MAGNETIC RESONANCE IN MEDICINE
27, 165-170 ( 1992)
NOTES Breast Coil Design for Low-Field MRI MARKKUKOMUAND MARTTIKORMANO Department of Diagnostic Radiology, University Central Hospital, 20520 Tiirku, Finland Received February 27, 1989; revised July 19, 1991; accepted October 28, 1991
Four double-breast coils were designed for the low-field resistive magnet MR imaging of female breasts at 0.02,0.04, and 0.1 T. The signal-to-noise ratio is optimized by shaping the coils, by constructing two different size coils at 0.02 T, and by using special wire for the turns of the solenoidal coils. The approximate linear dependence of the SNR on the LarmOr fnqUenCy is estimated. 0 1992 Academic Press, Inc. INTRODUCTION
There has been increasing interest in determining the efficiency of MRI as a tool in the diagnosis of human breast lesions ( I , 2). Several constructions for single- and double-breast coils have been reported (3-8). At low fields, the exposure of the patient to various magnetic fields is minimized, the proton relaxation time T1 is short and the T1-contrast is increased. The new methods, the spin-locking ( 9 , IO), chemicalshift imaging (CSI) fat/water discrimination ( 1 1 ) and magnetization transfer ( 1 2 ) techniques have also been applied at low fields. In order to find the value of breast MR imaging as a diagnostic modality, four coils have been constructed for the simultaneous imaging of both female breasts with low-field resistive magnet MRI systems. At low frequencies the signal-to-noise ratio (SNR) of the image is poor and is determined mainly by the quality of the receiving RF-coil, not by patient loading (13). Thus special attention has to be paid to optimizing the SNR. The preliminary results of the breast coil construction and the relaxation times of breast tissues at low fields have been reported ( 1 4 ) . In this paper, the low-field MRI breast coil technique is described and some illustrative breast images are presented. MATERIALS AND METHODS
The gist of breast coil design for low-field MRI is maximization of coil SNR, homogeneity of RF-field B1 over the imaging volume, extension of imaging area into the chest wall and laterally toward the axillae, and minimization of the artifacts caused by respiration and heart motion. The SNR of the coil can be improved by increasing the filling factor (sample volume/coil volume) and the quality factor Q of the coil. The filling factor can be optimized by shaping the coil according to the breast shape and by constructing several coils to match the different sizes of breasts. Low-field RFcoil design using Litz wire has been described ( 1 3 ) . Because of the horizontal field Bo, the solenoidal coil instead of the saddle coil can be constructed and this increases SNR ( 1 5 ) . 165
0740-3 194192 $5.00 Copynght 0 1992 by Academic Press, Inc All nghts of reproduction in any form reserved.
166
NOTES
Figure 1 shows the cross section of our double-breast coils. Four coils were designed for three MR Imagers (Instrumentarium Co., Helsinki, Finland) : two different size breast coils at 0.02 T and two coils at 0.04 and 0.1 T with dimensions equal to the larger coil at 0.02 T. They were constructed by winding Litz wire around two polyethylene plastic cups, which were mounted in a wooden or plastic support. The coils are connected in series. The wiring is partly extended on the hemispherical bottom of the cup, so the signal from the lower part of the coil is increased. The wiring was extended also on the underside of the cover sheet of the support. This rim helps in imaging the chest wall and the axillae. At 0.02 T the preamplifier is integrated in the receiver unit of the imager, whereas at 0.04 and 0.1 T the preamplifier of the receiver is built in the coil support near the coil. The Q value of unloaded and loaded coils was measured with the HP48 15A RF vector impedance meter ( 7). The SNR of the unloaded coils was determined by measuring the FID signal from the 5-ml pure water sample 2 cm above the bottom of the coil with the RF-pulse power on and off with a 1 kHz bandwidth of the receiver. SNR of the loaded coil was calculated from the relationship SNR a Q ' / 2 ( 1 5 ) . At 0.02 and 0.04 T, the RF-pulses were transmitted via the large-body RF-coil of the imager. The separate flat rectangular transmitting coil was built in the breast coil support of the 0.1-T MR scanner. The size of this four-turn coil was 3 1 X 48 cm2 and it was wound from 1.5-mm diameter copper wire. The homogeneity of the RF-field B , was tested by imaging a large water phantom inside the coil and measuring the signal intensity both as a function of the distance from the bottom of the coil and in the axial middle plane of the coil. During the MR imaging the patient is lying in a prone position on the patient couch with two extra pads added for patient comfort. The positions of the sagittal or coronal, contiguous multisectional breast images are selected from a single T1-weighted axial image of both breasts. The T 1 maps can also be generated from two inversion-recovery ( I R) images. The field-echo PS3D-sequences with 12 or 24 slices of 5-mm slice width are also used with the Gd-DTPA contrast agent (2). The displayed imaging matrix is
0
\ /
Litz-wire
0
0
20
LO 60 d (mm)
80
100
FIG. I . The cross section of the two-cup breast coils. The dimensions d , h, and 1 of the coils are listed in Table 1. FIG.2. The relative signal intensity as a function of the distance d from the bottom of the coils B and C at 0.02 T. The coils at 0.04 and 0.1 T behave similar to the coil C.
