An in Vivo NMR probe circuit for improved sensitivity

JOURNALOFMAGNETICRESONANCE

An in Vivo NMR

54,526~532 (1983)

Probe Circuit for Improved Sensitivity JOSEPH MURPHY-B•

ESCH

Department of Pharmaceutical Chemistry, School of Pharmacy, University of California, San Francisco, California 94143 AND ALAN

P. KORETSKY

Chemical Biodynamics Division, Lawrence Berkeley Laboratory, University of Cal$ornia, Berkeley, California 94720 Received May 18, 1983

Success in using NMR to study metabolic processes and to produce well-resolved images of living tissues has led to a rapid increase in its application to problems in biology and medicine (I, 2). A serious problem with all of these studies is the degrading effect that intact tissues have upon probe sensitivity. The desire to perform these kinds of experiments has therefore necessitated studies of loss mechanisms associated with conductive samples (3, 4). We have recently discovered a tuning scheme for coils which greatly minimizes these losses when used for in vivo experiments. The success of this scheme demonstrates the importance of coil-to-ground parasitic losses. We describe here a ‘model for this loss mechanism and show how a simple modification of the normal tuning scheme can minimize its effects. The improvement in sensitivity afforded by this scheme is illustrated using model circuits and 3’P spectra obtained in vivo, using implanted coils. Because of their high conductivity, biological tissues can produce large losses in the tuned circuit of an NMR probe, losses reflected in a reduction in the circuit Q. These losses are caused primarily by currents induced in the tissue by the electric and magnetic fields of the coil. While little can be done to reduce currents generated by the magnetic field, shielding of the electric field from the tissue can reduce dielectric losses. Previous workers have developed circuit models for dielectric losses assuming that the paths of electric fields which penetrate the sample extend in a distributed manner from one side of the coil to the other (3,4). Their models of these coil-to-coil parasitics can be reduced to a capacitance Cd and a resistance & connected in parallel with the coil, as shown in Fii. 1. & and Cd have the effect of lowering both the resonance frequency and the Q of the tuned circuit. This description gives good quantitative explanations for the behavior of the losses incurred when biological samples are placed within a standard probe of a high resolution spectrometer (4). For the types of coils used to study animal organs in situ, additional paths for the 0022-2364183 $3.00 Copyri&t 0 1983 by Academic Press, Inc. All rights of repreduction in any form reserved.

526

COMMUNICATIONS

577

FIG. 1. A probe circuit employing standard tuning and matching which models coil-to-coil influences of the sample. L is the inductance of the sample coil, Ci and C, are the tuning and matching capacitors. and R, models the coil resistance. C, and & model the coil-to-coil parasitics.

electric fields exist. In our probe, a laboratory rat with an implanted coil is supported by a metal cradle. For proper tuning and shielding, the cradle, the probe casing, and the tuning circuit must all be grounded. Hence, paths for electric fields exist between the coil within the animal and the probe ground. The influence of these coil-toground pat&tics can be modeled with two lumped element branches between each side of the sample coil and ground, as indicated in Fig. 2a. For simplicity, each branch consists only of a resistance in series with a capacitor, with the capacitance modeling both the insulation about the coil wire and the reactive influence of the sample. In this circuit, the second branch involving Rd2 and Cdz can be neglected, since it is shorted by the ground lead of the tuning circuit. As indicated in Fig. 2b, the remaining branch transforms to the equivalent parallel components R, and C,, given by

FIG. 2. Probe circuits employing standard tuning and matching which model (a) the additional inIhrences of coil-to-ground parasitics, and (b) an equivalent parallel circuit. The tuned circuit elements are defined as in Fig. 1, except that Rti represents the parallel combination of R, and &, and Ci incorporates C, into the tuning The branches 4, and Cdl, and & and C a. model the distributed influences of the coil-toground pan&tics. R, and C, are given by Eqs. [l] and 121.

528

COMMUNICATIONS

where w0/2a is the resonance frequency of the tuned circuit. The influence of these additional components on the tuned circuit is the same as that of the coil-to-coil parasitics. The total Q of the circuit can then be expressed as 1 -=es

1 ad+a

1

131

where Qpd = &fw& and (SOI= I?&&. Once recognized, the influence of coil-to-ground parasitics can be minimized with a balanced tuning circuit containing matching capacitors of equal size on both sides of the coil, as indicated in Fig. 3a. Since the matching capacitors are essentially in series, their values must be approximately twice as large as the single matching capacitor of the standard circuit. For the purposes of illustration only, two assumptions are made which permit simplification of the circuit. First, we assume that the Q of the sample circuit is high enough to make the impedance of the matching capacitors small compared with the 50 ohm input impedance. Thus, the parasitic elements link in series through ground, reducing the circuit to that of Fig. 3b. Next, we assume that the two branches are identical, that is, that Cdl = Cd2 and Rdl = Rd2.This is approximately correct since the sample coil is generally situated symmetrically within an NMR probe. The coil

Rdn

FIG. 3. Circuit models showing (a) the influence of coil-to-ground parasitic elements on the balanced matching and tuning scheme, (b) the reduced circuit assuming a high circuit Q, and (c) the equivalent parallel circuit. The circuit elements are defined as before.

