While MR is not the frontline emergency medicine tool that CT is, it can have value for those patients recovering from COVID-19. We chat today with Peder Larson, an Associate Professor at the University of California, San Francisco, who has worked extensively in the MR imaging of the lung to explore some of the potential uses in convalescence imaging.

Paul Weavers (PW): Currently we are in the midst of dealing with clinical symptoms and stabilizing the population – we have an epidemiological problem. As we look to once things are a bit quieter and we get back to regular outpatient behaviors: People are coming back with reduced lung function who have recovered from an acute COVID-19 case.

Is there a way we can determine whether this is an acute lung injury that can recover with therapy or if there is permanent damage? Is that a straight structural scan you can do with CT or ZTE MR? Does something with a contrast medium such as hyperpolarized gas provide more information? Or is there even a way to look at oxygen extraction?

Peder Larson (PL): I think it is in one of the latter areas, where a lot of the clinicians I have talked with have expressed the most interest in what MRI might be able to offer. In terms of structural imaging, it is hard to match the resolution and the speed of CT. However, radiation exposure is always a risk/benefit calculation. In this situation, I suspect you will be less concerned with the radiation exposure risk if somebody really has compromised lung function after COVID-19 recovery.

I think, the most potential and opportunity is what is considered more functional imaging. There are options with CT, for example ventilation imaging, but there are even more things you can do with MR. You are not only looking at the recovery of the structure. You can look at ventilation metrics where hyperpolarized gasses do a great job. It is basically how much you breathe in is getting to different parts of the lung. Hyperpolarized gas MR is often considered the gold standard for this type of functional imaging.

There are also MR based ventilation measures where you are looking at [the lungs] and doing timeseries of images during respiration, followed by image processing. You do this to try and extract if it is just moving or expanding and contracting with respiration, which would then indicate that it is actually being properly ventilated.

PW: It sounds like you need a lot of extra hardware, basically a hyperpolarizer and the right infrastructure to deliver this quickly to the patient and get them to the scanner. Do you know of standard clinical sequences where we can try to get similar information?

PL: You can get some of the information by just doing time or motion resolved imaging. The most basic would be using a fast 2D gradient echo sequence. You take the timeseries of data watching the lung expanding and contract and rely on image registration across the phases to assess whether you have a ventilation defects present or not. The most recent version of that is called PREFUL and has the advantages that it is basically all postprocessing – the acquisition is on every system. It is the postprocessing where the magic happens and it can get a bit more sophisticated. We for example work with 3D UTE and resolving images over the respiratory cycle. The Wisconsin group is really strong in this area as well, using time resolved imaging of normal respiration to get some measure of ventilation.

Hyperpolarized gas may be the most involved, needing the polarizer and the coils and obviously the gasses themself. For in between there are couple of other technologies. One is fluorinated gasses – and since fluorine resonates close to protons you can use proton hardware for the most part. There you just need the gas. They are basically inert gasses and you can use the same inhaled gas ventilation measurements with less modification.

The one that is probably the closest to running the scanner as is, is oxygen enhanced scanning. The protocol that most in this area have tried is alternating between breathing room air and breathing 100% oxygen. The oxygen introduces mostly a T1 change. If you run this in a T1 weighted sequence you will see a wash-in effect of the 100% oxygen versus the room air. Not only is the oxygen getting to a certain part of the lung but also a little bit of the dynamics. Is it a bit slower in a certain place, are the airways a little bit constricted or still getting a little ventilated? You can potentially visualize some things in that way. Every hospital will have 100% oxygen around.

PW: These are four different options on a continuum for getting lung function with MR. We can take these images and build this model of lung function. How does this then translate into a care decision? If we have a patient who is recovering lung function versus a patient who we scan once a day and there is no change in lung function? Do we know what treatment options you can take? Does that guide some invasive treatment to restore lung function versus watching lung function recover normally?

PL: That I am not that sure about (Ed. – we are all physicists here!). I had some discussions with pulmonologists not related to COVID but to other diseases where people use an inhaled steroid of some sort. You could potentially do a challenge if you have somebody come in after they have not used an inhaler, then image after they have taken the inhaler. There are also some treatments where you could have somebody come in without having done their usual treatment routine and look at the image before and after to see if they are really getting the benefit of improved ventilation from a certain treatment.

PW: Having a direct treatment-response assessment?

PL: I guess the other treatment options that I am aware of you have the extreme case of people on ventilator of course. They may be doing supplemental oxygen – that might be a step down from the ventilator. If the physician wants to be comfortable finding when they can come down a level of treatment, perhaps the imaging assessment can do that.

PW: Are there a lot of technical challenges other than workflow or making sure you have the right protocol set up to be able to do this?

PL: To some extent. Current these techniques just exist in the research world. None of these have widespread use, we are talking tens of institutions worldwide that have maybe run any type of polarized gas or fluorinating gasses, oxygen enhanced or just MR ventilation. It is a small number that have done it.

A lot of the fundamental technology is there. The question is how mature is that?  People have not done a lot of tests in terms of assessing reproducibility or test-retest, but current work is very promising.

PW: Maybe we can link the oxygen challenge and the 2D timeseries papers or something similar? There is no additional hardware required essentially.

PL: Absolutely. There is a nice review that came out in JMRI recently.

However, another thing that is timely and is even more emerging are low field imaging systems. The most prominent work recently was from NIH. Going to low field is great for lung imaging, you gain on T2* enough to negate the effects of the polarization going down. The lung images they have shown from the 0.55T system are really nice. If you are thinking about emergency situations you go to a lower field which is much more siteable with less safety issues. It is much more amenable to emergency department deployment. If I had one of those systems, I would be imaging lungs with it!

PW: Especially when we start thinking about long term follow-ups with repeated scans: the general radiation dose versus benefit burden if it is a repeated follow-up and we can do it with MR that is helpful.

PL: Yes, you can do it every single day for inpatients. That is the main reason we are doing lung MR’s is for avoiding the radiation burden: targeting kids and longitudinal studies. You are not going to second guess the exposure of a CT if somebody you scan is clearly very sick. Lung MR could be helpful for the understanding disease process, you could image people who have the virus but do not have the symptoms and look at their lungs to see if they have any disease, all without any concerns about radiation exposure.

People have done CT ventilation functional imagining where they try to capture images at different respiratory phases. In MR it is both a disadvantage that we have to scan for so long but it is also an advantage that we do not only get one set of snapshots but can resolve respiratory dynamics. This could be nicely combined with oxygen enhanced MRI – we could have people do this challenge several times and maybe get better results that way.