To get an accurate MRI, the patient must lie completely still for a long period in a confined space, be able to hold their breath on command, and withstand loud banging noises. That’s why it’s often very difficult to get young children to comply, even though they may need the scans for their healthcare.

In some cases, doctors give children anesthesia to temporarily suspend movement of their lungs and abdomen to acquire satisfactory MRI images. The child’s breathing is controlled throughout the scan by a ventilator, which is halted by the anesthesiologist for short intervals when the MRI technician needs the torso to be still. But that turns a relatively non-invasive procedure into something with larger risks and costs.

While a CT scan can be sufficient in many cases, an MRI often provides additional information that can influence treatment. And, unlike X-ray or CT, MRI doesn’t expose patients to ionizing radiation.

Researchers at Stanford University are using a multi-pronged, team-science approach that involves adapting MRI equipment for pediatric use, developing better motion correction strategies, and implementing state-of-the art image reconstruction techniques that have significantly sped up time it takes for a child to undergo an MRI scan.

Their first goal was to design and build MRI signal-receiving coils tailored to a child’s body. Signal-receiving coils surround the part of the body that is being imaged and are responsible for capturing the radiofrequency signal produced by the body during an MRI scan. The coils are designed to maximize the amount of true signal that is received while minimizing the noise or interference. However, standard coils are often larger than needed for children and pick up extra noise, causing images to become less sharp. In collaboration with GE Healthcare, they constructed parallel arrays of child-size receiver coils for imaging the abdomen. While the reduced size of the coils enhances image clarity, the parallel array layout speeds the scan time by allowing individual coils to pick up the signal from different parts of the body simultaneously, rather than sequentially.

In collaboration with an electrical engineer at U.C. Berkeley, they were able to implement a technique called compressed sensing to reduce scan times by gathering only a small fraction of the data conventionally needed to enable reconstruction of a complete magnetic resonance image, called under-sampling. The key to the technique is a special algorithm used after the scan that can reconstruct the full MR image from these few data points with high fidelity.