Advancements in reducing field strength are bringing MRI closer to populations unserved by the technology.
The confluence of deep learning MRI reconstruction, improved electronics, COVID-19, and recently low field strength MRI systems cleared by the U.S. Food & Drug Administration has created an “inflection point” for increased use of very-low (<0.2 T) field strength (VLFS) magnetic resonance imaging (MRI).
What is most impactful about portable, VLFS MRI is not just the reduced cost, ease of siting, reduced safety risk, and reduction of artifacts associated with higher field strengths, but the ability to substantially expand access of MRI to a much wider variety of clinical indications and environments, as well as bring it to areas of the world where MRI is prohibitively expensive or impractical. This will allow the specialty of diagnostic radiology to fundamentally re-imagine how MRI is used. For example, the American College of Radiology (ACR) appropriateness criteria for MRI could be considerably expanded with the introduction of VLFS MRI systems to operating rooms, intensive care units (ICUs), emergency departments (EDs), and in emergency transport vehicles.1
In order to make the most positive impact on patient care, radiologists and clinicians who understand the advantages and limitations of emerging VLFS systems will need to work together closely to plan, test, and practically implement new indications and protocols for MRI. Potential indications could include use for rapid intra-operative evaluation during cervical spine and other surgical procedures, patient monitoring in a neuro or other ICU, screening of trauma and other patients for cord contusion in the emergency department, rapid head scanning for stroke/trauma in an emergency transport vehicle, or even a quick and comparably expensive MRI scan as a substitute or adjunct for a subset of orthopedic conventional radiography studies.
The phrase “low field strength MRI” typically conjures up antiquated MRI scanners with noisy, fuzzy images that are limited to a subset of orthopedic and a few other applications. The general perception, at least in the United States, is that 1.5T MRI scanners are the minimum required for most applications and that 3T MRI systems are significantly better, especially for more demanding applications, such as neuroimaging and other more complex sequences. This perception is reinforced by insurance payers. While the ACR does not specify a lower limit for MRI field strength (although it does specify a “whole body” scanner), a subset of payers will not reimburse for studies performed on scanners with a field strength below 0.3T.2
Currently, in the United States, commercially available whole-body lower field strength MRI is mostly limited to a single open MRI scanner that operates at 1.2T with specialized MRI scanners below 1.0T marketed for very few and specific clinical indications, such as hand imaging, upright/weight-bearing extremity imaging, for real-time guidance in radiation oncology. High-field MRI (1.5T or higher) represents about 85 percent of the market size in Europe and North America.3
Around the world, the spectrum of field strength used for MRI scanners varies considerably. For example, in Europe and North America, 15 percent-to-18 percent of existing MRI systems are 3T and approximately 66 percent utilize 1.5T units. In the United States, 1.5T and 3T scanners are being purchased in approximately equal numbers and dominate new sales. In China, on the other hand, the percentages are substantially different with 8 percent of systems at 3T, 45 percent at 1.5T and the remaining 50 percent lower than 1.5T.4 These commercial MRI systems are relatively expensive to purchase and site at any field strength; throughout the world, consequently, only 10 percent of the population has access to MRI.5
Despite this general impression, low-field strength MRI actually does not necessarily mean low resolution or poor image quality. Much of this perception has been due to the strong trend toward higher field strength for new MRI scanners since the 1980s and, thus, is confounded as a comparison of new versus much older technology.However, the 1.5T accepted basic standard for field strength has not changed in the past 30 years, while the quality of MRI images has improved dramatically, underscoring that field strength is only one of many factors in today’s high standard of image quality.
The major limitation of low-field strength is reduced-bulk magnetization of nuclear spins, resulting in reduced signal compared to noise. This can be compensated for in many ways, but the most basic is signal averaging which increases the signal to noise by the square root of the number of averages. Unfortunately, this currently means an impractically long scan time for the current generation of low-field strength systems to maintain comparable signal-to-noise ratios.
The COVID-19 pandemic has accelerated the trend toward distributed imaging in diagnostic radiology. One of the best examples of this can be seen in the preference for performing portable chest radiography in the emergency department during the pandemic rather than PA and lateral chest radiographs in radiology. Similarly, VLFS MRI scanners located in clinical areas, such as the ICU and ED could offer the ability to minimize flow through a medical center of potentially infectious patients or those vulnerable to infection.
The fundamental driver of the “tipping point” in VLFS systems has been the recent application of artificial intelligence (AI)/Deep Learning in image reconstruction. Essentially, this involves the creation of a library of the same MRI images acquired as low noise, as well as high noise versions. An AI algorithm then “learns” how to predict the appearance of a low noise dataset based on the input of a noisier one. This can be accomplished in a variety of different ways using many different mathematical and computer vision-based techniques. This creation of higher quality images can theoretically be used to substantially reduce acquisition time.
For more coverage based on industry expert insights and research, subscribe to the Diagnostic Imaging e-Newsletter here.
