How do the requirements of interventional imaging differ from those of diagnostic imaging?

In interventional imaging, we are performing an invasive procedure with the patient in the bore of the MRI scanner and using the MRI information to help guide the procedure. So, there are a few differences from a regular diagnostic environment. For example, we don’t just image tissue like we would in a regular diagnostic exam; we also simultaneously image external devices, like guidewires and catheters and needles, that are being used inside the body for the procedure.

Also, imaging has to be very fast, given that we are using these images for device navigation and for procedural guidance. We cannot tolerate any latency in our real-time image reconstruction, and we cannot take data offline to do postprocessing or any sort of correction. The images need to be available to the interventionist immediately. There is also a lot of additional infrastructure compared to diagnostic imaging, such as real-time interactive imaging platforms, in-room displays, and communication devices.

In regards to the workflow, the clinicians who make the final decision on how to proceed are close to the patients during the intervention. How do they get the images?

I work specifically on MRI-guided cardiovascular procedures, where interventionists are navigating a device in the heart using the vessels from an access point in the femoral vein at the groin. They are standing at the foot of the patient, looking at images displayed in front of them. One additional piece of equipment we need is a display for the interventionists inside the room. That way, they can be beside their patients and navigate a device while looking at the images.

Can they also give feedback or control how the images are to be acquired?

The feedback is often given verbally. For example, they would ask the MR technologist to modify the image preparation pulse or change the slice position. They often do not have control over the imaging themselves to change parameters in real time, although researchers have explored this idea. Another thing which is simple, but very helpful, is a foot pedal to turn imaging on and off whenever they need it.

When does interventional imaging fail in regards to the purposes of the clinicians, and what are typical causes for that?

Cardiovascular interventional imaging has mostly stemmed from X-ray, which offers projection-style imaging where you can see the entire length of your device at once, but cannot see the soft tissue very clearly. As we move a procedure from an X-ray environment into the MRI environment, it is difficult to maintaining visibility of the entire length of the device in 3D space. As a result, it can be difficult during the course of a procedure to determine whether the device moved out of plane or went into the wrong spot. We have to make real-time decisions based on a limited imaging signature.

Interventional imaging violates some of the normal assumptions about the MRI system. Can you take one example and walk us through how you addressed it?

Interventional imaging truly pushes the limits of a traditional MRI scanner. I can think of two examples where we have overcome limitations. One of the limitations is related to fast imaging. To achieve the required frame rates, we moved to techniques such as spiral imaging which allowed us to get fast real-time images. However, we then had to deal with distortions from imperfect trajectories and from blurring in real time. Rather than take the data offline to  fix these artifacts in postprocessing, we streamed data to an inline reconstruction system, namely Gadgetron, to apply trajectory corrections using the system impulse response function, and to do real-time deblurring.

The second limitation has to do with device safety. The devices that we use for cardiovascular procedures are often long and metallic. When we put them into the MRI scanner and conduct real-time imaging, they run the risk of heating during scanning. This can be very dangerous. So unfortunately, we have not been able to use a lot of our normal tools from the cath lab in the MRI scanner. To overcome this, we have recently ramped our scanner down from a normal 1.5T to 0.55T – which is very counter-intuitive in the world of MRI. However, in the case of intervention it makes sense because heating goes with the square of the field strength. Thus, we can take some metal devices that would be used in any other intervention in the world and put them into the MRI scanner. This has opened up the door to a lot of new procedures. By ramping down a modern closed-bore system we are able to keep the high-performance gradients and the field homogeneity, meaning that we can do fast cardiac imaging really well at low field.

Do you encounter resolution issues because you are limited by the SNR at 0.55T?

Yes, we have to deal with the lower SNR and are therefore developing strategies to compensate for that. A lot of our SNR recovery depends on using methods such as spiral acquisitions, which can be very SNR-efficient. Of course, concerns about blurring, distortions, B0 or B1 inhomogeneity are significantly reduced at low field.

What additional capabilities does an MRI system need in order to serve as an useful interventional system?