529

COMMUNICATIONS

is thereby balanced with respect to ground. Using transformation equations similar to Eqs. [l] and [2], the circuit reduces to that of Fig. 3c. The total Q for the balanced circuit can be expressed as -=1 L+L t41 Qb Qpd 2Qa’ Comparing Eqs. [3] and [4], it is apparent that the new circuit configuration has reduced the influence of the coil-to-ground parasitics by approximately one-half Thus, the balanced matching circuit yields a higher circuit Q. Since sensitivity varies as Q”* for a fixed frequency (5), we can also expect an improved signal-to-noise ratio. To demonstrate the advantage of using the balanced tuning scheme in the presence of strong coil-to-ground parasitics, we constructed a series of model circuits, all employing a two-turn half-inch diameter coil. Ceramic capacitors, variable from 3 to 20 pF, were used to tune and match the circuits. All parasitic elements consisted of a 22 ohm carbon resistor placed in series with a 5 pF silver mica capacitor. These particular component values were chosen to simulate the variations in tuning and Q typically observed for our in vivo experiments. Q measurements were performed near 100 MHz with a Wavetek Model 1062 sweep generator and a 20 dB directional coupler used to measure the reflected power. The Q’s were calculated by dividing the resonance frequency of the circuit by the width of the resonance curve at the halfpower points. The oscilloscope used for these measurements was calibrated with I MHz frequency markers from the sweep generator. The variation in Q for different loss mechanisms and tuning schemes are shown in Table 1. The 6rst column shows that the Q of the unloaded coil is insensitive to the matching scheme. The second column in Table 1 shows that the balanced matching configuration does not reduce the effects of coil-to-coil parasitics. Only when the loading is placed from each side of the coil to ground does the balanced matching scheme significantly improve the circuit Q and reduce the influence upon tuning. The improvement is greater than predicted by our simple model indicating that some additional factors are at play. We have been performing 3’P spectroscopy on rat organs by implanting the sample coil around the organ of interest (6). The circuit we use, shown in Fig. 4, is a modTABLE MEASUREMENTS

1

OF CIRCUIT Q FOR THE STANDARD BALANCED CONFIGURATIONS

Tuning scheme

Lossless” circuit

Standard Balanced

59 59

Coil-to-coil’ loading 29 29

AND

Coil-to-ground’ loading 26 45

’ Tuned circuit with no loading elements. ’ Simulated with a single parasitic element across the coil. ‘Simulated with two parasitic elements between each side of the coil and ground.

530

COMMUNICATIONS

>

c

CZ ,.$Cl7

>

z -P----

ct II

FIG. 4. A modified, balanced tuning circuit for in vivo experiments where a length of transmission line is required between the coil and the tuning and matching capacitors C, and C,, respectively. C, and C, are mounted near the coil and perform both partial tuning and partial matching of the circuit. C, is placed in the ground lead to balance the coil.

ification of the balanced matching capacitor circuit just described. The circuit differs from that of Fig. 3 in several respects. First, the tuning capacitor C, is no longer placed directly across the coil leads, but instead has one of its leads connected directly to ground. C, and C, are small chip capacitors (American Technical Ceramics) placed as close to the coil as possible and insulated from the tissue with silicone sealer (Dow Coming). C, helps to confine the large circulating current of the tuned circuit to the vicinity of the coil, thus improving the filling factor. C, performs a partial transformation to a lower impedance, reducing the rf voltage and, therefore, the conductive losses across the transmission line. C, is chosen somewhat larger than C, to offset the imbalancing effect of C, and to assure that C, will always be within range, regardless of the variations in sample coil inductance. The net result is that (1) the sample coil remains relatively balanced with respect to ground, (2) the transmission line does not have undue influence upon the filling factor or the circuit Q, and (3) the entire circuit can be tuned and matched outside the animal. To demonstrate the improvement in Q obtainable with this scheme, measurements were made with a coil implanted around the kidney of a laboratory rat. The sample coil consisted of two half-inch diameter loops separated by approximately onequarter inch. The coil was wound with #22 gauge copper wire insulated with polyethylene tubing (Clay Adams PE-100). The transmission line consisted of approximately 10 cm of this wire in twisted pair. The capacitors C, and C, were insulated with silicone sealer; their values were 13 and 27 pF, respectively. The Q of this coil-capacitor arrangement was typically 60 prior to implantation. The arrangement was surgically implanted around a rat kidney following the procedure of Koretsky et al. (6). Two days after surgery the animal was anesthetized and positioned in the probe, and the leads were connected to the capacitors Ci and C,. The values for these capacitors were in the range of 5 to 15 pF. The measured Q of the tuned circuit was 38. The transmission line leads outside the animal were then reversed so that the capacitor configuration resembled that of the standard tuning circuit of Fig. 1. In this configuration the capacitor C, only performs a partial impedance match. The Q for this arrangement was found to be 13. This dramatic drop in Q is evidence that the coilto-ground pa&tics are the dominant loss mechanism for implanted coils. We attempted to obtain spectra from a single rat kidney using the two circuit arrangements described above to compare their sensitivities. However, because of (predicted) excess capacitance on the coil, the latter (low Q) configuration could not

