Computer Science Professor is Picking Up The Threads

He once created Mission Impossible-like imaging software for the intelligence community and the military, but now computer science professor Brent Munsell has his algorithms set on solving the biggest challenge of his career: aiding doctors in the discovery of new therapies for an exceedingly rare and debilitating neurological brain disease for his 5-year-old daughter, Maddy.

by Amanda Kerr
images by Mike Ledford

A maze of colorful threads stretches and bends across a convex plane on a computer screen. Each strand of red, orange, yellow, blue and green represents a connection – a critical channel that as a whole creates the mysterious ethos of the human brain. These aren’t the visual musings of a mad scientist, however. They are computer images of real cranial scans revealing the intricate networks that must exist for each of us to unlock our full potential.

“How does this happen – these tracks?” asks computer science professor Brent Munsell excitedly. “It’s a very specific imaging modality that follows water molecules through the white matter [connective tissue] regions of the brain. During an MRI, the data is basically watching what molecules diffuse through the white matter, and when they diffuse, it creates these tracks. And that tells you how much communication is happening between the brain’s two gray matter regions, which control cognition.”

The colorful array of fiber tracks holds the key to better understanding the nuances of neurological function – namely, why and how the brain works the way it does. And, what goes wrong when it doesn’t.

But designing highly specialized computational models to visually capture the tiniest of neurological threads within the brain isn’t a scientific challenge that Munsell took up for fun. He’s looking for patterns within these images that can help physicians better understand – and treat – conditions such as epilepsy, autism and Alzheimer’s disease.

Unlocking the mysteries of the human brain is kind of Munsell’s thing. Visually quantifying how well – or not – the human brain functions is what keeps him up at night. He’s hoping to one day peel back enough layers to find that one-in-a-million answer to some of the brain’s most perplexing neurological conditions. For science. For himself.

But, most importantly, for his daughter.

ONE IN A MILLION | No one suspected something was wrong when Madelyn June Munsell joined the world in January 2013. Sure, at 10 pounds, she was a big baby, but she certainly wasn’t record setting in size. In hindsight, Maddy was a little sluggish to move in the womb – a fact still not in and of itself concerning. Her mother carried a lot of amniotic fluid during the pregnancy – a sign sometimes correlated with Down syndrome. An amniocentesis, however, showed no abnormalities.

“I’m always looking back, thinking maybe there was a sign,” says Munsell.

“There was nothing that stood out, no red flags. But it was obvious early on that something was wrong.”

Just hours after she was born, Madelyn – Maddy to her parents and big sister – suddenly stiffened, her small frame going rigid and still. It was a frighteningly stark contrast to the warm, soft, pliable nature of a newborn baby.

“It looked like she was having a seizure,” recalls Munsell.

But further testing, as more episodes followed, showed something strange.

Although it looked like Maddy was having seizures, electroencephalogram (or EEGs) tests, which detect electrical activity in the brain, didn’t show signs of epilepsy. Whatever was going on in Maddy’s head, it wasn’t seizures.

“The doctors were totally confused,” says Munsell. “The episodes had the physical appearance of seizures, but the brain activity showed something else.”

Then, by happenstance, a colleague of his wife, Melissa, who is an emergency room nurse at Roper Mt. Pleasant Hospital, mentioned a rare neurological disease, wondering if Maddy’s condition could be related.

That night the couple typed the gobbledygook of a name into Google.

“When we saw it, we knew it – 100 percent – this was it,” recalls Munsell.

“There was no question in our minds.”

A genetic test at Duke University Hospital in Durham, N.C., confirmed it.

Maddy had Alternating Hemiplegia of Childhood (AHC), a condition so uncommon it is estimated at roughly only one in a million births.

According to the Alternating Hemiplegia of Childhood Foundation, the disorder can cause “transient attacks of hemiplegia,” or paralysis of a portion of the body, that can last for minutes, hours or even days. AHC can also cause developmental problems, such as fine and gross motor delays that can hinder things like walking, writing, dressing and eating. And the condition can impact cognitive function, including speech and language as well as behavioral issues akin to those associated with autism. There is no cure for AHC, and treatments are very limited.

