Agricultural MRI
ROOTS: This work, funded by ARPA-E, was a collaboration between Texas A&M, Harvard University, Soil Health Institute, NIST and ABQMR. The goal was to image plant roots through natural soils both in the agricultural field and in the greenhouse. There are relatively few breeding studies on plant roots because gathering information is extremely labor intensive. Three generations of MRI systems were built by ABQMR and deployed at Texas A&M and a fourth generation was conceptualized for above ground imaging in the field. In addition to imaging roots, proof-of-concept studies were done to show that the third-generation device could also produce high resolution images of washed roots and above ground parts of the plant.
It was known that the team would need to work at low magnetic field due to the inherent magnetism in natural soils. However, it was unclear what magnetic field strength would be optimal. From first to third generation, the field strength increased from 0.05 T (2 MHz) to 0.15 T (6.5 MHz) and finally 0.47 T (20 MHz). Through experiment, we have learned that there is not a perfect field strength for all options (it is somewhat soil dependent), but most soils work well up to 0.47 T. This means that there are minimal distortions in the images caused by magnetic particles in the soil.
First generation:
This system was designed for use in the agricultural field. The magnet was an end corrected solenoid wound from 16 gauge copper wire. For cooling, the magnet was submerged in a PVC bucket of circulated oil which dissipated heat via a chilled water loop of copper tubing. The magnet dissipated 2 kW of heat when energized to 0.05 T. This was the maximum field we could achieve while sufficiently cooling the magnet to avoid damage. This magnet had lifting hooks on top of the magnet which allowed the system to be moved by a tractor and lowered into the ground around a root bundle for imaging. In the bore of the magnet were three-axis gradient coils and an RF coil.
The root bundle was stabilized by hydraulically pressing a 25 cm outer diameter PVC pipe into the ground around the plant. A larger, sheet-steel cylinder was then driven into the ground centered around the PVC pipe. When removed, the annular region of soil would come with the steel cylinder. The MRI system could be lowered into this hole.
The system images a root bundle of approximately 25 cm in diameter and height. At 0.05 T (2 MHz), the signal to noise is very low. This means that acquiring a true three-dimensional image takes on the order of 10-15 hours. Instead, two-dimensional projections are acquired at 22.5° increments. When shown sequentially, the 3D architecture of the roots can be visualized. This full image set takes approximately 2.5 hours to acquire.
Second generation:
This system was born out of the need for more signal (i.e. faster images). Moving to higher magnetic field strength gives an increase in signal which means shorter imaging times. This system used a high temperature superconducting magnet built by HTS-110 in New Zealand. This magnet was cooled by liquid nitrogen and could be energized to 0.15 T (6.5 MHz). The magnet had lifting hooks at the top and could, similar to the first generation, be carried by a tractor for field use.
However, this magnet found it’s home mostly in the greenhouse scanning roots of potted plants. Images cover a 25 cm diameter x 25 cm tall region of a potted plant. The plants were typically grown in PVC pipe (which does not need to be removed for imaging) and scans can be taken sequentially along the pipe to see roots over a length more than the 25 cm FOV. A three-dimensional image of a root bundle in natural soil could be acquired in 1-3 hours depending on the desired resolution. An image with 2 mm per side on a voxel takes approximately 1 hour. An image with 1 mm per side on a voxel takes approximately 3 hours.
Third generation:
When the plethora of greenhouse applications became clear, ABQMR started on the third generation MRI system which was specifically designed for the greenhouse. Without the need for mobility in the field, the third generation is a much more user-friendly system. The closed cycle superconducting magnet was built by Superconducting Systems, Inc. This magnet can be energized up to a maximum of 1.98 T (85 MHz) but is routinely run at 0.47 T (20 MHz) to image root bundles up to 20 cm in diameter x 20 cm tall. Imaging times and resolutions are similar to the 6.5 MHz system but the images have less noise.
Above ground device:
Though the first and second-generation devices are field compatible, it is still necessary to remove soil from around the root bundle to lower the magnet into the ground. Of course, something that could image the roots from above the ground would be an ideal solution. This phase of the project was slated to use a pancake magnet (also built by Superconducting Systems, Inc.) which would sit on top of the ground and project an imaging region 20 cm in diameter centered at 12 cm below the ground. Gradient coils and an RF coil were designed but not fabricated.
The first question was whether it was possible to build such a magnet. SSI was successful in building the magnet which is designed to be transported in the field and operates at 0.04 T (1.8 MHz), just above the AM radio band. The magnet is 2.1 meters in diameter and 2000 lbs!
High resolution imaging:
Though the third generation MRI system is used at 0.47 T (20 MHz), since it is a superconducting magnet, it is capable of reaching much larger field strengths. With higher field strength comes more signal – and more distortions if magnetic soils are present in the system. When run at full field of 1.98 T (85 MHz), the samples must be washed of soil, but images with resolutions down to 30 um in-plane can be achieved.
Soil water imaging:
One of the benefits of MRI is that the imaging is very tunable. All of the images shown and discussed above were tuned to image the root water and ignore any signal from the much larger quantity of soil water present. However, it is also possible to acquire images of soil water plus root water to see heterogeneity in soil water distribution. The images here are slices out of a 3D image showing the top region of a soil column with a sorghum plant. The soil column was sealed to prevent evaporation and drainage. Over 10 days, the soil water was imaged as the plant transpired water.
Ka Man Jasmine Yu, Joel Oliver, Brian McKinley, Brock Weers, Hilary T. Fabich, Nathan Evetts, Mark S. Conradi, Stephen A. Altobelli, Amy Marshall-Colon, John Mullet, "Bioenergy sorghum stem growth regulation: intercalary meristem localization, development, and gene regulatory network analysis", the Plant Journal, 112(2), 2022
G. Cody Bagnall, Neha Koonjoo, Stephen A. Altobelli, Mark S. Conradi, Eiichi Fukushima, Dean O. Kuethe, John E. Mullet, Haly Neely, William L. Rooney, Karl F. Stupic, Brock Weers, Bo Zhu, Matthew S. Rosen, Cristine L. S. Morgan, "Low-field magnetic resonance imaging of roots in intact clayey and silty soils," Geoderma 370, 114356 (1 July 2020).