An MRI method that can scan for the different physical properties in various body tissues and diseases could offer an efficient way to diagnose cancers, multiple sclerosis, heart disease and other conditions, according to an article published online March 13 in Nature.

The technique, magnetic resonance fingerprinting (MRF), has demonstrated it can differentiate white matter from gray matter from cerebrospinal fluid in the brain in about 12 seconds, according to researchers at Case Western Reserve University (CWRU) and University Hospitals (UH) Case Medical Center, both in Cleveland.

“Our end goal is to really generate tissue maps instead of MR-specific maps,” Mark Griswold, PhD, radiology professor at CWRU and UH Case Medical Center explained in an email to Health Imaging. “If we can add enough physical parameters, we should be able to specifically identify every tissue in the human body and a host of diseases based solely on their MR fingerprint.”

These MR fingerprints are generated by simultaneously varying different parts of the input electromagnetic fields that probe the tissues. These variations make the received signal sensitive to four physical properties that vary from tissue to tissue, and these differences become evident when applying pattern recognition programs using the same math in facial recognition software.

Patterns are then charted, and instead of looking at relative measurements from an image, quantitative estimates are used to tell one tissue from another, according to Griswold. As the technology progresses, and more physical properties are able to be interrogated, these results will determine whether tissue is healthy or diseased, how badly and by what.

“Just as in a normal criminal case, fingerprints aren't ever perfect--there are always errors, like partial prints, or smudges, etc.,” said Griswold. “However, the people at the FBI are still able to make a positive identification with an overwhelming accuracy. If we take this analogy to MRI, in many ways conventional MRI cannot tolerate errors, such as motion and noise, that degrade our signal from its ideal shape. In MRF, we remove all of the rigid criteria and ask simply ‘which is the most likely tissue?’”

The technique also could drastically shift MR workflow. Unlike conventional MRI scanners with dozens of controls and imaging sequences that can be altered to extract specific pieces of information, MRF can extract multiple pieces of information simultaneously from a single randomized sequence, according to Griswold.

“The vast majority of the important controls and sequence design could happen once at the factory, and the end user would just be presented with a big ‘scan’ button,” he said. “This could clearly provide a practical increase in throughput just by simplifying the use of the scanner.” Moreover, since the results of MRF are quantitative, they could be repeatable from site-to-site and day-to-day, and attending radiologists could be more efficient at reading the scans by making quantitative comparisons.

“This would be very much more in line with the process for CT, where the Houndsfield unit provides a solid bedrock. MRF could be viewed as a multidimensional Houndsfield unit,” said Griswold.

Researchers plan to reduce scanning time and continue building the library of MR fingerprints over the next few years.