Researchers in France and the US have solved an important mystery surrounding magnetic resonance imaging (MRI). The team claims that the breakthrough, which explains why some MR images turn-out better than predicted by theory, could also lead to new imaging techniques.

MRI works by “flipping” the direction of nuclear magnetic moments using a sequence of radio pulses and watching as the moments relax to their equilibrium positions. According to theory, this flipping must be done in a controlled manner to ensure that the nuclei relax in an ordered way — resulting in a high-quality image. However, many control sequences that should not work in theory actually result in perfectly good images.

While this apparent good fortune has not overly concerned the MRI community, it has left some researchers wondering if the theory could be overhauled. Now, Philip Grandinetti of Ohio State University and colleagues at University of Orleans/CNRS and the University of Lyon in France have come up an alternative mathematical framework for describing the nuclei's behaviour while being flipped (J. Chem. Phys. 129 204110).

Spinning around

A MRI scans involves applying a strong magnetic field to a sample, which aligns its nuclear magnetic moments in a specific direction. Then a sequence of radio-frequency pulses is applied flip the moments so they point in the opposite direction. The way in which the nuclei return to their equilibrium positions depends on their local environment — and this can be used to create an image of tissue or other material.

It was just a matter of finding the right mathematical framework to follow the process as it goes along this superadiabatic trajectory Philip Grandinetti, Ohio State University

Inversion is usually performed adiabatically, a method that "locks" the moments and drags them into the flipped state. The speed of the adiabatic process is critical. Too slow and signal will be lost — too quickly and the magnetization spirals out of control, compromising the desired inversion.

What had puzzled researchers was how it was possible to carry out perfectly good adiabatic inversions at speeds that theory said was too fast. The reason, according to Grandinetti and colleagues, is that when the nuclei appeared to be uncontrollable they were still under control but behaving superadiabatically. This means that instead of guiding the moments in one orderly movement from one orientation to another, the moments take a much more indirect (and seemingly more disorderly) route but still end up at the right destination. "It was just a matter of finding the right mathematical framework to follow the process as it goes along this superadiabatic trajectory," he explained.

Less than perfect magnets

This new mathematical framework should help researchers develop even faster adiabatic inversion processes, thus boosting the MRI signal and improving image contrast. The revised theory could also be used to generate clinical-quality images in situations where the magnetic field is weaker or less homogeneous. One area of interest is stray field imaging, that is, performing MRI scans outside the magnet. Another possible application is the development of portable MRI systems, using magnets that are less "perfect" than those needed for existing clinical scanners.

"With this concept of superadiabaticity, we may be able to have better control and be able to manipulate the system in such a way that we can compensate for field inhomogeneities," Grandinetti said.

Put to good use?

The team has no immediate plans to pursue any practical applications themselves. However, the revised theoretical framework could be put to good use immediately. Many MRI systems will indicate on a display panel if a programmed inversion process will not work well. This allows operators to alter the parameters accordingly before beginning imaging. These predictions of inversion success do not account for superadiabatic behaviour, and time can be wasted "fixing" perfectly good pulse sequences.

"If the companies did implement this, they could reduce the amount of time it takes for operators to optimize these processes. At the moment operators are guessing blind and saying: 'I know I could do this faster than what the instrument is telling me.' Now we understand why, we could fix that software," Grandinetti said.