SQUID-Detected In Vivo MRI at Microtesla Magnetic Fields

SQUID-Detected in Vivo MRI at Microtesla Magnetic Fields Michael Mosle, Whittier R. Myers, SeungKyun Lee, Nathan Kelso, Michael Hatridge, Alexander Pines and John Clarke the spins in the sample and to achieve a spin precession frequency of tens to hundreds of MHz. The spins precess at a frequency f 0 = γ B 0 /2π , and the resulting signal is detected by Faraday induction; γ is the gyromagnetic ratio. For 1 H, γ /2π = 42.6 MHz/T. Because the signal from Faraday induction scales as the precession frequency and the spin polarization increases linearly with B 0 , in conventional MRI the detected signal scales as B 0 2 . Thus there is great incentive to operate at high magnetic fields, and B 0 typically ranges from 1-4 T in clinical imaging. Since the early 1990s, however, an alternative approach of polarizing the spins with a pulsed field B p and detecting them in a much lower field has attracted interest [3], [4]. This approach has a number of advantages. First, because resolution limitations caused by field inhomogeneity scale with the magnitude of B 0 , the relative homogeneity requirement of B 0 is much relaxed by reducing B 0 . Since B p does not require high homogeneity, one can use a weak, moderately homogenous B 0 field and a strong, inhomogeneous B p field; both fields can be generated from copper wire coils and do not require shimming. Second, when one encodes images in a low field (< 0.1 T) artifacts caused by susceptibility changes in the sample and by chemical shift are largely eliminated. Finally, in contrast to the case for superconducting magnets, the field from resistive magnets can be easily varied, enabling the measurement of NMR properties of the sample over a wide range of fields. In particular, image contrast caused by differences in the spin-lattice relaxation time (T 1 ) is often enhanced at low magnetic fields [5], [6]. To avoid the sensitivity loss inherent in low-frequency Faraday detection, one can use a SQUID detector [7]-[9]. Coupled to a superconducting flux transformer, a SQUID can have a field sensitivity of ~1 fT Hz -1/2 at frequencies ranging from several megahertz down to a few hertz [10]. Combining SQUID detection with prepolarization allows frequency- independent MRI detection. We used this combined approach of SQUID detection and prepolarization with low-field encoding in our earlier work [11, 12] to acquire two-dimensional images of water and oil phantoms and of vegetables with 2-4 mm spatial resolution in an encoding field of 132 μT. In this paper, we employ a three-dimensional phase-encoded pulse sequence to produce three-dimensional in vivo images of human arms and fingers in microtesla fields. B B B B B B B B B B B Abstract—We use a low transition temperature (T c ) Super- conducting Quantum Interference Device (SQUID) to perform in vivo magnetic resonance imaging (MRI) at magnetic fields around 100 microtesla, corresponding to proton Larmor frequencies of about 5 kHz. In such low fields, broadening of the nuclear magnetic resonance lines due to inhomogeneous magnetic fields and susceptibility variations of the sample are minimized, enabling us to obtain high quality images. To reduce environmental noise the signal is detected by a second-order gradiometer, coupled to the SQUID, and the experiment is surrounded by a 3-mm thick Al shield. To increase the signal-to- noise ratio (SNR), we prepolarize the samples in a field up to 100 mT. Three-dimensional images are acquired in less than 6 minutes with a standard spin-echo phase-encoding sequence. Using encoding gradients of ~100 μT/m we obtain three- dimensional images of bell peppers with a resolution of 2 x 2 x 8 mm 3 . Our system is ideally suited to acquiring images of small, peripheral parts of the human body such as hands and arms. In vivo images of an arm, acquired at 132 μT, show 24-mm sections of the forearm with a resolution of 3 x 3 mm 2 and a SNR of 10. We discuss possible applications of MRI at these low magnetic fields. Index Terms—In vivo images, magnetic resonance imaging, nuclear magnetic resonance, SQUID I I. I NTRODUCTION n magnetic resonance imaging (MRI), a spatial map of the nuclear magnetic resonance (NMR)-active nuclei is obtained by detecting the magnetic signal from nuclear spins precessing in the presence of a magnetic field gradient [1], [2]. In addition to the image encoding gradients, traditional MRI utilizes a strong, uniform field B 0 for two purposes: to polarize B Manuscript received October 4, 2004. This work was supported by the Director, Office of Science, Office of Basic Energy Sciences, Division of Materials Sciences and Engineering of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098. M. Mosle is with the Physics Department, University of California and the Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720 USA (phone: 510-642-3634; fax: 510-642-1304; e-mail: mjmoessle@lbl.gov). W. R. Myers (email: wmyers@socrates.berkeley.edu), S-K. Lee (email: lsk@socrates.berkeley.edu), N. Kelso (email: kelso@socrates.berkeley.edu), M. Hatridge (email: hatridge@berkeley.edu) and John Clarke (email: jclarke@socrates.berkeley.edu) are with UC Berkeley and LBNL, Berkeley, CA 94710, USA. A. Pines is with the Chemistry Department, UC Berkeley and the Materials Sciences Division, LBNL, Berkeley, CA 94720, USA (email: pines@berkeley.edu).