Ultra-low Field MRI Research Articles

MRI with an atomic magnetometer suitable for practical imaging applications

Conventionally implemented MRI is performed in a strong magnetic field, typically >1 T. The high fields, however, can lead to many limitations. To overcome these limitations, ultra-low field (ULF) [or microtesla] MRI systems have been proposed and implemented. To-date such systems rely on low-Tc Superconducting Quantum Interference Devices (SQUIDs) leading to the requirement of cryogens. In this letter, we report ULF–MRI obtained with a non-cryogenic atomic magnetometer. This demonstration creates opportunities for developing inexpensive and widely applicable MRI scanners.

Microscopic investigation of the resonant mechanism for the implementation of nc-MRI at ultra-low field MRI

In this paper the possible use of the resonant mechanism between some spectral components of the neuronal activity and the spin dynamics in ultra-low field MRI experiments – for the implementation of the nc-MRI techniques and proposed by Kraus et al., 2008 – is investigated by means of “realistic” simulations of the neuronal activity of a modelled neuronal network. Previously characterized digital neurons are used to reproduce neuronal currents based on biophysical details and the distribution of the local magnetic field inside a MRI cubic voxel (having a dimension of 1.2 mm) is evaluated. The properties of the water proton spin dynamics as a consequence of the neuronal field and of external applied fields are extrapolated integrating the Bloch equations. The characteristics of the expected MR signals are discussed in relation to the specifics of the NMR sequence used and to the properties of the neuronal activity. The great potentialities of the technique are provided by: a) the possible easy implementation of the technique, b) the possible cheap instrumentation required; c) the flexibility of the ultra-low field systems.

SQUID-based instrumentation for ultralow-field MRI

Magnetic resonance imaging at ultralow fields (ULF MRI) is a promising new imagingmethod that uses SQUID sensors to measure the spatially encoded precession ofpre-polarized nuclear spin populations at a microtesla-range measurement field. In thiswork, a seven-channel SQUID system designed for simultaneous 3D ULF MRIand magnetoencephalography (MEG) is described. The system includes sevensecond-order SQUID gradiometers characterized by magnetic field resolutions of1.2–2.8 fT Hz−1/2. It is also equipped with five sets of coils for 3D Fourier imaging with pre-polarization.Essential technical details of the design are discussed. The system’s ULF MRI performanceis demonstrated by multi-channel 3D images of a preserved sheep brain acquired at46 µT measurement field with pre-polarization at 40 mT. The imaging resolution is2.5 mm × 2.5 mm × 5 mm. The ULF MRI images are compared to images of the same brain acquiredusing conventional high-field MRI. Different ways to improve imaging SNR arediscussed.

Toward SQUID-Based Direct Measurement of Neural Currents by Nuclear Magnetic Resonance

Modern high field (HF) MRI uses magnetic fields greater than 1.5 T to yield exquisite anatomical features. We have also seen an explosion in functional MRI in the last decade that measures hemodynamic responses that are ultimately sluggish (~one sec) and only indirectly related to electrophysiological processes. Magnetoencephalography (MEG) is a direct measure of the external fields generated by neuronal currents with exquisite temporal information (less than one msec). Spatial localization, however, is inferred from modeling priors, making MEG ldquoimagingrdquo only indirect at best. Ultra low field (ULF) MRI has recently been demonstrated with 2-3 mm resolution using fields in the microtesla regime. While the nuclear magnetic resonance (NMR) signal at ULF is dramatically weaker than at HF, we acquired high signal-to-noise measurements for a variety of samples at ULF using SQUID technology. Several researchers have proposed that electrophysiological activity may interact with the nuclear spins in a volume of interest, causing measurable variations in the NMR signal. We have developed a new approach to directly measuring neuronal activity with SQUID-based ULF-NMR techniques based on the hypothesis that interactions between the spin population and neural activity in cortex can be dominated by resonant mechanisms unique to ULF. We have experimentally demonstrated the feasibility of this approach via ULF-NMR using a single-channel SQUID system.

Breast tumor imaging with ultra low field MRI.

Ultra low field MRI was evaluated for breast tumor imaging. Twenty-seven patients with a known or suspected breast malignancy were imaged preoperatively with a T2-weighted gradient-echo sequence. Additional IR- and gradient-echo sequences were used for measurements of T1 and T2 relaxation times. Seventeen resected specimens were also imaged within four hours after removal. Nineteen of 23 carcinomas detected at mammography were also detected with MRI (sensitivity 83%), and one of four benign cases was evaluated as pathologic (specificity 75%). Most of the tumors had T1 and T2 relaxation times longer than those of normal breast parenchyma. The T1 values of the tumors were significantly different from the T1 values of fat (p = .0002) and breast parenchyma (p = .0151). The tumor T2 values differed significantly from the T2 of breast parenchyma (p = .0223) but not from that of fat (p = .1475). In vitro tumor measurements were similar to the in vivo measurements. Despite these statistically significant findings, the calculated relaxation times were not considered sufficiently reliable for the clinical evaluation of breast tumors.

Magnetic resonance imaging of the uterus at an ultra low (0.02 T) magnetic field

In vivo pelvic imaging of 39 women and in vitro relaxation time measurements of four uterine specimens were performed using an ultra low field (0.02 T) MRI unit. Average T 1 times measured in vitro at 37°C for the myometrium and endometrium were 206 ms ( SD 47 ms) and 389 ms ( SD 21 ms), respectively. Corresponding T 2 times were 95 ms ( SD 20 ms) and 167 ms ( SD 13 ms). The proton relaxation of almost all myometrial specimens proved to be biexponential, but of all endometrial specimens was monoexponential. Contrast measurements between endometrium versus myometrium and myometrium versus the junctional zone were performed after imaging 18 volunteer women using different pulse sequence parameters. Normal uterine structures were optimally demonstrated by SE 700 70 . Relatively short repetition times could be used, because spin-lattice relaxation times were short at the low magnetic field. Consequently, the short repetition times allowed averaging of four excitations to create adequate images within an acceptable scanning time. In addition to T 2-weighted images a T 1-weighted inversion recovery sequence with a short inversion time of 50 ms (IR 1000/50/40) adequately differentiated the three uterine zones. Although pathologic lesions of the uterus including leiomyomas, anomalies and carcinomas were well demonstrated, especially with the T 2-weighted spin echo pulse sequence, further investigations are needed to evaluate the optimal technique for ultra low field MR imaging of uterine tumors.

Examination of infant brain maturation using ultra low field MRI.

The brains of 42 newborn infants were examined with MRI at 0.02 T field, and regional variations of T1 relaxation time were measured from the images. There were three groups: 1. full term infants (9), 2. preterm infants (10) and 3. SGA (= small for gestational age) infants (20). Relaxation times showed a correlation to myelination of the brain. The brain of SGA infants showed a large variation in their T1 values.