Success Story
Improving Treatment for Patients with Multiple Sclerosis
Neurology
New imaging technologies help distinguish between active and chronic lesions without using contrast agents.
Multiple sclerosis (MS) is a neurological disease that affects approximately 18,000 people in Switzerland with onset typically seen between the ages of 20 and 40. In this disease, the immune system attacks the myelin sheaths that insulate nerve fibers in the brain, causing inflammation and lesions. This disrupts the flow of information and leads to clinical symptoms such as vision problems, walking difficulties or numbness and tingling.
There is considerable variation in the disease course between patients. Some patients progress quickly, while others remain stable for years. Monitoring brain lesions is therefore important, especially for distinguishing between chronic and active lesions. While active lesions tend to grow and cause further damage, chronic lesions remain more or less stable. Currently, gadolinium-enhanced MRI is the standard way to distinguish between the two. However, gadolinium, a chemical element, is toxic and washes out slowly, so a safer and less invasive technique is needed.
A PHRT grant enabled Elda Fischi-Gomez, a computer scientist at the EPFL, to develop new medical imaging technologies that can determine whether the lesions are actively changing, or whether they’re old scars that aren’t causing new problems. “We combined two different types of MRI scans to get information on demyelination and degeneration without using contrast agents”, she explains. One is a multi-echo T2 sequence, which provides information on myelin. The other is diffusion MRI, which detects the movement of water in brain tissue. “By combining these two pieces of information, we have created a kind of toolbox that helps us differentiate between active and chronic lesions based on natural properties of the tissue”, she says.
Fischi-Gomez has tested her new technique on 60 patients. “The results look promising”, she says. Using machine learning, the lesions could automatically be classified as active or chronic with good accuracy. “This approach could eventually help clinicians make more informed treatment decisions without the need for contrast agents”, she says.
Her technique could also solve the problem of clinical-radiological mismatch: Sometimes an MRI shows several lesions, but the patient seems to be functioning well. Or a person has only few lesions but is severely impaired. “We are trying to understand this mismatch between what the MRI shows and the patient’s actual condition”, she says.
Fischi-Gomez hopes to expand her work beyond MS. She is currently working on a project involving premature births. “With newborns, especially preterms, we need to scan quickly”, she explains. “We are developing fast MRI sequences that capture high-quality data. This may help us understand early brain development and predict neurodevelopmental disorders later in life.”
“I’m a computer scientist, not a physician”, she says. “I rely on clinicians to tell me what really matters in patient care.”
Fischi-Gomez’s research illustrates how one innovative idea can have multiple applications. And it’s a perfect example of how technology can improve health, which is precisely the goal of the PHRT initiative. “I’m a computer scientist, not a physician”, she says. “I rely on clinicians to tell me what really matters in patient care. We need each other. Together we can develop tools that are both scientifically robust and capable of advancing medicine.”