Field-cycled proton-electron double-resonance imaging of free radicals in large aqueous samples

David J. Lurie*, James M.S. Hutchison, Lawrence H. Bell, Ian Nicholson, David M. Bussell, John R. Mallard

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

79 Citations (Scopus)

Abstract

We have recently published a new method of imaging free radicals in aqueous
solutions called proton-electron double-resonance imaging (PEDRI) ( I ). In this
technique a conventional proton NMR image is collected while the EPR resonance
of a free radical solute is irradiated. If the EPR irradiation has sufficient power, the
NMR signal from those protons being relaxed by the paramagnetic solute is enhanced, and the parts of the sample containing free radical exhibit greater intensity
in the final image. Unlike EPR imaging (2) the sample size in PEDRI is not constrained by magnetic field gradient requirements. In this Communication we present
the first results of an extension of PEDRI which uses magnetic field cycling, allowing
larger samples to be imaged with lower levels of applied radiofrequency power.
PEDRI is an imaging version of a dynamic nuclear polarization experiment (3-
5). The enhancement of the NMR signal upon irradiation of the EPR of the solute
may be written empirically as
ill
where AZ and A0 are the NMR signals with and without EPR irradiation. Assuming
that the main relaxation mechanism for protons is a dipolar interaction with the free
radical’s unpaired electron, the enhancement is given by the relationship
[21
where ys and yp are the electron and proton gyromagnetic ratios, B2 is the EPR irradiation RF magnetic field in the rotating frame, and T I and r2 are the electron relaxation
times. E is frequently negative, indicating that the NMR signal changes phase by 180
upon irradiation of the EPR resonance, while its magnitude changes by the factor 1 El.
With a given sample, a particular RF magnetic field strength B2 must be applied
to achieve a certain enhancement. We may further write
431 0022-2364/89 $3.00
Copyright 0 1989 by Academic Press, Inc.
All rights of reproductmn in any form reserved.
432 COMMUNICATIONS
B; oc P, 131
where P is the power of the applied EPR irradiation. Assuming a constant sample
conductivity, the power is related to the EPR frequency v2 and the volume of the EPR
resonator Vas
or
PKu;v [41
P c$ B:V,
where B. is the static magnetic field strength.
[51
If we impose a requirement for a certain level of enhancement with a particular
type of sample, the maximum volume of the sample is constrained by the power
available for the EPR irradiation (provided RF penetration into the sample is not a
limiting factor). If biological samples are to be imaged, or if PEDRI is to be used in
vivo, it is desirable to keep the power per unit volume as low as possible to avoid
excessive sample heating. Initial PEDRI experiments were performed using a static
magnetic field of 0.04 T, giving an EPR frequency of 1123 MHz ( 1). Using a sample
of 2 mM TEMPOL free radical solution (4-hydroxy-2,2,6,6-tetramethylpiperidineI-oxyl) (Aldrich Chemical Co.) an enhancement of E = -8 was obtained with a
resonator volume of 1.5 ml by applying a power of 0.9 W. Assuming a good filling
factor, the applied power per unit mass of the sample was approximately 600 W/kg,
too high for experiments on biological samples.
Equations [ 21 to [ 5 ] indicate that a large decrease in applied power per unit volume
can be achieved by reducing Bo. A 4-fold reduction in B,, for example, will yield a
16-fold decrease in P/V. Provided B2 remains constant, the enhancement level will
be maintained at the lower field strength. If the irradiation power is limited by the
available hardware, the volume of the sample can thus be increased by a factor of 16
in this case.
The price which must inevitably be paid for decreasing B,, is a degradation of the
image signal-to-noise ratio. Assuming a B. 3’2 dependence, a fourfold reduction in B.
would decrease the SNR by a factor of eight. While signal averaging can be used to
recover SNR, scan times may become unacceptably long if B. is reduced too far.
Taking the figure of 600 W/kg from our 0.04 T PEDRI experiments, Eq. [ 51 indicates that a reduction of the static field to 0.0025 T should bring the applied power
down to less than 2.5 W/kg, an acceptable level for biological experiments. The SNR,
however, would be reduced by a factor of approximately 64, requiring 4096 averages
to restore image quality. Fortunately the conflicting requirements on Bo, namely the
need for a low value to reduce the applied RF power and a high value to maintain
SNR, are not mutually exclusive if the powerful technique of magnetic field-cycling
is used in conjunction with PEDRI.
Field-cycling has been used for a number of years to study relaxation phenomena
at very low magnetic field strengths ( 6). A field-cycling (FC) experiment includes
three distinct periods during each of which the static field B. has a different value:
polarization at Bg (high field), evolution at Bg (low field), and detection at BF
