Low-cost low-field NMR and MRI: Instrumentation and applications

Abstract

While NMR and MRI are often thought of as expensive techniques requiring large institutional investment, opportunities for low-cost, low-field NMR and MRI abound. We discuss a number of approaches to performing magnetic resonance experiments with inexpensive, easy to find or build components, aimed at applications in industry, education, and research. Opportunities that aim to make NMR accessible to a broad community are highlighted. We describe and demonstrate some projects from our laboratory, including a new prototype instrument for measurements at frequencies up to ∼200 kHz and demonstrate its application to the study of the rapidly advancing technique known as inhomogeneous magnetization transfer imaging, a promising method for characterizing myelin in vivo.

Keywords: Arduino; FPGA; Inhomogeneous magnetization transfer; Microcontroller; Software-defined radio; Two-photon excitation.

Graphical abstract

Fig. 1

The pyPPM project. (A) shows an Earth’s field NMR spectrum of water following the rapid shut-off of a strong polarization field. A block diagram of the pyPPM spectrometer system is shown in (B). (C) shows a completed pyPPM circuit board and the coil design is shown in (D). Images from the pyPPM project courtesy Bradley Worley, used with permission.

Fig. 2

Low-cost pulsed Earth’s field NMR. (A) Block diagram of the arduino Uno based system. The transmitter and receiver each consist of three op-amps. The polarization coil current is provided from a computer power supply. The addition of gradient coils allows imaging; a representative image of a 500 ml water bottle with an extruded plastic  + shape is shown in the inset. Complete schematics of all the components are supplied in . (B) and (C) Two-photon excitation NMR experiments performed with this Earth’s field instrument. The coil assembly is rotated about the horizontal axis, as shown in (B) and the resulting signal appearing in the coil is shown in C. (A) is reprinted from Ref. , used with permission.

Fig. 3

(Left) A small sensor built with a barrel magnet and integrated rf coil. (Right) CPMG signals obtained using the barrel magnet sensor from a coffee (black) and an espresso (red) drink. The inset is the 6–10th echoes after 2 scans. The echo spacing is 250 $\mathrm{\mu }$s. Reprinted from Ref. , Copyright 2019, with permission from Elsevier. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

(A) and (B) NdFeB H-yoke magnet built for low-cost table top educational imaging system. The 13 kg magnet assembly produces an 0.18 T $B_0$ with 50 ppm homogeneity over a 1 cm diameter spherical region. (C) Photos of the complete tabletop system including rf console and gradient power supplies along with 3D RARE images of a 3D printed “M-I-T” phantom and a mouse brain. See original article for details. Reprinted from Ref. , Copyright 2019, with permission from Elsevier.

Fig. 5

Halbach magnet array for pediatric imaging. (a) The array is constructed from two layers of magnet elements organized in 23 rings. (b) A side view shows the variation in diameter of the rings in order to optimise homogeneity. (c) The final constructed array. Each ring was constructed separately using a PMMA holder fixed together using threaded brass rods. (d) Slices from a T1-weighted 3D image data set of an “avocado in watermelon” head phantom. Reprinted from Ref. , Copyright 2019, with permission from Elsevier.

Fig. 6

The LimeSDR crowd-sourced software-defined radio board as a high performance magnetic resonance transceiver. (a) and (b) show the fid and spectrum of ¹H in CaSO₄$·$$\frac{1}{2}$H₂O, acquired with a spectral width of 50 MHz. The $\sim$100 kHz wide Pake pattern shown in the inset of (b) appears as a sharp peak when the entire spectral width is displayed. (c) Real part of the ESR FID and echo at 149.4 MHz from a degassed trityl radical solution. CW EPR of (d) Lithium phthalocyanine and (e) a 500 $\mathrm{\mu }$M solution of trityl radical. (a) and (b) reprinted from Ref. , Copyright 2018 Wiley Periodicals, Inc, used with permission. (c)–(e) reprinted from Ref. , Copyright 2019 A. Doll. Licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Fig. 7

Arduino Due based low-field NMR spectrometer. (A) shows the circuit diagram for the transmit and receive circuitry not provided by the microcontroller. The 470 pF capacitor in the dashed box can be used to create a series resonance while the transmitter is active, resulting in a significantly higher $B_1$, or removed to create weaker pulses with no ring-up or ring-down. (B) is a block diagram of the complete system. Connections between the Due and the transmit/receive circuit include the DAC output, ADC input, two digital logic lines to control relays, +5 and +3.3 V supplies. (C) is an oscilloscope capture of the rf output at 33.6 kHz in which two phase shifts, of 180 and 90° can be observed. The upper trace comes directly from the microcontroller where the 2 $\mathrm{\mu }$s sample frequency can be observed. The lower trace is the output of the transmitter amplifier which incorporates a low-pass anti-aliasing filter.

Fig. 8

Low-field ihMT experiments. (A) ihMT-CPMG pulse sequence. For single-sided MT experients, $f_1=f_2$. For dual-irradiation experiment $f_1=-f_2$. Experiments here were performed with ${\tau }_1$ = 2 ms, ${\tau }_2$ = 1 ms and ${\tau }_3$ = 5 ms. (B) NMR echo data from the Due-based spectrometer at 63.6 kHz. The top trace shows a train of 40 echoes with 32 ms echo spacing. The lower trace shows the Fourier transform of the echo train. (C) The signal intensity of a hair conditioner sample following magnetization transfer prepulses ($f_1=f_2$), dual-frequency prepulses ($f_1=-f_2$) and ihMTR. All signals normalized for direct excitation of the water signal. The signal suppression of the MT data were fitted to a Lorentzian lineshape, eMT data were fitted to a Gaussian lineshape and ihMT to the difference. (D) shows MT, eMT and ihMT from a lamellar lipid sample at 500 MHz. (D) reprinted from Ref. , Copyright 2016, Wiley Periodicals, Inc, used with permission.