The real-time image visualization software plays a big role in offering the ability to interactively change slice orientation and image contrast. There is other peripheral equipment such as a communication system, which allows the interventionist to communicate with the team in the control room, and in-room displays.  Particularly for cardiovascular interventions, it is important to have diagnostic-quality ECGs rather than the ones used for triggering within MRI and hemodynamic recording capabilities.

I have been wondering how you create physical access to the patient. How does the medical personnel tolerate working with the constraints of the bore opening?

Wider bores are definitely helpful. For cardiovascular procedures, we often use an incision in the groin to navigate a device in the heart. But, for other procedures, patient access can be a real challenge. The MRI environment is not very friendly to this, in general. Obviously, we have to be very careful about safety, especially if we are moving a patient from the cath lab into the MRI scanner. Also, just the acoustic noise of the MRI scanner can be very difficult for the interventionist. Noise-cancelling headphones with a microphone are part of the communication systems that we use.

Can any standard MRI system be used as an interventional system with the right additional devices?

Yes, absolutely. In hospitals, more people are moving into this model because they want to do an MRI-guided procedure on their patient but do not necessarily have a dedicated suite that has an adjoining cath lab. Instead they just take a conventional MRI system and run with fast imaging and their standard diagnostic sequences. They can use hand signals and patient communication equipment. It is challenging, but it can be done.

Are open MRI systems commonly used for interventional procedures?

They are available but not commonly used, in my experience. The limitations in terms of image quality caused by the poor field inhomogeneity and gradient performance were prohibitive. In the field of cardiovascular interventions, most people use closed-bore systems.

What are current unmet needs in interventional imaging? Where are the biggest opportunities?

For me, it is the devices. In order to do complex procedures in the MRI scanner, we need the right tools. Right now, we do not have the tools that are always safe and easily visible. If we had the appropriate devices, or an intrinsically safe MRI scanner, we could easily build the imaging technology to perform complex procedures.

Are people currently working towards solving that need?

There is a lot of development of non-metallic versions of tools or safer metal versions. Historically, there are many decades of engineering that have gone into building the dozens of different devices that have the exact right mechanical properties for each individual anatomy and each individual procedure. The task of rebuilding that entire catalogue of devices to be MRI-safe is tremendous. While there is certainly more work to be done in that direction, the combination of safer imaging with new tools is going to allow us to do very interesting things in the future.

Would it be a benefit for the interventionist to have diagnostic imaging quality during the scans?

Yes, it would be a benefit. In fact, the real motivation for using MRI to guide these procedures, given all of the cumbersome additional equipment and the difficult working environment – is the ability to use the diagnostic capabilities of MRI. We want to be able to use 3D anatomy to do more complex procedures, which we can only get with MRI and not with traditional imaging modalities. Also, we want to be able to measure physiology or visualize the tissue changes as we are performing the intervention. The possibility to see that our ablation or biopsy is targeted in the right place and to confirm the effect is what we expect.

Adrienne Campbell-Washburn: You work on high-field spine imaging. Thinking about the interventional world, where do you see potential for spinal cord interventional MRI?

Johanna Vannesjö: As I do not have the clinical experience, this is somewhat speculative on my part. However, I can imagine that it would be helpful to determine successful decompression of the cord in interventions where you release pressure on the spinal cord. There are a lot of patients who suffer from cord compression, e.g., because of disc herniation. Thus, it is a quite common type of surgery. One question regarding treatment, which is still difficult to answer, is which patients will benefit from surgery. It is more of a pre-intervention question where more information would be needed. MR images could potentially deliver information that could help the clinicians to make these kinds of decisions.

I can also imagine that interventional MRI would be useful in pediatric cases with malformations, e.g., when the cord gets tethered to the surrounding meningeae given it should normally be more or less free-floating in the CSF. The guidance of interventional imaging to visualize the anatomy and see where to make the intervention more exact may be helpful.

One big issue is that the spinal cord is very small. To be able to visualize the structures, or the tissue integrity, one needs high resolution in the images. This is one of the challenging cases where a combination of fast imaging and very high resolution would be needed to make interventional MRI useful.