167
NOTES
- -
128 X 256 or 256 X 256 and the pixel size of these images is 0.7 * 1.4 mm. The slice thickness is 5, 7, or 10 mm. The complete examination of both breasts can be accomplished in 1 h. RESULTS
The properties of the breast coils are presented in Table 1. The difference between the Q-values of the coils at 0.02 T and higher fields is due to the loading effect of the receiver preamplifier. This effect may cause small error in the values of SNR (loaded) at 0.02 T. The SNR of the loaded coils seems to behave almost linearly as a function of the Larmor frequency. From the values of the SNR of the coils at 0.02 T (coil C), 0.04 and 0.1 T, the approximate proportionality SNR a f A.2 can be calculated. This frequency dependence is in satisfactory agreement with the results of Hoult and Richards ( 1 5 ) and Edelstein et al. (16) considering the small frequency range and the patient loading effect. The relative signal intensity as a function of the distance from the bottom of the coils B and C at 0.02 T is illustrated in Fig. 2. The standard deviation of the pixel intensity over the 10-cm-diameter circular area in the coronal middle plane of coil C was about 15%. Figure 3 shows the sagittal image of a normal volunteer imaged with the breast coil C at 0.02 T. The field-echo PS3D-image of a normal breast at 0.04 T is illustrated in Fig. 4 and the IR-image of large breast carcinoma can be seen in Fig. 5. Axial and coronal imaging of both breasts is thus possible simultaneously with decreased spatial resolution. TABLE 1 Properties of the Breast Coils at 0.02, 0.04, and 0.1 T
0.02 T Coil n
d (cm) h (cm) 1 (cm)
L (PW Q (unloaded) Q (loaded) SNR (unloaded) SNR (loaded) SNR-improvement" Sh ( k H 4
B
C
32 18 12 15 6.0 1.5 2.5 1.5 81 28 540 480 350 330 28 23 23 19 9 8 6.4 2.0
0.04 T
0.1 T -
12 15 7.5 1.5 14 195 125 52 42 3
8 15 7.5 1.5 6.4 310 230 140 121 6 5.2
7.1
N o k . n, total number of turns; d, h, I, dimensions in Fig. I ; L, inductance of the coil; 4fi, Effect of coil loading on resonance frequency . SNR of the loaded breast coil compared with the SNR of the large whole-body coil of the imager.
168
NOTES
FIG.3. The sagittal PS 500/60 image of a normal volunteer at the nipple plane. The field strength is 0.02 T, the FOV is 128 X 179 mm2, and the slice.width is 10 mm.
---
Because of the short T I ( Y 120 160 ms) of fat, the PS sequence with TR of 150 * 200 ms and TE of 40 ms gives the best image quality. The fat and fibroglandular tissue are clearly distinguishable, but it is not possible to separate glandular tissue from fibrous tissue with standard sequences. Tumors located within the fat can be recognized quite easily, despite the smallest being interpreted at only 7 mm in diameter in X-ray mammography. Tumors within the glandular tissue were not as evidently demonstrated by standard pulse sequences used, but may be recognized from calculated TI maps, from the spin lock or magnetization transfer images with a fat/water separation technique or by using Gd-DTPA as a contrast agent.
