Low field MRI has seen a recent renaissance in interest as a novel and low-cost method of MRI for lung, brain, and other organ systems. Low field MRI may play a role in identifying and understanding damage to the lung caused by COVID-19. Today we talk with Matt Rosen, Director of the Low Field MRI and Hyperpolarized Imaging Laboratory at the Athinoula A. Martinos Center for Biomedical Imaging. He has extensive experience in milli-tesla MRI instrumentation as well as in free radical imaging enabled by operation at low field. We discuss the potential for low field MRI to contribute to both clinical and research efforts surrounding COVID-19.

Cameron Cushing (CC): What we are trying to do is get an understanding of what contributions MR imaging might be able to make to the current COVID-19 pandemic.

What do you see as important information to gather as we go forward in terms of COVID response – Lung anatomy, function? Short term diagnosis and long term follow ups would seem to have different goals.

Matt Rosen (MR): It is interesting how we are hearing a lot now about how what early on was thought of mostly as an acute respiratory disease is more of a systemic disease. The so-called cytokine storm associated with the immune response seems to be the real problem.

It reminds me a little of what I know about reperfusion injury in stroke, and similarly of so-called secondary injury in traumatic brain injury. The classic presentation of say military TBI is that you are concussed, and during that impact or explosive blast there is neuronal shearing and focal injury in the brain. Once the force stops, that primary injury stops. In the minutes to hours to days following the initial injury, what was focal damage can turn into global injury. The general thinking is that an overproduction of reactive oxygen species (ROS) or free radicals at the site of primary injury lead to global edema and cell death.

In these cases, if one could somehow obtain a tomographic map of the quantity and type of ROS that are present, you could imagine titrating in a therapeutic agent until the ROS was at the appropriate level. This is the story I have been telling for stroke and traumatic brain injury, where the therapeutic might be an antioxidant or some other free radical scavenger. Conventional MRI has no ability to see these ROS or radicals which may be present at millimolar or smaller concentrations in vivo. That is why we have been using Overhauser DNP at low field to generate hyperpolarization in vivo with the source of the hyperpolarization coming from aqueous free radicals. (Sarracanie 2014, Waddington 2018). In the work we have done, we have used exogenous free radicals as a tracer – sort of a proof-of-concept. The amazing thing is that by operating at low magnetic field, you can access the electron resonance in vivo; at 6.5 mT its around 140 MHz (and our 1H frequency is way down at 276 kHz!)

CC: How about pre-clinical imaging, are there any potential opportunities there for studying COVID-19?

MR: Is there an animal model of COVID-19? As a physicist I am not up to date on that, but it would be very interesting to think about if we could somehow understand that sector of the immune response that is modulated by free radicals.

Some work that we are doing in collaboration with Thomas Theis at North Carolina State University may also be useful in preclinical studies. We have been using SABRE, a parahydrogen mediated hyperpolarization approach. We have been showing that SABRE is very unique in its ability to hyperpolarize; not just helium or xenon or pyruvate or one specific thing like you can do with other techniques that I have been working with for the last 25+ years. SABRE is a very flexible formalism and you can polarize broad classes of molecules — carbon compounds, nitrogen compounds, proton compounds, you name it! (Shchepin 2019) SABRE also is maximally efficient at 6.5 mT, where happens to be where we work in my lab. SABRE really unlocks the ability to do molecular imaging with MRI, and doubtless this will be important as a tool to understand COVID19 processes in vivo.

CC: For someone thinking about clinical MR where the minimum field strength is usually at 1.5T – is there any relevance at that field strength to some of these techniques you have been talking about?

 MR: I think doing free radical contrast via in vivo Overhauser DNP like we do in lab but at 1.5 T is not going to happen because that is requires something like a 50 gigahertz electron saturation field – you are going to microwave that person! The techniques I am talking about with SABRE use injected hyperpolarized tracer compounds. You can do that kind of hyperpolarized molecular imaging at any field strength. You just have to find the right pathway that you want to study, and find an agent that will target the process, and make sure it can be hyperpolarized. If someone has a conjecture like, “oh it is this particular metabolite and it is suppressed or it is enhanced from COVID”, one can look at the dynamics of a hyperpolarized version of that metabolite.

