Radiomics-Based Muscle Quality Assessment from Multi-Site Multi-Vendor MRI
This study demonstrates that radiomics analysis of multi-site, multi-vendor Dixon MRI data, particularly using fat-fraction features, significantly improves the prediction of muscle strength in older adults compared to traditional cross-sectional area measurements by capturing critical muscle quality and heterogeneity changes associated with aging.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
As people grow older, their muscles often weaken, a condition that can lead to falls, loss of independence, and a decline in overall health. This process, known as sarcopenia, is not simply about muscles shrinking in size; it is also about the quality of the tissue itself. Just as a wooden beam can rot from the inside while still looking solid on the outside, muscle fibers can become infiltrated with fat and scar tissue long before the muscle visibly shrinks. For decades, doctors have relied on measuring the cross-sectional area of a muscle—essentially counting how much space it takes up in a scan—to guess how strong a person is. While size matters, it tells only part of the story. It misses the subtle, internal changes in texture and composition that signal a muscle is losing its ability to generate force.
A new study from researchers at New York University, Stanford University, and the National Institute on Aging explores a different way to look at muscle health. They turned to a sophisticated type of magnetic resonance imaging, or MRI, that can separate water from fat within the muscle tissue. By using a method called radiomics, which treats medical images as rich sources of data rather than just pictures, the team analyzed the tiny patterns and textures inside the thigh muscles of 141 adults. They wanted to see if these hidden patterns could predict muscle strength better than simply measuring the muscle's size, especially in older adults where the link between size and strength often breaks down.
The researchers gathered data from three different groups of people scanned on different MRI machines in different locations. One group consisted of healthy adults ranging from their early twenties to their late nineties, another group was made up of healthy adults in their thirties to seventies, and the third group included older adults, some of whom were healthy while others showed signs of frailty or muscle loss. The team measured the actual strength of the participants' legs by having them push against a machine with their knees, recording the maximum force they could produce. They then compared two ways of predicting this strength: the traditional method of measuring the muscle's area on the scan, and a new method that used computer algorithms to extract hundreds of detailed texture features from the MRI images.
When the researchers looked at the healthy, younger groups, the traditional measurement of muscle size worked quite well. A larger muscle generally meant a stronger leg, and the computer models based on size performed similarly to models that looked at texture. However, the story changed when the older adults were included in the analysis. In this mixed group, simply measuring the size of the muscle became a much weaker predictor of strength. The muscle might look large on a scan, but if it was filled with microscopic fat deposits or had a disorganized internal structure, it was significantly weaker than expected.
The radiomics approach, which analyzed the texture and composition of the muscle, proved to be far more effective in these older populations. The computer models that focused on the fat content within the muscle tissue were particularly successful. These models could detect the subtle, chaotic patterns of fat infiltration that a human eye or a simple size measurement would miss. When the researchers combined the data from all three groups, the texture-based models predicted muscle strength with a level of accuracy that was significantly higher than the size-based models. The fat-related features were the most consistent indicators, showing that the more disordered and fatty the muscle tissue appeared in the scan, the weaker the muscle was likely to be.
The study also revealed that the most informative details were not just about how much fat was present, but how that fat was arranged. The computer analysis identified specific patterns of irregularity and non-uniformity within the muscle tissue that correlated strongly with weakness. It found that muscles with a more chaotic internal landscape, where fat and water were mixed in complex, uneven ways, were less capable of generating force. This suggests that the organization of the tissue is just as important as the amount of tissue itself. The researchers noted that while the traditional method of measuring muscle size is still useful for healthy people, it fails to capture the early signs of decline in older adults. The new texture-based method offers a way to see these early changes before they become visible as shrinkage.
One of the strengths of this work was its ability to handle data from different types of MRI machines and different locations, which is often a challenge in medical research. The team found that the fat-based texture analysis remained reliable even when the scans came from different manufacturers. They discovered that the specific type of fat map used mattered; maps that provided a precise, quantitative measure of fat percentage were more stable and accurate than simpler methods. This consistency suggests that the technique could be robust enough to be used in real-world clinical settings, even if the equipment varies from hospital to hospital.
The findings point toward a future where doctors might be able to assess a patient's risk of frailty or muscle loss simply by looking at the texture of their muscles on a routine MRI scan. This could allow for earlier intervention, helping to identify people who are losing muscle function before they lose muscle mass. While the study did not test this method on every possible patient group, and the specific MRI sequences used are not yet standard in every hospital, the results provide a strong proof of concept. The research demonstrates that the hidden language of muscle texture holds valuable clues about strength and health, offering a more complete picture of muscle quality than size alone can ever provide.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.