(intermediate field). The nuclear magnetization is allowed to build up during the
COMMUNICATIONS 433
polarization period which usually lasts longer than the NMR T, of the sample. Relaxation occurs during the evolution period which lasts approximately as long as T, .
Finally & is switched to an intermediate level and the NMR signal is detected by
applying a 90“ pulse in the usual way. The field is switched between levels in a time
much less than T, . The gain in sensitivity of an FC experiment over a conventional
NMR experiment performed at a constant field of BE ranges from B,P BF/ Bg2, if the
evolution period tE is short compared with T, , to B$‘/ Bz, if tE is long compared with
T, (6).
In field-cycled PEDRI (FC-PEDRI) the polarization and detection magnetic fields
Bz and Bt are chosen to give an acceptable SNR. The EPR irradiation takes place
during the evolution period with BE depending on the desired EPR irradiation frequency, which can be chosen independently of Bfl’ and BF in order to achieve an
acceptable power per unit volume figure.
We have implemented FC-PEDRI on our homebuilt whole-body proton NMR
imager. This instrument has a four-coil, vertical-field, side-access, resistive magnet
which normally operates at 0.04 T, giving an NMR frequency of 1.7 MHz ( 7). For
this work we have modified the apparatus to operate at 0.0 1 T; this field was used for
both polarization and detection, giving an NMR frequency of 425 kHz. Circuitry was
constructed to down-convert the RF pulses from 1.7 MHz to 425 kHz and to upconvert the 425 kHz NMR signals to 1.7 MHz. This approach allowed us to use the
original transmit and receive electronics without readjustment when the imager was
used at the lower field strength. A miniature split-solenoid RF coil (diameter 85 mm)
was used for transmission and reception at 425 kHz. It was connected to the RF
transmitter and receiver via a passive transmit/receive switch.
Field-cycling was achieved using a field compensation technique (6). We did not
attempt to switch the current in the imager’s magnet since this would have placed
unacceptable demands on the magnetic power supply and coil insulation due to the
large inductance of the whole-body magnet. Instead, the field from the main magnet
was held constant at 0.01 T and the current in a much smaller secondary magnet coil
situated inside the imager’s gradient coil tube was switched on and off. The field
produced by the secondary coil was arranged to be in opposition to that from the
main magnet so that when the secondary coil was driven the net field at the center of
the coils was reduced. The secondary coil was an air-cored, water-cooled, two-coil
Helmholtz design with an internal diameter of 22 cm and an inductance of 24 mH.
Each coil of the Helmholtz pair had 188 turns of 2.5 mm diameter copper wire and
the two coils were connected in series, requiring a current of 3.67 A to produce a field
of 0.005 T at the center of the magnet. A constant-current power supply (HewlettPackard 6269B) was used for the secondary coil, and the current was switched under
control from the imager’s pulse programmer using power MOSFET transistors. The
switching time of the secondary coil was less than 10 ms.
The field compensation approach to field-cycling has the advantage that only the
main magnet is driven during the detection period, when the greatest demands are
placed on the spatial homogeneity and temporal stability of the magnetic field at the
sample. In FC-PEDRI the homogeneity of the magnetic field during the evolution
period need only be good enough to irradiate the EPR line of interest throughout the
sample: in these experiments the linewidth was more than 4 MHz at an EPR fre-
434 COMMUNICATIONS
quency of 160 MHz so that a variation of BE of more than f 1% over the sample
volume could be tolerated, while the calculated homogeneity of the Helmholtz pair
was better than k 1000 ppm over the sample volume. The disadvantage of field compensation is the inevitable interaction between the primary and the secondary magnet
coils caused by their close proximity: in our apparatus this gave rise to an instability
of the primary magnetic field when the current in the secondary coil was switched.
The effect became more serious as the primary and secondary field strengths were
increased, so that we were restricted to an upper limit of 0.0 1 T for BE and Bg .
A synthesized frequency generator (Farnell Instruments, UK) provided the EPR
excitation signal which was amplified by a 1 W broadband amplifier (Farnell
Instruments) before being applied to the sample by a single-turn loop resonator situated inside the NMR coil. FC-PEDRI experiments were performed on a phantom
containing a 2 mM aqueous solution of the nitroxide free radical TEMPOL at room
temperature. The EPR irradiation frequency was fixed at 160 MHz and FC-PEDRI
experiments were carried out by irradiating one of the characteristic EPR lines of the
nitroxide triplet which were observed at BE values of 0.0037, 0.005 1, and 0.0072 T;