COMMUNICATIONS

531

be tuned to 97.3 MHz, the 3’P frequency of the spectrometer. We therefore obtained a spectrum from a second rat kidney using a tuning coil containing only a single, 10 pF tuning capacitor placed close to the coil. The spectrum obtained with this scheme is shown in Fig. 5a, while that for the balanced scheme is shown in Fig. Sb. The signal-to-noise ratio for the @ ATP peak was 16 for the former and 41 for the latter. This is a greater improvement in sensitivity than is predicted by the change in circuit Q (13 vs 38) and probably results from the increased filling factor obtained with the partial matching capacitor for the balanced scheme. While employing a Faraday shield may be the most desirable technique for eliminating dielectric losses (3), this is not always a practical solution, as is the case with

FIG. 5. ,‘P spectra of two rat kidneys at 97.3 MHz using (a) an implanted coil with only a single tuning capacitor mounted near the coil, and (b) an implanted coil employing the mod&d, balanced tuning scheme. The experiments were performed on a homebuilt spectrometer using a Nicokt 1180 data system and a Cryomagnet Systems wide bore magnet. The spectra were obtained in 2600 scans using 45” pukes and 130 msec recycle times. A 30 Hz exponential filter was applied to the FID befbre Fourier tram&m&on. The peaks shown are (1) methylene diphosphonic acid (Sigma Chemical Co.), pH 8.9, in acapillary mount& on the coil, (2) sugar phosphates, (3) inorganic phosphate, (4) urine phosphate and phospbodiasters, (5) y-ATP, (6) (Y-ATP,NAD (H), and (7) &ATP.

532

COMMUNICATIONS

implanted coils. Our balanced matching scheme is an easily implemented alternative to reduce the losses associated with living tissue. Since the balanced matching circuit actually performs better than predicted, our model and assumptions only serve as a starting point for a more complete description. Since the coils used for our performance tests are similar to those used as surface coils (7), we expect an improvement in sensitivity similar to that obtained for implanted coils. Furthermore, similar tuning and loss considerations hold for imaging configurations, and therefore the balanced matching circuit should yield improvements for whole body imaging and spectroscopy. ACKNOWLEDGMENTS We gratefully acknowledge the support of Dr. Michael W. Weiner, Dr. Thomas L. James, the Cardiovascular Research Institute, and the Medical Services of the Veterans Administration. We also thank Dr. Sam Wang for performing the coil implantations and Dr. Melvin Klein for some stimulating discussions. REFERENCES

1. DAVID G. GADIAN, “NMR and Its Applications to Living Systems,” Clarendon, 2. P. B~?-~oMLEY, Rev. Sci. Instrum. 53(9), 13 19 (1982). 3. D. I. HOULT AND P. C. LAUTERBUR, J. Mugn. Reson. 34, 425 (1979). 4. D. G. GADIAN AND F. N. H. ROBINSON, J. Magn. Reson. 34, 449 (1979). 5. D. I. HOULT AND R. E. RICHARDS, J. Mugn. Reson. 24, 71 (1976).

Oxford,

1982.

6. A. P. KORETSKY, W. STRAUSS, V. BASUS, J. MURPHY, P. BENDEL, T. L. JAMES, AND M. W. WEINER, “Acute Renal Failure” (H. E. Eliahu, Ed.), pp. 42-46, John Libbey, London, 1982. 7. J. J. H. ACKERMAN, T. H. GROVE, G. C. WONG, D. G. GADIAN, AND G. K. RADDA, Nature (London) 283, 167 (1980).