As you know, despite its clinical successes, MRI is incredibly insensitive. We image protons at 110 molar, but try looking at things in the sub-micromolar concentration! Modalities like PET have us beat in terms of absolute sensitivity since you can count single photons! So this is why those have been the techniques of choice for molecular imaging. However, adding hyperpolarization to MRI is going to be the thing that changes all this! It will allow MR images to have the sensitivity and specificity to understand the etiology of dynamic disease processes. I am pretty encouraged by that even though I do not know what molecular target or targets are of interest right now for COVID19.

Paul Weavers: For this we need to find someone [a researcher] who says I have [a specific] target and this labeling compound?

MR: Exactly! What we are going to say is we have this technology to broadly enhance a signal from large classes of molecules. What would be your dream? What is a molecule that is present at single micromolar concentrations or greater in vivo? Hopefully they make me a short list of compounds they want to get some spatial and temporal data from. We can go through that list and be like “we can SABRE that one!”. That is pretty exciting, and I would love to have that conversation with folks who understand the details of COVID. Maybe some of that already happens in the lung as well.

I had the opportunity to speak with a pulmonologist here at MGH early on in the pandemic. I told her that I used to do helium and xenon lung imaging, and I wondered if obtaining ventilation and perfusion maps of the lungs would help. She did not think so, because even though those images might be interesting, they would not advise treatment currently. The goal of everything I do is not just to make pretty pictures but instead to build tools in response to clinician scientists who tell me: “You know, if I could see this, it would indicate something that we can use to take therapeutic action”. Currently, people are doing all sorts of x-ray imaging of COVID patients, and Radiologists are well aware what the lesions in COVID look like. Knowing the [molecular] target that we want and then going after that with SABRE hyperpolarization and MRI would just be an awesome tool, even preclinically.

CC: What about the clinical workflow of MR imaging, what changes might we see there?

MR: The thing I hear a lot – that is a consequence of what is going on now – our MR suites are just not used to having to operate in the presence of widespread infectious disease. So as of right now, the workflow for MR scanners has really changed. You have to go in and disinfect everything in between exams. In the case of say MR triage of stroke where the patient might be COVID positive, it is going to require downtime after the exam for that scanner for cleaning.

I do wonder if this is this going to change or lead to a significant kind of rethinking of the kind of architectural designs of scanners? At the moment, what is done includes things like placing blue pads between the patient and the scanner to make sure a person is not in direct contact with the bore. That is going to change going forward! I am curious and hopeful that it might get people thinking about how to manage this, and it may very well lead to smaller, purpose-built abdominal, cardiac or head only scanners.

CC: Then it becomes an engineering challenge. What key developments would have to happen to bring that market?

MR: To build a scanner that looks different and is capable of being at the patient bedside for instance? In general terms, what it requires is kind of a fundamental consideration of the signal equation and MRI. Especially if we are talking about smaller scanners, with different kinds of geometries, and maybe at lower magnetic fields or with weird non-linear gradients as an example. Now, we are lucky in that there is compute machinery to do generalized reconstructions, and to solve general inverse problems (like our work using AUTOMAP) [Zhu, Nature, 2018]. This will really allow us to rethink new scanners from the ground up, asking the serious question of how good does the imaging actually have to be, and can we attain good enough image quality by leveraging the compute side rather than hardware requirements such as homogeneity, linearity, and other performance metrics.

CC: Do you foresee much challenge with adapting clinicians to this idea?

MR: To the idea of weird looking purpose-built scanners? You might have two months ago. But today, if I brought in some weird looking hand held single-sided magnet and said this can do cardiac MR, and you can sterilize it in a jiffy, I think my colleagues at the MGH would not hesitate to use it. These will be imaging tools advised by the physicians who are going to make decisions about care, not necessarily by people who have trained their whole lives to read say T2-weighted 1.5 T MRI.

Physicians are very practical people in general and they will do what it takes to help their patient. If new MR-based tools are able to be used to answer specific questions, as decision tools, you will get people on board. Let us look at point-of-care ultrasound as an example. Ultrasound has had a huge resurgence lately with the $2000 Butterfly scanner. Why? Because it is so inexpensive that people are using it to look at everything with it, including the lungs.

I have spent years thinking about all kinds of weird use cases for different shaped MR scanners. However, until now, I never thought about how to sterilize an MRI scanner. It will be interesting to add this to the mix of scanner design needs.

CC: Almost makes you wonder if the big infectious disease research hospitals have any input here.

MR: I wonder. It does get you thinking!? A lot has been learned over the last hundred years to sterilize our equipment and it is remarkable to see how much that field is still evolving. There are whole sectors along these lines that are in flux, such as new approaches to sterilization of our scanners!