Skeletal muscle measurements by MRI

The following is the abstract of the article discussed in the subsequent letter:

	ABSTRACT

TOP ABSTRACT LETTER REFERENCES REPLY R-REFERENCES

Mitsiopoulos, N., R. N. Baumgartner, S. B. Heymsfield, W. Lyons, D. Gallagher, and R. Ross. Cadaver validation of skeletal muscle measurement by magnetic resonance imaging and computerized tomography. J. Appl. Physiol. 85(1): 115-122, 1998.Magnetic resonance imaging (MRI) and computerized tomography (CT) are promising reference methods for quantifying whole body and regional skeletal muscle mass. Earlier MRI and CT validation studies used data-acquisition techniques and data-analysis procedures now outdated, evaluated anatomic rather than adipose tissue-free skeletal muscle (ATFSM), studied only the relatively large thigh, or found unduly large estimation errors. The aim of the present study was to compare arm and leg ATFSM cross-sectional area estimates (cm²) by using standard MRI and CT acquisition and image-analysis methods with corresponding cadaver estimates. A second objective was to validate MRI and CT measurements of adipose tissue embedded within muscle (interstitial adipose tissue) and surrounding muscle (subcutaneous adipose tissue). ATFSM area (n = 119) by MRI [38.9 ± 22.3 (SD) cm²], CT (39.7 ± 22.8 cm²), and cadaver (39.5 ± 23.0 cm²) were not different (P > 0.001), and both MRI and CT estimates of ATFSM were highly correlated with corresponding cadaver values [MRI: r = 0.99, SE of estimate (SEE) 3.9 cm², P < 0.001; and CT: r = 0.99, SEE = 3.8 cm², P < 0.001]. Similarly good results were observed between MRI- and CT-measured vs. cadaver-measured interstitial and subcutaneous adipose tissue. For MRI-ATFSM the intraobserver correlation for duplicate measurements in vivo was 0.99 [SEE = 8.7 cm² (2.9%), P < 0.001]. These findings strongly support the use of MRI and CT as reference methods for appendicular skeletal muscle, interstitial and subcutaneous adipose tissue measurement in vivo.

	LETTER

TOP ABSTRACT LETTER REFERENCES REPLY R-REFERENCES

Skeletal muscle measurements by MRI

To the Editor: I was deeply interested when I read the paper by Mitsiopoulos et al. (3) describing careful validations of skeletal muscle measurement. Although computerized tomography (CI) and magnetic resonance imaging (MRI) are more and more widely used for body-composition determination, publications describing validation of those techniques are scarce (1, 3, 4). However, the major source of error in surface determination by MRI, the magnetic field gradient nonlinearity, has been omitted in these otherwise very complete controls. Magnetic field gradients are used to label signals as a function of their spatial origin. They are supposed to behave linearly through the field of view inside the magnet. This is true with a good approximation at the center of any magnetic resonance system, typically over a 25-cm field of view upon which the calibration of magnetic field gradient is performed when a system is installed. The farther a sample is located away from this central zone, the less accurate this calibration (5). Peripheral distortions due to magnetic field gradient nonlinearity cause a majoration of surfaces determined at a large distance from the center of the magnet, thus leading to an overestimation of subcutaneous fat volumes in obese subjects. To illustrate this, I performed a quick test of surface comparisons for a cylindrical object at two positions, 0 and 15 cm away from magnet center in direction x. Corresponding surfaces were 110.1 and 108.04 cm².

Nonlinearities are corrected in some MRI systems but not all of them. One should check this point, as we did in a recent body-composition study (2), since precise information on this subject is not easily obtained from manufacturers.

	REFERENCES

TOP ABSTRACT LETTER REFERENCES REPLY R-REFERENCES

1.   Abate, N., D. Burns, R. M. Peshock, A. Garg, and S. M. Grundy. Estimation of adipose tissue mass by magnetic resonance imaging: validation against dissection in human cadavers. J. Lipid Res. 35: 1490-1496, 1994[Abstract].

2.   Leroy-Willig, A., T. Willig, M. Henri-Feugeas V. Frouin, E. Mariner, A. Boulier, F. Barzic, E. Schouman-Claeys, and A. Syrota. Body composition MR study in patients with DMD, SMA and normal subjects. Magn. Reson. Imaging 15: 737-744, 1997[Medline].

3.   Mitsiopoulos, N., R. Baumgartner, S. B. Heymsfield, W. Lyons, D. Gallagher, and R. Ross. Cadaver validation of skeletal muscle measurement by magnetic resonance imaging and computerized tomography. J. Appl. Physiol. 85: 115-122, 1998[Abstract/Free Full Text].

4.   Ross, R., L. Léger, R. Guardo, J. De Guise, and B. G. Pike. Adipose tissue volume measured by MRI and CT in rats. J. Appl. Physiol. 70: 2164-2172, 1991[Abstract/Free Full Text].

5.   Saint-Jalmes, H., J. Taquin, and Y. Barjhoux. Design data for efficient axial gradient coils: application to NMR imaging. Magn. Reson. Med. 2: 245-252, 1985[Medline].

Anne Leroy-Willig, Institut de Myologie 75013 Paris, France

	REPLY

TOP ABSTRACT LETTER REFERENCES REPLY R-REFERENCES

To the Editor: We welcome the comments of Leroy-Willig that address distortions in magnetic resonance image reconstruction due to nonlinearity in magnetic field gradients. In practice, nonlinearity effects are small, well known (2), and automatically corrected for by the image-reconstruction software used by our scanner (General Electric Signa Advantage 1.5 Tesla, software version 5.4.2, Milwaukee, WI). Errors in determination of the cross section of objects such as human limbs are more likely to depend on main-field inhomogeneities that are introduced by the presence of the object inside the field of view. These imperfections have been thoroughly studied in the MRI literature, and means for their correction have also been extensively discussed (see Ref. 1 for a recent account). Such corrections usually require the acquisition of ancillary data from which a spatially dependent correction factor is derived. In our study, the use of phantoms and two different imaging modalities mitigates our concerns about such Bo (main field) inhomogeneities. As previously suggested (3), we estimate that most of our experimental errors arise from other sources such as through-plane partial volume artifacts and radio-frequency inhomogeneities. We appreciate Leroy-Willig bringing this often-unappreciated MRI phenomenon to our attention. As noted, the potential influence of distortions in magnetic resonance image reconstruction due to nonlinearity in magnetic field gradients is understood and now controlled for within most clinical scanners.

	REFERENCES

TOP ABSTRACT LETTER REFERENCES REPLY R-REFERENCES

1.   Alley, M. T., G. H. Glover, and N. J. Pelo. Gradient characterization using a Fourier-transform technique. Magn. Reson. Imaging 39: 581-587, 1998.

2.  Glover, G., and N. J. Pelo. Method for Correcting Image Distortion Due to Gradient Nonuniformity. US Patent 4591789. Issued 1983.

3.   Ross, R., L. Léger, D. Morris, J. De Guise, and R. Guardo. Quantification of adipose tissue by MRI: relationship with anthropometric variables. J. Appl. Physiol. 72: 787-795, 1992[Abstract/Free Full Text].

Robert Ross, Queen's University School of Physical and Health Education Kingston, Ontario, Canada K7L 3N6