--
DISCUSSION
The quality of our specialized breast coils has turned out to be sufficient for the practical imaging of female breasts even at very low fields. The SNR of the breast coil is significantly larger than the SNR of the large whole-body coil. In order to maximize
NOTES
169
FIG. 4. The axial PS3D 125/25 image of the normal volunteer at 0.04 T. The slice width is 5 mm and FOV is 343 X 452 mm’. FIG. 5. The axial IR 1000/30/ 18 image of the breast carcinoma at 0.1 T. The slice width is 7 mm and the FOV is 358 X 358 mm’.
the SNR of the image at lowest fields, the construction of different size two-cup coils is desirable. The lack of ionizing hazards, simultaneous imaging of both breasts with reduced examination time, elimination of the motion artifacts, and enlarged imaging area toward the chest wall and the axillae, the clear distinction between the fat and fibroglandular tissue and the new spin-locking and magnetization transfer modalities with the CSI fatlwater separation are the main advantages of the present low-field breast MR imaging technique. Work on relaxation parameters T1 and T2 of breast tissues, evaluation of the MR images in diagnosis of breast lesions, and development of the CSI fat / water separation, spin-locking, and magnetization transfer techniques is in progress. ACKNOWLEDGMENTS This work was supported by the Sigrid Juselius Foundation, Helsinki, Finland, and the Instrumentarium Research Foundation, Helsinki, Finland. We thank Drs. Kirsti Dean, Peter Dean, and Marja-Leena Majurin for valuable discussions and help in the breast imaging. REFERENCES 1. S. J. EL YOUSEF,R. H. DUCHESNEAU, R. J. ALF’IDI,J. R. HAAGA,P. J. BRYAN,AND J. P. LIPUMA, Radiology 150, 761 (1984).
2. S. H. HEYWANG-KOBRUNNER, “Contrast-Enhanced MRI of the Breast,” Schering, Munich, 1990. 3. G. M. BYDDER,W. L. CURATI,D. G. GADIAN,A. S. HALL, R. R. HARMAN,P. R. BUTSEN,D. J. AND I. R. YOUNG,J. Comput. Assist. Tomogr. 9,987 ( 1985). GILDERDALE, 4. C . B. STELLING, P. c. WANG, A. LIEBER,s. s. MATTINGLY,W. 0. GRIFFEN,AND D. E. POWELL, Radiology 154, 457 (1985). 5. N. T. WOLFMAN,R. MORAN,P. R. MORAN,AND N. KARSTAEDT, Radiology 155,241 ( 1985). 6. J. P. HORNAK, J. SZUMOWSKI, D. RUBENS,J. JANUS,AND R. G. BRYANT, Radiology 161,832 ( 1986). 7. M. A. SMITHAND D. W. WE,Magn. Reson. Imaging 4,455 ( 1986). 8. P. W. MCOWANAND T. W. REDPATH,Phys. Med. Biol. 32,259 ( 1987). 9. R. E. SEPPONEN, J. T. SIPPONEN,J. A. POHJONEN, AND J. I. TANTTU,J. Comput. Assist. Tomogr. 9, 1007 (1985).
NOTES 10. G. E. SANTYR,R. M. HENKELMAN, AND M. BRONSKILL, J. Magn. Reson. Med. 12,25 (1989). I ! . J. I. TANTTU,J. A. POHJONEN,AND R. E. SEPPONEN,“Proceedings, SMRM, 6th Annual Meeting, 1987,” p. 365. 12. J. I. TANTTU,C. E. KAHN,JR., R. E. SEPPONEN, E. A. HOLLAND,E. TIERALA, AND M. J. LIPTON, Radiology 177 ( P ) , 245 (1990). I S . M. SAVELAINEN AND M. SEPPANEN, “Proceedings, SMRM, 6th Annual Meeting, 1987,” p. 840. 14. M. KOMU,K. DEAN,P. DEAN,AND M. KORMANO “Proceedings, SMRM, 9th Annual Meeting, 1990,” p. 102. 15. D. I. HOULTAND R. E. J. RICHARDS, J. Magn. Reson. 24, 71 ( 1976). 16. W. A. EDELSTEIN, G . H. GLOVER,C. J. HARDY,AND R. W. REDINGTON, Magn. Reson. Med. 3,604 (1986).