the intermediate resonance was used for most experiments. Non-field-cycled PEDRI
experiments were also performed at a constant field of 0.0 1 T, in which case the EPR
irradiation frequency of the intermediate resonance was 288 MHz; some of these
experiments used a homebuilt 4 W RF amplifier for the EPR irradiation.
Figure 1 shows a typical FC-PEDRI pulse sequence. For simplicity, we have made
the polarization and detection magnetic field strengths equal in our experiments (BE
= BF = 0.01 T). Table 1 summarizes the observed enhancements obtained with 2
mMTEMPOL samples in the PEDRI experiments at 0.04 and 0.0 1 T and in the FCPEDRI experiments with Bz = 0.005 1 T. Also listed are the power per unit mass
figures for enhancements of E = -5 with aqueous samples. The variation of the applied power with EPR frequency agrees reasonably well with Eq. [ 41.
COMMUNICATIONS 435
TABLE 1
Summary of PEDRI and FC-PEDRI Results
Field EPR
strength frequency
CT) (MHz)
Applied
power
W)
Power per
Power unit mass of
Observed required Volume of sample for
enhancement forE= -5 resonator E= -5”
E W) (ml) W/k)
0.04 1123 0.87 -7.6 0.60 1.5* 400
0.01 288 I.0 -8.9 0.58 36.0' 16.1
0.01 288 4.0 -7.1 2.91 129.0d 22.6
0.0051' 160 I.0 -7.1 0.72 129.0d 5.6
’ Assuming 100% filling factor.
* Single-turn loop resonator; diameter 10 mm, length 20 mm.
’ Single-turn loop resonator; diameter 42 mm, length 26 mm.
d Single-turn loop resonator; diameter 63 mm, length 40 mm.
’ Field-cycled, with fr 9 T,
Figure 2 shows FC-PEDRI images of a resolution phantom; a field-cycled nonPEDRI image is also shown for comparison. The center of the phantom consisted of
five tubes of internal diameters 15, 9, 5, 4, and 3 mm filled with 2 mM TEMPOL
solution. These were enclosed in a cylindrical container of diameter 4 cm which was
filled with water doped with copper sulfate to give the same T1 as that of the free
radical solution (650 ms at 2.5 MHz). Around the outside of the cylinder were attached 14 sample tubes with internal diameters of 8 m m, alternate tubes being filled
with 2 m A4 TEMPOL solution or copper sulfate-doped water. The overall diameter
of the phantom was 6 cm, about the size of a small rat. The three FC-PEDRI images
were obtained using tE values of 750, 1000, and 1500 ms with a TR of 2000 ms, and
the average observed enhancement factors were -4.0, -5.3, and -7.1, respectively.
The instantaneous power level in the EPR irradiation was approximately 7 W/kg,
while the average applied power ranged from 2.7 to 5.3 W/kg depending on the pulse
sequence timing.
In FC-PEDRI the enhanced versus unenhanced image intensity ratio depends not
only on the power of the EPR irradiation but also on the relative timing of the polarization and evolution intervals, and the values of Bi and BF. At the beginning of the
evolution period the size of the magnetization depends on the length of the polarization period compared with the sample’s T, at Bg. During the evolution period the
magnetization decays at a rate determined by the sample’s T, at B,E . Meanwhile the
magnetization in regions of the sample containing free radical increases, again at a
rate depending on T, , toward an equilibrium value which depends on the EPR irradiation power. The sequence timing parameters will therefore influence the detectability of the free radical and must be optimized for each FC-PEDRI experiment, particularly if very low concentrations of free radicals are being sought.
In conclusion we have shown that the sample volume in PEDRI is constrained by
the power available for the EPR irradiation. Lowering the magnetic field strength
reduces the power required to achieve a given enhancement level. FC-PEDRI allows
the applied power to be reduced to levels which are acceptable for biological experi-
436 COMMUNICATIONS
FIG. 2. Images of phantom containing 2 mMTEMPOL solution and copper sulfate-doped water. Images
are 64 X 64 sections of original 128 X 128 images, which had 15 cm field of view, 15 mm slice thickness.
All images are field-cycled, with B: = BF = 0.01 T, BE = 0.005 1 T, saturation-recovery NMR sequence
with ra = 2000 ms, four averages. Image A has tE = 1000 ms, no EPR irradiation. Images B-D have EPR
irradiation at 160 MHz, power 1 W. Image B has fE = 750 ms; observed enhancement factor E = -4.0.
Image C has tE = 1000 ms; E = -5.3. Image D has fr = 1500 ms; E = -7. I.
ments without compromising the SNR of the image. By scaling up the field-cycling
apparatus it should be possible to image free radicals in larger experimental animals,
providing a useful tool for biological and medical research. We believe that this is the
first time that field-cycling has been used in an imaging experiment above the Earth’s
field strength. Field-cycled NMR imaging is itself likely to be of use in studying relaxation in vivo at extremely low magnetic fields.
Original languageEnglish
Pages (from-to)431-437
Number of pages7
JournalJournal of Magnetic Resonance (1969)
Volume84
Issue number2
DOIs
Publication statusPublished - 1 Jan 1989

Cite this