Features:

TUNING IN TUMORS
by Trish Barker, NCSA science writer

Magnetic resonance imaging is one of the most valuable diagnostic tools available to today's physicians. MRI gives medical practitioners highly detailed views of the inside of the human body and aids in the diagnosis of a wide range of diseases and injuries.

A revised understanding of nuclear magnetic resonance that has developed over the past decade holds the potential to make MRI an even more sensitive tool, one that can provide unprecedented contrast and resolution and could potentially detect tumors at an earlier, smaller stage. A team of researchers led by Warren S. Warren (http://www.princeton.edu/~wwarren/), a researcher at Princeton and the University of Pennsylvania Medical School, is using computational modeling on NCSA's IBM p690 system to support unusual experimental results that have added an important new chapter to established NMR theory.

Minuscule magnets

The technique of nuclear magnetic resonance was first developed in the 1940s and is based on the quantum mechanical property of spin. Spin makes particles- -protons, electrons, and atomic nuclei--behave like miniature magnets.

In NMR, a strong magnetic field is applied to a sample (water, for example, or the tissues of the body in the case of MRI). In response to this magnetic field, the nuclei in the sample align either with the field or in opposition to it. A radio frequency pulse is then applied, which knocks some of the nuclei into a new position, their spins askew. The axis of each tipped nucleus rotates, or precesses; the speed of the rotation varies depending on the type of nuclei and the strength of the fields that are involved.

The minute magnetic fields of these nuclei oscillate because of this rotation, and this creates a radio frequency current. These NMR signals can be detected and harnessed to create detailed images of the interior of the human body.

How NMR produces images

The human body is full of hydrogen atoms, which are ideal subjects for MRI because they have a strong tendency, termed magnetic moment, to line up in the direction of a magnetic field. The magnet in an MRI machine forces those hydrogen atoms to line up either in sync with or in opposition to its magnetic field.

For the most part, the alignments of these atoms cancel each other out--for each one aligned toward the patient's head another will be aligned toward the feet. But out of the millions of hydrogen atoms in the body, some will not be canceled out. These "extra" atoms are those that will produce the MR images.

A radio frequency pulse is applied to the area of the body to be examined. The pulse pushes the "extra" atoms to spin in a different direction and at a particular frequency. When the radio frequency pulse is turned off, these atoms slowly begin to return to their former state, releasing a signal as they do so. This signal is detected and is converted, through a mathematical operation known as a Fourier transformation, into an image.

A shift in NMR theory

That was the understanding of NMR and the state of MRI technology until the early 1990s, when Warren noticed experimental results that didn't fit the established theory. Simple pulse sequences that NMR theory predicted would generate no signal instead produced strong signals. The results were so unexpected, they initially were labeled "crazed" experiments, but continued experimentation established that the signals were no fluke. Instead, Warren realized that a key assumption of NMR theory was wrong.

Scientists had long known that two spins within the same molecule could sometimes react to a magnetic field in tandem, precessing differently as a pair than they would independently. This phenomenon is known as multiple- quantum spin coherence. Warren realized that the strange signals he had found were the result of spins from widely separated molecules teaming up. He labeled this new phenomenon intermolecular multiple-quantum coherence (IMQC).

"Ordinarily you assume that individual molecules don't interact with each other. That is where we found the mistake," he explained. Because of Warren's experiments, NMR theory was adjusted to include the IMQC phenomenon.

Creating sharper images

In addition to this shift in NMR theory, Warren's discovery also presented opportunities for improved MRI techniques.

The frequency of the signal received from teamed spins correlates with the amount of oxygen in the sample, which is a key indicator of what processes are occurring there. Tumors are often oxygen guzzlers, so a detailed map of tissue oxygenation could provide a precise technique for cancer detection.

Spin coherence could also help MRI technicians create more detailed images at a higher resolution. In standard MRI, NMR signals are detected from blocks of tissue that are a few millimeters in height, width, and depth. But by leveraging IMQC, technicians may soon tune MRI machines to detect signals from spins that are separated by as little as 100 micrometers. This capability translates into more refined images, which could enable physicians to detect tumors that are too small to be found with standard MRI technology. Earlier detection means treatment can begin sooner and increases the chances of recovery.

The new imaging technique could also provide a more precise map of a tumor's position and shape, information that could help physicians and pharmaceutical researchers target tumors more effectively.

"The problem with targeted medicine is you need to be able to see the target," Warren explained. The enhanced MRI technique could be extremely valuable in the drug development process because researchers would be able to determine at a molecular scale whether or not the drug being tested was having an effect.

Warren's research group is now trying to "tune in" the finer resolution by determining which pulse sequences produce the sharpest images. "The basic issue is trying to find a methodologies that do a better job of telling you something at the molecular level," Warren said.

The research team uses an NMR simulation program to test pulse sequences; the code models how various sequences will affect a fat-water system that is similar to human tissue. The goal is to determine which pulse sequences yield the most useful information and the most detailed images.

Warren's research group had been running the simulations on a workstation at his Princeton lab, but "to do any meaningful calculation it was taking six months or even more," said Prasad Lakkaraju, co-principal investigator on the current NMR study at NCSA and a professor at New Jersey's Georgian Court College.

While on a research sabbatical with Warren's group, Lakkaraju worked on parallelizing the simulation program and porting it to NCSA's p690 cluster, where the group has an allocation of 20,000 hours. "What it will take is running a large number of datasets," Warren said.

Funding statement

This research is funded by the National Institutes of Health.