Heart muscle cells are arranged in a precise, mosaic-like microstructural pattern, and disruptions to that pattern can be an early indicator of many diseases. Unfortunately, doctors have never been able to see any of this directly in a living patient — only through biopsy, which is invasive and samples only a sliver of tissue — complicating efforts to study the diseases.
Now, a team at Massachusetts General Hospital has found a way around that: a noninvasive MRI method that reads out the cellular pattern as a kind of microscopic “bar code,” continuously and without a needle. They describe the work in a recent issue of Nature Biomedical Engineering.
Building on Decades of Cardiac MRI Research
The new method builds on diffusion tensor imaging (DTI), a technique that uses MRI to track the microscopic movement of water molecules through tissue. Because water tends to diffuse along the length of cardiac muscle cells rather than across them, DTI can reveal how these cells are oriented, effectively mapping the heart’s internal architecture.
A team of Martinos researchers led by Dr. David Sosnovik have been pioneers in adapting DTI for the heart. It’s a technically demanding task: a beating heart moves far more — and much faster — than the water molecules the scan is trying to track. As a result, cardiac DTI has historically produced images with limited spatial resolution: sharp enough to confirm the technique works, but too coarse to capture the fine detail needed to study many heart conditions in depth.
A Nearly Tenfold Leap in Resolution
To overcome this, in the study reported in Nature Biomedical Engineering, Dr. Sosnovik’s team combined several technical advances developed at the Martinos Center and by their collaborators: a custom-built, 64-element radiofrequency coil designed specifically for cardiac imaging; an MRI technique that measures water diffusion while filtering out the “noise” of the heart’s own motion; and a post-processing method that allows scans to be collected while a patient breathes normally, with the images sorted afterward by breathing cycle rather than requiring the patient to hold their breath.
Together, these advances improved the spatial resolution of cardiac DTI by nearly tenfold. At that resolution, structural details that had previously been blurred into a single averaged signal suddenly came into focus.
“It became very clear to us that we were seeing a level of microstructural detail in these images that would allow us to assess the microstructure of the heart in new ways,” says Dr. Sosnovik, senior author of the study.
From Blurred Averages to a Cellular “Bar Code”
Earlier cardiac DTI studies had combined data from the entire heart into a handful of summary values. Because the new images were so much sharper, the team realized they could instead analyze each individual voxel — the 3D equivalent of a pixel — on its own.
“When we analyzed each voxel individually, we could now detect interesting properties and relationships between them,” says Dr. Christopher Rock, the study’s first author.
For every voxel, the researchers measured 10 distinct microstructural properties, turning what used to be a handful of measurements per heart into thousands. “The next challenge became how to interpret this wealth of data and derive useful metrics from it,” Dr. Sosnovik says. To make sense of it, the team turned to analytical methods borrowed from computational biology, where similar techniques are used to sort through large genetic datasets.
Applying these methods, the researchers found that each voxel’s set of 10 properties formed a kind of signature, or “bar code,” describing the local microstructure at that location in the heart. Sorting these bar codes revealed that heart tissue consistently fell into one of four distinct microstructural categories.
Testing the Method in Heart Disease
The team first applied their technique in healthy volunteers, then in patients with aortic stenosis, a condition in which the heart’s main outflow valve becomes markedly narrowed, forcing the heart muscle to work harder to push blood through it. Over time, the heart muscle (myocardium) thickens in response to this added demand.
“It has always amazed me that the heart is able to cope with the load imposed on it by aortic stenosis for as long as it can,” Dr. Sosnovik says. “We wondered, however, whether the thickened myocardium would maintain its microstructural integrity or show signs of microstructural damage.”
The researchers compared patients with aortic stenosis to healthy, age-matched control subjects. As expected, the myocardium was significantly thicker in the aortic stenosis patients, and at first glance, its microstructural pattern still looked essentially normal. All of the patients’ hearts were pumping normally, and none had yet reached the point of needing valve replacement.
But the voxel-based bar-coding analysis picked up something a conventional look would have missed. “In the subjects with aortic stenosis,” Dr. Rock says, “we saw an increase in the number of voxels showing signs of transition from a higher to lower degree of microstructural order.”
Dr. Sosnovik adds: “The myocardium, while thickening to deal with the increased load it faced, largely managed to maintain its intricate microstructural pattern.” “However, our microstructural bar-coding technique did show that subtle signs of damage were starting to appear.”
“We are now at a point, at least in the research setting, where we can examine in detail how the cells in the heart are arranged without needing a biopsy,” Dr. Sosnovik says. “This has the potential to help us better understand heart disease, detect it with greater accuracy, and help guide the development of new therapies.”
He cautions that these findings are a research result and should not be used to guide patient care, which should continue to be based on current professional guidelines and the recommendations of a patient’s own physicians.
He also credited the collaborative environment at MGH for making the work possible. The study was conducted jointly through the Martinos Center for Biomedical Imaging and the Cardiovascular Research Center at MGH. “The infrastructure, expertise and collaboration provided by both centers was key to the successful execution of this study,” he says.
The study was funded by a research grant from the National Institutes of Health (NIH).
