Behind the Breakthroughs

By the time a patient has a tumor removed, they have been through the proverbial wringer. After the diagnosis, life is a series of appointments, waiting rooms, scans, and worst of all, worry. After surgery and being wheeled into recovery, the ordeal is not necessarily over. Patients can face another lingering worry: did the surgeon get it all? Tissue samples are sent to a pathology lab, where pathologists aim to determine whether cancerous cells remain at the margins. If the results are unclear, a patient may face additional surgery and everything it entails.

Margin Diagnostics, or MarginDx™, brings a fusion of microscopic imaging and artificial intelligence (AI) into the active operating room to answer such questions while the patient is available, so surgeons can act. The imaging technology at the center of MarginDx was not new to Maria Jaromin, interim director of the Healthcare Innovation Program Office at the National Center for Supercomputing Applications (NCSA), part of the University of Illinois Urbana-Champaign. She did not build it. She does not run it. But years earlier, before she worked at a supercomputing center, it sat in the biomedical imaging patent portfolio she managed for the university.

Today, several NCSA teams collaborate in support of an Advanced Research Projects Agency for Health (ARPA-H) project led by Dr. Stephen Boppart, contributing expertise that spans UX design, software development, data pipeline engineering, AI workflow management, and advanced computing. Collaborations like this do not assemble themselves.

“It’s Not Organic”

Jaromin’s office, known as HIPO, connects health researchers, clinicians, and health organizations on campus and beyond with expertise from across NCSA. That expertise ranges across high-performance computing, AI, data science, software engineering, visualization, genomics, and more. NCSA’s technical groups build the systems and tools. HIPO does the work around them: defining needs, assembling teams, and finding the funding that keeps projects moving. Jaromin’s team works with researchers who arrive looking for resources and are not always certain what to ask for.

“There aren’t two cases that are alike,” Jaromin explains, noting that researchers range from experienced users of supercomputing and its myriad applications to people who “feel like they need a deeper technical background before engaging with NCSA.” She says it’s often not obvious how everything fits together.

“Translating between experts at NCSA and the researchers takes time,” she says. “We need to learn the language. We probably have different work styles. So it’s not organic. Things like this sometimes happen organically, but it’s not organic. And it requires a little bit of facilitation and a little bit of time.”

The time to get a researcher from initial contact to an operational state can be considerable. It is not unusual, she says, for a researcher to meet with her and then disappear for months, later to emerge with a project years later “because maybe the data wasn’t ready, or maybe they came to us too soon, or maybe the funding wasn’t there.”

In between, Jaromin says, there is a lot of listening. Researchers describe their science, the obstacles they face, and what they believe they need. Jaromin listens for underlying needs then matches them with the expertise available across NCSA’s network of experts, looking for precedents in projects NCSA has already done. 

“My job is to think about who has the expertise and who is going to be the right fit,” she said.

Jaromin’s own route to this station in her career was nonlinear, she says. She studied power engineering in Poland then went to Stockholm on a European exchange program and returned there for graduate school, modeling fluid flow and heat transfer in nuclear reactors at the KTH Royal Institute of Technology. She followed her husband’s faculty appointment to Illinois and continued in modeling and simulation. She enjoyed the work, yet other people’s work kept catching her attention. 

“I was always interested in other things that other people worked on,” she says. “I always felt limited by doing my own work, my own research. It felt very narrow.”

Jaromin recalls seeing “a big gap between the basic science that we do and how we translate that into the real world,” and she was determined to create the bridge. 

“I thought, well, I don’t know how to do it, but I need to learn,” she added.

With this insight, she earned an MBA and moved to the University of Illinois Office of Technology Management, where much of her work involved managing a technology portfolio that included medical and biomedical imaging. Later, she joined NCSA, where her experience came together as she assembled teams and helped others advance their research.

Guarding Sensitive Data

The researchers who bring Jaromin their questions typically work with sensitive, heavily regulated data, and health information systems control how it is accessed and used. 

To address this, NCSA developed the Advanced Computing Health Enclave. There, the Nightingale high-performance compute cluster was established for researchers working with regulated data, providing them with granular individual data rather than aggregated population data. Jaromin’s part was connecting health researchers and clinical collaborators to it and helping projects and collaborations get started. 

“While building and running a system like this is far from a trivial undertaking, NCSA is well positioned to rise to the challenge,” she says, noting the collaborative effort that went into building a Health Insurance Portability and Accountability Act of 1996 (HIPAA)-compliant cluster that gives her researchers access to everything from health records to voice recordings and video.

The Importance of the Work

MarginDx™ is the project Jaromin cites when asked why the work matters. Principal investigator Stephen Boppart, a Beckman Institute faculty member, developed the imaging technology. This ongoing collaboration among researchers at Illinois, NCSA, the Mayo Clinic, Eleuthra Photonics, and Carle Health, funded through ARPA-H’s Precision Surgical Interventions program, has a payoff that impacts tumor patients, in the moment, on the operating table.

A 2024 insurance claims analysis by UTHealth Houston, published in Annals of Surgical Oncology, found one-year reoperation rates of 21.1 percent in a commercially insured cohort and 14.9 percent in a Medicare cohort of women who had undergone breast-conserving surgery between 2017 and 2019. The collaborators behind MarginDx aim to dramatically reduce those numbers by giving surgeons microscopic information before the patient leaves the operating room and goes to recovery.

“This is an extremely challenging project,” Jaromin says. “It takes a tremendous collaborative effort. People across NCSA, the university, and many researchers come together to make it possible. Our collaborators suffer many sleepless nights solving difficult problems.” The hard work comes with a satisfying pay-off, she says. “How amazing is the improvement in the quality of life for the patient who doesn’t have to undergo a second surgery?”

The National Center for Supercomputing Applications on the campus of the University of Illinois Urbana-Champaign. The Healthcare Innovation Program Office connects health researchers, clinicians, and health organizations to expertise across the center. Photo courtesy of NCSA.

“I don’t think our work is really that visible on the outside,” she says. “When people talk about exciting research projects, the spotlight often falls on the principal investigator and the major contributors or the technical experts. There is a big group that does a lot of legwork.”

Jaromin believes curiosity is a key factor that unites the people involved in these projects.

 “I’m curious,” she says. “I want to learn more about other people’s work. I want to understand it, and understand the gaps, where the problem is, and how we can go around the problem, or solve it, or remove it.”

HPC Unites at SC26