27, 165-170 ( 1992)
NOTES Breast Coil Design for Low-Field MRI MARKKUKOMUAND MARTTIKORMANO Department of Diagnostic Radiology, University Central Hospital, 20520 Tiirku, Finland Received February 27, 1989; revised July 19, 1991; accepted October 28, 1991
Four double-breast coils were designed for the low-field resistive magnet MR imaging of female breasts at 0.02,0.04, and 0.1 T. The signal-to-noise ratio is optimized by shaping the coils, by constructing two different size coils at 0.02 T, and by using special wire for the turns of the solenoidal coils. The approximate linear dependence of the SNR on the LarmOr fnqUenCy is estimated. 0 1992 Academic Press, Inc. INTRODUCTION
There has been increasing interest in determining the efficiency of MRI as a tool in the diagnosis of human breast lesions ( I , 2). Several constructions for single- and double-breast coils have been reported (3-8). At low fields, the exposure of the patient to various magnetic fields is minimized, the proton relaxation time T1 is short and the T1-contrast is increased. The new methods, the spin-locking ( 9 , IO), chemicalshift imaging (CSI) fat/water discrimination ( 1 1 ) and magnetization transfer ( 1 2 ) techniques have also been applied at low fields. In order to find the value of breast MR imaging as a diagnostic modality, four coils have been constructed for the simultaneous imaging of both female breasts with low-field resistive magnet MRI systems. At low frequencies the signal-to-noise ratio (SNR) of the image is poor and is determined mainly by the quality of the receiving RF-coil, not by patient loading (13). Thus special attention has to be paid to optimizing the SNR. The preliminary results of the breast coil construction and the relaxation times of breast tissues at low fields have been reported ( 1 4 ) . In this paper, the low-field MRI breast coil technique is described and some illustrative breast images are presented. MATERIALS AND METHODS
The gist of breast coil design for low-field MRI is maximization of coil SNR, homogeneity of RF-field B1 over the imaging volume, extension of imaging area into the chest wall and laterally toward the axillae, and minimization of the artifacts caused by respiration and heart motion. The SNR of the coil can be improved by increasing the filling factor (sample volume/coil volume) and the quality factor Q of the coil. The filling factor can be optimized by shaping the coil according to the breast shape and by constructing several coils to match the different sizes of breasts. Low-field RFcoil design using Litz wire has been described ( 1 3 ) . Because of the horizontal field Bo, the solenoidal coil instead of the saddle coil can be constructed and this increases SNR ( 1 5 ) . 165
0740-3 194192 $5.00 Copynght 0 1992 by Academic Press, Inc All nghts of reproduction in any form reserved.
166
NOTES
Figure 1 shows the cross section of our double-breast coils. Four coils were designed for three MR Imagers (Instrumentarium Co., Helsinki, Finland) : two different size breast coils at 0.02 T and two coils at 0.04 and 0.1 T with dimensions equal to the larger coil at 0.02 T. They were constructed by winding Litz wire around two polyethylene plastic cups, which were mounted in a wooden or plastic support. The coils are connected in series. The wiring is partly extended on the hemispherical bottom of the cup, so the signal from the lower part of the coil is increased. The wiring was extended also on the underside of the cover sheet of the support. This rim helps in imaging the chest wall and the axillae. At 0.02 T the preamplifier is integrated in the receiver unit of the imager, whereas at 0.04 and 0.1 T the preamplifier of the receiver is built in the coil support near the coil. The Q value of unloaded and loaded coils was measured with the HP48 15A RF vector impedance meter ( 7). The SNR of the unloaded coils was determined by measuring the FID signal from the 5-ml pure water sample 2 cm above the bottom of the coil with the RF-pulse power on and off with a 1 kHz bandwidth of the receiver. SNR of the loaded coil was calculated from the relationship SNR a Q ' / 2 ( 1 5 ) . At 0.02 and 0.04 T, the RF-pulses were transmitted via the large-body RF-coil of the imager. The separate flat rectangular transmitting coil was built in the breast coil support of the 0.1-T MR scanner. The size of this four-turn coil was 3 1 X 48 cm2 and it was wound from 1.5-mm diameter copper wire. The homogeneity of the RF-field B , was tested by imaging a large water phantom inside the coil and measuring the signal intensity both as a function of the distance from the bottom of the coil and in the axial middle plane of the coil. During the MR imaging the patient is lying in a prone position on the patient couch with two extra pads added for patient comfort. The positions of the sagittal or coronal, contiguous multisectional breast images are selected from a single T1-weighted axial image of both breasts. The T 1 maps can also be generated from two inversion-recovery ( I R) images. The field-echo PS3D-sequences with 12 or 24 slices of 5-mm slice width are also used with the Gd-DTPA contrast agent (2). The displayed imaging matrix is
0
\ /
Litz-wire
0
0
20
LO 60 d (mm)
80
100
FIG. I . The cross section of the two-cup breast coils. The dimensions d , h, and 1 of the coils are listed in Table 1. FIG.2. The relative signal intensity as a function of the distance d from the bottom of the coils B and C at 0.02 T. The coils at 0.04 and 0.1 T behave similar to the coil C.