For the Munsells, it was a relief to have a name for their daughter’s affliction. But with that knowledge came a flood of questions wrapped in fear and sadness at what the future would hold for their little girl. Would Maddy ever walk? Would she be able to talk? Would she get to laugh and play and jump with her big sister, Abbylee?

“It was kind of a good day and a bad day,” says Munsell, choking back tears. “It was tough.”

VIRTUAL REALITY | A small class of six students quietly ponders the image of what appears to be a statue of Buddha on the projector screen at the front of the classroom in the College’s Harbor Walk campus along the Cooper River. Sunlight from a row of windows at the rear of the room is all that illuminates the space, as cargo ships coming and going from the Port of Charleston quietly slip in and out of view.

Brent Munsell asks the students in his computer graphics class to ponder this question: Is the image they’re looking at really a statue?

“This is all mathematically generated; it’s all computer generated, right?” he asks. “I’m just manipulating some information that’s in the data.”

With a few clicks of the keyboard, the details of the statue – its round paunch, neat clothes and soft, sweet face – vanish, leaving only a blank silhouette. Munsell taps the keyboard again, and, once more, the details of the Buddha take shape.

“When you apply a lighting model, you can see his clothes, his stomach, some of his facial expressions,” says Munsell. “Lighting is extremely important to computer graphics. We don’t really have an eye per se. We don’t have the sun or a light source. Everythingis simulated.”

Vectors and diffusion, speculative light and reflectance: all equations and codes that create the illusion of color, depth, texture and dimension.

Munsell demonstrates the various techniques to create dimples in a virtual golf ball and give life to an image of a creepy, purple spider.

But these techniques can do more than create novel images that could appear in video games and the latest sci-fi blockbuster. With another swipe of the computer mouse, a brightly colored array of green, blue, purple and red lines fills the projector screen. A few more clicks of code and the familiar curves of the human brain drape across a section of the colorful lines, which now jut out like a web of fibers from within the mind.

“Think about all the fiber tracks that are in there – it’s actually quite responsive,” says Munsell, referring to the “imposter approach” that allows him to quickly create the myriad of fiber tracks based on real MRI data.

“It’s pretty crazy, huh?”

The students mumble quietly among themselves about these visual revelations, as one student softly whispers “cool.”

FOR MADDY | As a Ph.D. student in computer science at the University of South Carolina, Munsell focused on medical image analysis and computer vision techniques that could, with fine detail, display the shape of organs such as the heart or liver. He wanted doctors – before they opened a patient up – to be able to see if an organ was diseased.

But after Maddy’s birth, his research interests changed.

“When Maddy came along, I really focused more on the brain and how it was connected,” says Munsell. “I pivoted and said, ‘I’m going to use the things I’ve learned with shape analysis, and I’m going to apply them to brain connectivity.’ I wanted to know as much as I could about the brain.”

At the time, Munsell had been teaching at Claflin University, his first teaching position after earning his doctorate in 2009. But with Maddy’s tenuous health, the father of two couldn’t justify the three-hour round-trip drive from Mt. Pleasant to Orangeburg every day. So, when a visiting assistant professor position opened up at the College of Charleston, he jumped at the chance to work closer to home.

Munsell landed the job and joined the College’s Department of Computer Science in the fall of 2013. Two years later, he was promoted to a tenured track position of assistant professor and began working with his department chair to create the Machine Learning and Medical Image Analysis Lab – a small hub of computers stashed away in the College’s Harbor Walk campus dedicated to research that refines how we see – and interpret – medical imaging data.

