects. Steroids 73: 430–440, 2008.[CrossRef][Web of Science][Medline]
Copyright © 2008 by the American Physiological Society.
TOP
ABSTRACT
MATERIALS AND METHODS