For example, if major reductions, such as a 256-fold decrease in acquisition time for a particular sequence could theoretically be achieved, it could potentially allow for a 16-fold reduction in magnetic field strength while preserving overall image quality. This could, in this example, allow a 0.094T scanner to provide images with a comparable signal-to-noise-to-current 1.5T systems with similar scanning times. We see a variant of this hypothetical example being implemented with the use of Deep Generative adversarial neural networks (GANs) for compressive sensing, which can result in much shorter image acquisition time. Extensive experiments that were based on a large cohort of abdominal MRI data with evaluations provided by expert radiologists confirm that these GANs are able to retrieve images with substantial diagnostic quality improvement with up to 300-times faster speeds than state-of-the-art CS-MRI toolboxes.6 These image acquisition advancements hold promise for low-field strength imaging.
Low-Field Strength MRI
Reduced-field strength improves patient safety and comfort, including improved compliance with a more open system, reduced projectile risk, and the possibility to image metallic implants at a closer distance due to reduced magnetic susceptibility artifacts. There is also a substantial reduction in acoustic noise due to the reduced force on the gradient coil windings.7
There is ongoing research into the use of specialized contrast agents, such as super-paramagnetic iron oxide nanoparticles (SPIONs) for low-field MRI. These have the potential to substantially decrease image quality differences between VLFS and higher-field strength MRI. Ferumoxytol, for example, shortens T1, T2, and particularly T2* relaxation rates, and when combined with a high-efficiency balanced steady-state free precession sequence, results in high-quality images with low-field MRI. It also offers a prolonged blood pool phase and delayed intracellular intake, making it ideal for use as a contrast agent when moving towards lower field strength.9
Coffey, et. al., documented both theoretically and experimentally, the ability of a VLFS MRI scanner at 0.0475 to achieve a higher signal-to-noise ratio than a 4.7T scanner using non-equilibrium hyper-polarization schemes and frequency-optimized MRI detection coils.10 This suggests that by using this approach with hyper-polarized contrast agents, such as hyper-polarized noble gases for lung scanners or 13C-labeled metabolites, VLFS can provide extraordinarily high (and superior to high-field strength MRI) signal for diagnosis.
Even without deep learning enhancement or hyper-polarized or other contrast agents, multiple studies have documented the advantages of intra-operative low-field strength systems. A 2016 study deployed a 0.15T MRI system for assisting cranial surgeries. It was found to be a useful diagnostic tool that did not impact anesthetic and surgery times.11 Another study demonstrated the use of intra-operative MRI at 0.15T in patients undergoing tumor resection.12 More than 103 procedures were performed with the use of the mobile ultra-low field MRI unit. The intra-operative imaging resulted in continued tumor resection in 30 percent of the patients with glioblastoma due to unexpected residual tumors.
Companies that previously spearheaded the “high-field MR revolution” are now exploring lower-field strength technology. Siemens’ Magnetom FreeMax with a field strength of 0.55T when compared with 1.5T, this low-field MRI system demonstrated increased safety for patients with metallic interventional devices, which allow an MRI-guided right heart catheterization in patients with metallic guide wires. This system was also found to reduce image distortion in the upper airway, cranial sinuses, lungs, and the bowel. 13
At this time, the only commercial, portable MRI system that has received 510(k) clearance by FDA as a portable ultra-low field imaging device is available from Hyperfine Research, Inc. This system can be wheeled to a patient’s bedside, plugs into a standard wall outlet, and can be operated using a wireless tablet. The system has a field strength of 0.064T, utilizes a permanent magnet, and weighs approximately 1,400 pounds.
The system, thus, has the potential to be sited in the ICU, ED, or in orthopedic or other clinics and, potentially, in a mobile vehicle, such as an ambulance. Initial clinical testing has been performed at Yale-New Haven Hospital, University of Pennsylvania Hospital, New York-Presbyterian Brooklyn Methodist Hospital, and Good Samaritan Hospital Medical Center where the scanner has been used for brain imaging.
In many ways analogous to the emergence of ultra-portable ultrasound systems, portable and distributed MRI systems will challenge our current thinking on MRI scan reimbursement, accreditation, utilization and appropriateness criteria, “turf” issues, and credentialing issues.
There will be ongoing questions about whether the studies should be acquired by non-MRI, sub-specialty technologists and whether they will be interpreted by radiologists, as well as the potential for substantial over-utilization and the potential for a falsely negative study resulting in a more comprehensive, higher field strength MRI scan not being performed. On the other hand, the potential for a major expansion of MRI into new hospitals and outpatient locations and even onto emergency vehicles for point-of-care service, into new indications for cost-effective yet higher accuracy patient evaluation, and as a more sensitive indicator of disease than CT or ultrasound or conventional radiography is exciting.
Radiologists and our clinical colleagues have demonstrated incredible creativity and innovation in the application of MRI to an extraordinary number of anatomic, vascular, functional, and interventional applications. Portable, ultra-low field MRI will undoubtedly add a whole new dimension to this potpourri of patient care options and will eventually bring MRI to that 90 percent of the world’s population that has been waiting for it to finally arrive.
Burham A. Khan, M.D., is currently an intern with Hyperfine, and Eliot Siegel, M.D., FACR, is also a professor and Vice Chair of Information Systems at the University of Maryland School of Medicine.