167
NOTES
- -
128 X 256 or 256 X 256 and the pixel size of these images is 0.7 * 1.4 mm. The slice thickness is 5, 7, or 10 mm. The complete examination of both breasts can be accomplished in 1 h. RESULTS
The properties of the breast coils are presented in Table 1. The difference between the Q-values of the coils at 0.02 T and higher fields is due to the loading effect of the receiver preamplifier. This effect may cause small error in the values of SNR (loaded) at 0.02 T. The SNR of the loaded coils seems to behave almost linearly as a function of the Larmor frequency. From the values of the SNR of the coils at 0.02 T (coil C), 0.04 and 0.1 T, the approximate proportionality SNR a f A.2 can be calculated. This frequency dependence is in satisfactory agreement with the results of Hoult and Richards ( 1 5 ) and Edelstein et al. (16) considering the small frequency range and the patient loading effect. The relative signal intensity as a function of the distance from the bottom of the coils B and C at 0.02 T is illustrated in Fig. 2. The standard deviation of the pixel intensity over the 10-cm-diameter circular area in the coronal middle plane of coil C was about 15%. Figure 3 shows the sagittal image of a normal volunteer imaged with the breast coil C at 0.02 T. The field-echo PS3D-image of a normal breast at 0.04 T is illustrated in Fig. 4 and the IR-image of large breast carcinoma can be seen in Fig. 5. Axial and coronal imaging of both breasts is thus possible simultaneously with decreased spatial resolution. TABLE 1 Properties of the Breast Coils at 0.02, 0.04, and 0.1 T
0.02 T Coil n
d (cm) h (cm) 1 (cm)
L (PW Q (unloaded) Q (loaded) SNR (unloaded) SNR (loaded) SNR-improvement" Sh ( k H 4
B
C
32 18 12 15 6.0 1.5 2.5 1.5 81 28 540 480 350 330 28 23 23 19 9 8 6.4 2.0
0.04 T
0.1 T -
12 15 7.5 1.5 14 195 125 52 42 3
8 15 7.5 1.5 6.4 310 230 140 121 6 5.2
7.1
N o k . n, total number of turns; d, h, I, dimensions in Fig. I ; L, inductance of the coil; 4fi, Effect of coil loading on resonance frequency . SNR of the loaded breast coil compared with the SNR of the large whole-body coil of the imager.
168
NOTES
FIG.3. The sagittal PS 500/60 image of a normal volunteer at the nipple plane. The field strength is 0.02 T, the FOV is 128 X 179 mm2, and the slice.width is 10 mm.
---
Because of the short T I ( Y 120 160 ms) of fat, the PS sequence with TR of 150 * 200 ms and TE of 40 ms gives the best image quality. The fat and fibroglandular tissue are clearly distinguishable, but it is not possible to separate glandular tissue from fibrous tissue with standard sequences. Tumors located within the fat can be recognized quite easily, despite the smallest being interpreted at only 7 mm in diameter in X-ray mammography. Tumors within the glandular tissue were not as evidently demonstrated by standard pulse sequences used, but may be recognized from calculated TI maps, from the spin lock or magnetization transfer images with a fat/water separation technique or by using Gd-DTPA as a contrast agent.
