A group of doctors and engineers solve problems that prevent doctors from ordering an MRI for some children.
Today’s MRIs are anything but kid-friendly. But if a group of doctors and engineers have its way, that problem will be a thing of the past.
A typical MRI exam involves the patient lying down on a narrow bed that is pushed into the bore of a large magnet. As part of the exam, the body is subjected to a strong magnetic field which makes the protons in the body’s water molecules align their spins with the magnetic field. The protons in the area that needs to be studied are then selectively manipulated to give off radio waves. A heavy receiver coil is placed on the area, acts like an antenna, and picks up the radio wave signals.
The MRI signal is relatively small and must be sampled long enough for clinicians to detect it in a large sea of various radiofrequency signals. It is also critical for the patient to be motionless, as voluntary movement, or even that from a rapid breath or heartbeat, might corrupt the data.
All this is bad news for kids. The coils are heavy and scary, children can’t lie still for long, and they have faster breathing and heart rates than adults and often need to be sedated. To avoid these challenges, pediatricians will often choose alternative procedures, such as CT scans and ultrasound. But those may not provide the same sensitivity or specificity to diagnose diseases. Shreyas Vasanawala, MD/PhD, a pediatric radiologist at Stanford University, along with a group of interdisciplinary engineers, believes there’s an alternate solution: Instead of skipping an important imaging tool such as an MRI, solve the stumbling blocks that deter medical professionals from ordering one for children.
Vasanawala’s approach involves addressing two of the biggest challenges: the intimidating weight of the coils and the amount of time needed to capture accurate scans. These interdisciplinary efforts to make MRIs more kid-friendly involve a number of key researchers.
John Pauly and Greig Scott, electrical engineers at Stanford, for example, are working on making the coils lightweight and flexible. They are developing new methods of transmitting signals without the requirement for bulky cables. Vasanawala is also working with Ana Claudia Arias and Michael Lustig at University of California, Berkeley, to develop flexible coils that can be printed with conductive inks on fabric, so the child only need wear a snug T-shirt housing the embedded coils.
“Our vision is to have it be bespoke, tailored for individual patients,” said Vasanawala, who also serves as director of MRI at Stanford Children’s Health. “Right now we have developed different sizes that are flexible and lightweight, but the overall goal for the project is to have the whole system embedded in cloth so the child wouldn’t know there is hardware being placed on.” The child-sized coils are not only more comfortable, but they also cut down on background noise and interference, thereby delivering better quality images, Vasanawala said.
Next comes the scanning time. On this front, Vasanawala is focusing on doing more with less. Many medical images contain a lot of unnecessary or corrupted data that clinicians can discard without compromising the image. The challenge is to figure out where and how to collect the right data. Vasanawala is focused on improving this process of “compressive sensing” using machine learning and the expertise of Dr. Joseph Cheng, a senior scientist at Stanford.
“Over the years we have done many long conventional [MRI] exams and have a lot of data to work with to arrive at very good-quality image reconstruction,” Vasanawala said. “So now we can subsample this data and train neural networks to learn how to reconstruct images with very little data.” This enables faster scanning. Vasanawala credits graduate students Christopher Sandino and David Zeng with advancing imaging speeds by an order of magnitude. Essentially, machine learning and artificial intelligence help connect the dots.
Work is also being done in figuring out algorithms that will reconstruct images using motion-correction strategies, something that is invaluable for MRI use in kids. Scans that used to take an hour can now be done in ten minutes.
All this begs the question: Wouldn’t faster scans be advantageous for everyone, kids and adults? Why not work on applying these technologies across the board?
“Sure, there are definitely a lot of benefits for adults and they have flowed over to the adult side,” Vasanawala said. But our motivations are driven by kids and to make their experiences easier instead of having to adapt adult solutions to pediatric problems.”
Poornima Apte is an independent writer who focuses on technology.