Crop Root MRIs Open New Doors in Agriculture
A team of scientists led by Texas A&M AgriLife is taking a page from the medical imaging world and using MRI to examine crop roots in a quest to develop crops with stronger and deeper root systems.
The team from Texas A&M AgriLife Research, Harvard Medical School, ABQMR Inc. and Soil Health Institute developed a novel MRI-based root phenotyping system to nondestructively acquire high-resolution images of plant roots growing in soil and established the Texas A&M Roots Lab to further develop this technology as a new tool for assessing crop root traits.
The “Field-Deployable Magnetic Resonance Imaging Rhizotron for Modeling and Enhancing Root Growth and Biogeochemical Function” is a part of the Rhizosphere Observations Optimizing Terrestrial Sequestration, ROOTS, program funded through U.S. Department of Energy’s Advanced Research Projects Agency-Energy program.
Nithya Rajan, Ph.D., AgriLife Research crop physiologist/agroecologist in the College of Agriculture and Life Sciences Department of Soil and Crop Sciences, Bryan-College Station, is leading this multidisciplinary project team.
“We are applying this technology to see if we can sense roots growing in agricultural soils and characterize them,” she said. “To date, imaging roots in soil has been challenging because the soil is complex, with solids, moisture and roots. We just want to image the roots.”
From concept to applications, in sorghum and beyond
The project was initially funded for three years with a $4.6 million grant. The second phase of funding was approved this year at $4.4 million.
“In the first phase, we developed the proof of concept and initial prototypes, and in the second phase we developed a low-field MRI rhizotron for high throughput imaging and applications in a wide variety of crops in addition to energy sorghum,” Rajan said.
Also on the team with AgriLife Research are Bill Rooney, Ph.D., sorghum breeder and Borlaug-Monsanto Chair for Plant Breeding and International Crop Improvement in the Department of Soil and Crop Sciences, and John Mullet, Ph.D., biochemist and Perry L. Adkisson Chair in Agricultural Biology in the Department of Biochemistry and Biophysics.
Rooney and Mullet are using the MRI system to advance bioenergy sorghum genetics. Brock Weers, Ph.D., and Will Wheeler, Ph.D., are support scientists working with the AgriLife Research team.
“We need to develop crop root systems that store more carbon in soil,” Mullet said. “In addition, deeper root systems can take up more water from soil profiles, increasing crop drought resilience.”
From a crop improvement perspective, Rooney added, this technology is essential to effectively screen crop germplasm for specific genotypes with enhanced root systems.
Getting to the root of the matter, without disturbing the soil
Using MRI allows researchers to gather root images without damaging plants, unlike traditional methods such as trenching, soil coring and root excavation, Rajan said.
The AgriLife Research team is working with ABQMR Inc., a group of MRI scientists in Albuquerque, New Mexico, who are involved in designing and building the system.
“With low magnetic fields, MRI can be used to image roots in natural soils,” said Hilary Fabich, Ph.D., president of ABQMR. “The low magnetic fields also mean there is less of a safety risk working with the sensor in an agricultural setting.”
Using “machine learning” to see through the noise
Matt Rosen, Ph.D., is the co-principal investigator of the project. He is director of the Low-field MRI and Hyperpolarized Media Laboratory and co-director of the Center for Machine Learning at the Martinos Center for Biomedical Imaging at Harvard. Rosen and his team bring their experience with both low-field MRI physics and state-of-the-art artificial intelligence techniques to the project.
The Rosen lab pioneered the use of deep learning for processing MRI data. Neha Koonjoo, Ph.D., a postdoctoral fellow in the Rosen lab, has been leveraging the AUTOMAP — Automated TransfOrm by Manifold Approximation — deep learning-based image reconstruction approach to reduce the influence of environmental noise in root MRI images. Her approach was described in a recent research article.
Bragi Sveinsson, Ph.D., a postdoctoral fellow working with Rosen, developed the first prototype of a software named “MIDGARD” — MRI 3D seGmentation and Analysis for Root Description — for extracting quantitative root trait information from MRI images of roots.
The team plans to release MIDGARD as an open-source software after further testing.
“Using MIDGARD, we can extract quantitative root trait information, and this data will be used for selection of ideal plant characteristics,” Rosen said. “In the future, MIDGARD will also have the ability to three-dimensionally image soil water content, a key property that drives root growth and exploration.”
Technology to market
Technology-to-market activities of this project are led by Cristine Morgan, Ph.D., chief scientific officer of Soil Health Institute, Research Triangle Park, North Carolina, and principal investigator of the first phase of the project when she was at Texas A&M. To foster collaborations with industry partners, the Soil Health Institute established the company Intact Data Services.
“I am excited to translate this technology for phenotyping at scale, as well as the ability to use MRI to 3D-image soil water intact,” Morgan said.
Reference: Koonjoo N, Zhu B, Bagnall GC, Bhutto D, Rosen MS. Boosting the signal-to-noise of low-field MRI with deep learning image reconstruction. Sci Rep. 2021;11(1):8248. doi: 10.1038/s41598-021-87482-7
This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source.