--
DISCUSSION
The quality of our specialized breast coils has turned out to be sufficient for the practical imaging of female breasts even at very low fields. The SNR of the breast coil is significantly larger than the SNR of the large whole-body coil. In order to maximize
NOTES
169
FIG. 4. The axial PS3D 125/25 image of the normal volunteer at 0.04 T. The slice width is 5 mm and FOV is 343 X 452 mm’. FIG. 5. The axial IR 1000/30/ 18 image of the breast carcinoma at 0.1 T. The slice width is 7 mm and the FOV is 358 X 358 mm’.
the SNR of the image at lowest fields, the construction of different size two-cup coils is desirable. The lack of ionizing hazards, simultaneous imaging of both breasts with reduced examination time, elimination of the motion artifacts, and enlarged imaging area toward the chest wall and the axillae, the clear distinction between the fat and fibroglandular tissue and the new spin-locking and magnetization transfer modalities with the CSI fatlwater separation are the main advantages of the present low-field breast MR imaging technique. Work on relaxation parameters T1 and T2 of breast tissues, evaluation of the MR images in diagnosis of breast lesions, and development of the CSI fat / water separation, spin-locking, and magnetization transfer techniques is in progress. ACKNOWLEDGMENTS This work was supported by the Sigrid Juselius Foundation, Helsinki, Finland, and the Instrumentarium Research Foundation, Helsinki, Finland. We thank Drs. Kirsti Dean, Peter Dean, and Marja-Leena Majurin for valuable discussions and help in the breast imaging. REFERENCES 1. S. J. EL YOUSEF,R. H. DUCHESNEAU, R. J. ALF’IDI,J. R. HAAGA,P. J. BRYAN,AND J. P. LIPUMA, Radiology 150, 761 (1984).
2. S. H. HEYWANG-KOBRUNNER, “Contrast-Enhanced MRI of the Breast,” Schering, Munich, 1990. 3. G. M. BYDDER,W. L. CURATI,D. G. GADIAN,A. S. HALL, R. R. HARMAN,P. R. BUTSEN,D. J. AND I. R. YOUNG,J. Comput. Assist. Tomogr. 9,987 ( 1985). GILDERDALE, 4. C . B. STELLING, P. c. WANG, A. LIEBER,s. s. MATTINGLY,W. 0. GRIFFEN,AND D. E. POWELL, Radiology 154, 457 (1985). 5. N. T. WOLFMAN,R. MORAN,P. R. MORAN,AND N. KARSTAEDT, Radiology 155,241 ( 1985). 6. J. P. HORNAK, J. SZUMOWSKI, D. RUBENS,J. JANUS,AND R. G. BRYANT, Radiology 161,832 ( 1986). 7. M. A. SMITHAND D. W. WE,Magn. Reson. Imaging 4,455 ( 1986). 8. P. W. MCOWANAND T. W. REDPATH,Phys. Med. Biol. 32,259 ( 1987). 9. R. E. SEPPONEN, J. T. SIPPONEN,J. A. POHJONEN, AND J. I. TANTTU,J. Comput. Assist. Tomogr. 9, 1007 (1985).
NOTES 10. G. E. SANTYR,R. M. HENKELMAN, AND M. BRONSKILL, J. Magn. Reson. Med. 12,25 (1989). I ! . J. I. TANTTU,J. A. POHJONEN,AND R. E. SEPPONEN,“Proceedings, SMRM, 6th Annual Meeting, 1987,” p. 365. 12. J. I. TANTTU,C. E. KAHN,JR., R. E. SEPPONEN, E. A. HOLLAND,E. TIERALA, AND M. J. LIPTON, Radiology 177 ( P ) , 245 (1990). I S . M. SAVELAINEN AND M. SEPPANEN, “Proceedings, SMRM, 6th Annual Meeting, 1987,” p. 840. 14. M. KOMU,K. DEAN,P. DEAN,AND M. KORMANO “Proceedings, SMRM, 9th Annual Meeting, 1990,” p. 102. 15. D. I. HOULTAND R. E. J. RICHARDS, J. Magn. Reson. 24, 71 ( 1976). 16. W. A. EDELSTEIN, G . H. GLOVER,C. J. HARDY,AND R. W. REDINGTON, Magn. Reson. Med. 3,604 (1986).