“When you think of medical studies, you do not typically associate computer science; however, with computer science being pervasive in all facets of modern life, software development, algorithms and data analytics are omnipresent in any serious research study,” says Sebastian van Delden, chair of the College’s computer science department. “We partner with the Medical University of South Carolina in a few such capacities, with Dr. Munsell bringing computer vision and image processing algorithms to the table in order to process and analyze medical images and data. This meritorious work has a direct impact on people’s lives and the betterment of society.”

| From left: Munsell’s computational models create images from brain scans showing white matter connections between different gray matter regions of the brain, brain networks, blood oxygen levels in gray matter regions and (bottom right) an outline of white matter within the entire brain. |

A software engineer by trade and an electrical engineer by education (he holds undergraduate and master’s degrees in electrical engineering), Munsell spent several years in the private sector, following a five-year stint in the Navy, creating software for defense contractor Lockheed Martin and later Charleston-based Scientific Research Corporation. Using computer vision technology, he created Mission Impossible-esque software to help suss out images and patterns from surveillance video, photos and data for intelligence agencies and various branches of the military. His skills were so sharp, the CIA once offered him a job.

Obviously, Munsell knows how to create algorithms that can visualize things, so it was only natural that he would channel those skills into finding a way to help Maddy and others with similar neurological conditions.

Munsell’s wife, Melissa, says being a mad scientist holed up with a computer is just who he is. Maddy’s condition simply added an intensity and deeper purpose to his work.

“I definitely think with Madelyn, it has made him more focused and maybe given him a little more drive to go after what he wants, but with the goal of relating it to what he thinks would be helpful to others,” she says. “He wants to be able to help others any way he can, whether that’s helping better understand the disorder Maddy has, or helping those in the autism community and finding solutions for them.”

She adds, “I think when you have a child with special needs, you’re really more aware of others with special needs and what needs to be done to help those kids.”

And in the three years since launching the lab, Munsell, in partnership with researchers from universities across the country and around the world, has developed ground-breaking computational models for brain imaging that have the potential to change the way we diagnose and understand conditions such as autism, epilepsy and Alzheimer’s disease.

In 2016, he traveled to Greece with researchers from the University of North Carolina at Chapel Hill and the School of Software at Tsinghua University in Beijing, China, for the Medical Image Computing and Computer Assisted Intervention Society’s annual conference. The group presented data on a new imaging technique that better maps connectivity between gray and white matter regions of the brain, a tool that could help diagnose such conditions as Parkinson’s disease, Alzheimer’s disease or autism. In 2017, Munsell’s work, as part of the Infant Brain Imaging Study, a consortium of eight universities across the United States and Canada, was published in the journal Nature. That research, which made international news, focused on a computational model that can detect subtle abnormal brain overgrowth patterns in infants. In a study of more than 400 MRI scans, the technique correctly predicted the diagnosis of autism in infants with such overgrowth patterns with 94 percent accuracy.

And while much of his work has been funded by the College, Munsell and a group of researchers led by Professor Jane Joseph at MUSC recently received a grant from the Department of Defense to study connectome (map of neural connections) biomarkers for predicting Alzheimer’s disease in traumatic brain injury patients. He’s also on another team led by Dr. Leonardo Bonilha at MUSC that is seeking additional grant funding for epilepsy research from the National Institutes of Health.

This torrent of potentially life-changing research has done something else, too. It’s opened doors for computer science students at the College to have a front-row seat to this cutting-edge work. Matt Adamson ’18 and Eric Hofesmann ’17 have both worked alongside Munsell in the Machine Learning Lab after taking his course in data structures and algorithms.

Adamson, who majored in computer science, has spent the last two years working with Munsell to refine machine learning algorithms that can predict cognitive ability in children using brain scans. It’s an opportunity that he never thought he would get as an undergraduate student.

“I always thought that machine learning was kind of a black box – something that only Ph.D.’s could do,” he says. “Dr. Munsell showed me that someone as young as a sophomore can use machine learning to make valuable predictions in the medical field.”

Hofesmann’s research with Munsell on using brain connectivity patterns as a neurological fingerprint took the “Best of the Best” award at the School of Sciences and Mathematics Undergraduate Research Poster Session in 2017. The experience, says the physics and computer science double major, inspired him to pursue his doctorate in computer vision at the University of Michigan, where he is currently a research engineer.

