To the layperson, an image—say, an MRI of the brain—is a diagnostic test. Yet to MIT Professor Polina Golland and her team, that individual image is just one tiny piece in a complicated puzzle where no one yet knows the full picture.
“When you look at 10 people who have the same disease, it’s oftentimes very hard to generalize what really is underlying the symptoms. When you look at 10,000 . . . that gives hope to create insights. For example, how normal aging affects the brain versus different diseases.”
Using advanced computational tools to analyze thousands of patient images, Golland’s team detects subtle patterns of change. Through analysis and applied mathematics, they hope to help clinicians understand how, where and when disease affects human organs.
“Quite recently the field of statistical inference applied to very large data sets has bloomed,” Golland says. “What’s exciting is, in the last 10 years, the computational methods for performing the analysis have dramatically improved as well, because suddenly they were relevant.”
In a current project with neurologist Dr. Natalia Rost at the Mass General Hospital (MGH) Acute Stroke Service, Golland’s team uses such methods to help answer how genetics impacts stroke, neurodegeneration and white matter disease. Working in a tight cluster, MGH scans and collects DNA from patients ages 30 to 90 with acute stroke; The Broad Institute sequences the DNA; and Golland oversees analysis of the images plus genetic information. Preliminary evidence from the study shows that chronic white matter damage accumulation is correlated to recovery from stroke and overall cerebral vascular health.
“Today’s computational power enables us to perform analysis that I couldn’t even dream about when I was starting. As you acquire more images, based on a larger population . . . it’s no longer a ‘typical person’ from the clinical or normal cohort, it’s actually down to different subpopulations in the clinical cohort that characterize heterogeneous effects.”
In this way, as research goes big data, the path is being paved toward personalized medicine. “One of the implications is that we are getting closer to being able to produce precise answers for every individual.”
In addition to her analytical role, Golland says she also helps clinicians take full advantage of presurgical imaging.
“The biggest challenge right now is that really good quality images still provide very diffuse information . . . . Yet there is more to extract and present to clinicians. We’ve built software that transforms ‘slices’ of images into a three-dimensional mesh of an object that helps clinicians plan what they’ll do in surgery.”
Golland’s group collaborates actively with MGH, Brigham & Women’s Hospital and Boston Children’s Hospital, as well as with the Brain & Cognitive Sciences researchers at MIT. Such interdisciplinary collaborations between engineers and clinicians can sometimes be quite challenging due to vastly different approaches, but “that’s the fun of it,” she says. Early in her PhD study at MIT in Electrical and Computer Engineering, with no prior background in medical imaging, she simply jumped in. “I showed up to a meeting with the clinical partner of my PhD advisor and said, ‘I know how to program.'”
Today, Golland is the PhD advisor, bringing on new recruits.
“The way I describe it to potential graduate students is: first, we get together with a potential collaborator who might have some problems that are clinically important and technically challenging. That’s a good combination for us. And they say, ‘This is what we would love to do, but of course there is no way to do it.’ That’s when we get interested.”
Recruiting MIT students who want to join in that challenge is a joy for Golland. “These are students from applied mathematics, computer science, physics, electrical engineering—disciplines that think about modeling the world. They may have a curiosity about what the medical profession is like, without wanting to go to medical school. They are engineers at heart, but want to make a difference for people. That’s what’s important to us.”