“Working on that research project with Dr. Munsell was by far the most important thing I got out of my undergraduate experience,” says Hofesmann, noting that he’d never even heard the term “machine learning” before he started working with Munsell. “Once you actually dig into those types of complex problems, you learn way more than just memorizing something for class.”

The experience for students is invaluable, says van Delden.

“In his lab, Dr. Munsell and his students work side by side to understand and implement complex algorithms,” says the computer science department head. “This is technical work that involves a lot of student mentorship on Dr. Munsell’s part so that they can contribute to the project. His efforts have a big impact on our students.”

IRON WILL | Maddy Munsell giggles with delight as her father slowly pushes her wheelchair along the winding asphalt of the James Island Expressway. A mess of blond, curly hair bounces on top of the child’s head as she reaches out a small hand, fingers wiggling, trying to reach the brown-and-white dog that’s trotting by with another runner during the Charleston James Island Connector Run.

“She loves it,” says Munsell of his youngest daughter’s delight of racing with her dad.

The visible pleasure of the 5-year-old girl dances across Munsell’s tired heart as he makes his way to the other side of the bridge, feet clomping as he steers Maddy up the span. He knows she’s one in a million. And that’s why he decided to start running four years ago.

“I basically said to myself, ‘What am I sitting here doing?’” reflects Munsell. “She wasn’t walking, and I just felt like I had to do something. I’m doing it because she can’t, and in a much larger picture for those of us who are healthy, we should thank our lucky stars. And, we should get out there and be active for the people who aren’t so fortunate.”

His epiphany came on the eve of Maddy’s first birthday as he and his wife confirmed that their little girl had AHC. Faced with an uncertain future for his daughter, Munsell needed to do something for Maddy, for himself. So, he decided to run. And swim. And bike.

In 2014, the computer science professor, who admits he was more than a little overweight at 260 pounds, partnered with Chris Bailey ’12 (M.S., M.P.A. ’15), associate director of strategic initiatives and communications for the Honors College and a coach with the endurance fitness company Without Limits, and Catherine Hollister with Blue Sky Endurance, to begin the slow, grinding process of training for a triathlon. It would take two years of grueling workouts before he would be ready to tackle his first half-ironman event.

“A half-ironman event requires a high level of commitment,” says Bailey.

“I’m always impressed with his ability to keep up with seven to eight hours of training a week while balancing his career, his family and all he does to give back to our community.”

Since then, Munsell has gotten involved with Racers for Pacers, an organization that raises funds to provide running chairs for children with disabilities to participate in races. And, when his oldest daughter, 10-year-old Abbylee, wanted to start doing triathlons, the father-daughter team joined Palmetto Tribe, a triathlon training program for youth.
At each race or triathlon, the proud dad, who often holds fundraisers for AHC research before these events, wears what he calls his “uniform,” a trisuit that says “One Mission END AHC,” an outfit he’ll don this summer when he takes on a half-ironman in Santa Rosa, Calif.

“We have an incredible supportive running and triathlon community,” says Hollister, “through fun runs and raffle tickets together we’ve raised hundreds of dollars to support the One Mission END AHC campaign.”
Melissa Munsell says watching her husband find meaningful ways of bonding with his daughters amid the sometimes crushing weight of Maddy’s AHC diagnosis has been inspirational. “It got him way more involved in the lives of the girls,” she says. “It kind of all fell together.”

These days Maddy can walk a little on her own with the help of a walker and ankle braces. She can say “Dada” and “bye-bye,” among a few other words. She struggles with a lot of limitations compared to other 5-year-olds, but that only makes Munsell work harder and race faster.

In March 2017, Munsell, Maddy and Abbylee, all participated in the Kiawah Island Triathlon event for children. As Abbylee competed alongside her sister, Munsell carried Maddy across the swimming pool, pushed her on a special bicycle he made for her and guided her as she crossed the finish line with a walker.

The trio finished the race exhausted, but happy. The scene was kind of a metaphor for the roller coaster ride of highs and lows that come with Maddy’s illness. With the bad, AHC has also brought threads of good.