The call came from out of the blue. It was a normal, busy day at the General Electric office in Philadelphia.
“He said something like, ‘I’m calling from the National Library of Medicine,’” recalled George Thoma, PhD, unsure of what he had heard.
“The National Laboratory of what?” Thoma asked.
“No, the National Library of Medicine,” the caller said.
Thoma’s reaction: Why is a library calling me?
Good question with a good answer.
Earl Henderson, the former chief of the Communications Engineering Branch of NLM’s Lister Hill Center, was in touch because Thoma had expertise the Library needed.
“The Lister Hill Center was engaged in a very large telemedicine experiment to connect NIH and major medical centers with clinicians in several sparsely populated states,” explained Thoma. “This was before the internet, so long-distance communication was very difficult unless you used satellites.”
At the time, Thoma was designing “earth stations”—the equipment on the ground that processes communication signals through satellites—at General Electric.
“It was almost the same job, so I thought, hey, I’d like to move to DC,” recalled Thoma.
Thoma, the current chief of the Communications Engineering Branch of NLM’s Lister Hill Center, initially planned to stay a year or two.
The year was 1974.
The right stuff
Thoma came of age in the late 1950s and early 1960s. “This was a period of tremendous excitement for the space program,” said Thoma.
He wanted in on the action.
“The space program had a lot of spinoffs. You could work in rocketry or satellites, which could be used for navigation and communications,” said Thoma. “I was interested in communications so electrical engineering was the logical thing to major in.”
If his choice for a major seemed logical, his choice of college was a bit unusual for someone who grew up in Charlotte, North Carolina. It also represented a different career path from his parents, who were immigrants from India. Thoma’s mother was a physician and his father taught philosophy. On a chance encounter, his father met a dean from Swarthmore College and thought the school would be a good match for his son.
“It was serendipity,” said Thoma. “Originally started by Quakers, Swarthmore is known for liberal arts, but ever since the 1850s they’ve always had an engineering department, although back then it was referred to as the practical arts.”
Upon graduating from Swarthmore, Thoma continued his studies at the University of Pennsylvania, where he received his PhD in electrical engineering. “My dissertation had to do with satellite-based navigation systems that would be detected by ships or aircraft on earth,” he said. “It’s hard to believe now that back then a ship would leave New York and maybe for only 100 miles have real navigation data. Satellites could provide information over huge distances that would make a big difference in navigation for ships and airplanes.”
His chosen profession suited Thoma, who said, “It’s lifelong learning because technologies change and dramatically so.”
Upon graduating, Thoma stayed in Philadelphia, first designing navigation equipment for a small, high-tech start-up and then accepting a job at General Electric (GE). He said, “It was an exciting time because GE was building satellites and earth terminals.”
But back in 1974, the opportunity to help build a telemedicine research program at NLM was potentially more exciting.
Meeting the challenge of telemedicine
“I put everything in my Volkswagen Beetle—all my possessions fit—and beetled on down,” Thoma said about his move from Philadelphia to Bethesda.
For the next five or six years, telemedicine was Thoma’s life.
Satellites and terminals needed to be designed and built so clinicians in remote areas could communicate with major medical centers.
“Until that point, earth terminals—the things that send and receive signals to satellites—were massive. There was one in West Virginia for instance with an antenna that was 100 feet in diameter,” he explained. “That wouldn’t work for a small remote health clinical in the middle of Alaska.”
A different type of communications system was needed.
“We faced tremendous challenges in the design of the amplifiers and transmitters that send signals up and the low noise receivers that receive signals back from satellites,” he said.
Thoma and his team met the challenge.
“We proved that remote regions could connect for health care purposes,” explained Thoma.
The earth terminals they built across several sparsely populated states—Washington, Alaska, Montana, and Idaho—allowed remote clinics to link to NIH as well as major medical centers. NLM even had a studio, thanks to his colleague, Jim Main.
The project was a success.
Urgent tasks of the day
Thoma was ready to take on “other urgent tasks of the day.”
In the early 1980s, that meant both preserving documents and making interlibrary loans more efficient.
“The acidic paper in books printed in the mid-nineteenth century on up was falling apart. Of course, you could microfilm the material, but the problem with microfilm was that it’s not easy to use and you can’t share it very well,” explained Thoma.
The solution was document imaging.
This required new technology.
“We had to build document scanners from scratch,” explained Thoma.
And build them they did—right on the 10th floor of the NLM’s Lister Hill Center building.
Actually, Thoma and his colleagues built both the hardware and the software, but most of their efforts were focused on hardware. “You could open a book and have the pages scanned and then stored on what they called optical discs,” said Thoma. Those discs were 12 inches in diameter, but they did the trick, running on the computers of the day.
They were also cost efficient. Back in the early 1980s, a scanner or printer could cost a few hundred thousand dollars, explained Thoma. “We built it for much less, maybe a few thousand dollars.”
The new technology helped in other ways. “There was an interest in electronic ways of doing interlibrary loans,” explained Thoma. They built a system that not only allowed for scanning, it could automatically send electronic files from one library to another by fax.
These technologies also helped Thoma and his colleagues, particularly Dr. Daniel Le, develop the Medical Article Records System, an automated system that extracts bibliographic text from journal articles, in both paper as well as electronic forms. “That system is still in place, except in its third generation,” said Thoma.
From document imaging to medical imaging
After using technology to improve document imaging, Thoma and his team shifted their focus to medical imaging by the mid 1980s.
Currently, Thoma and his team are tackling the complexity of medical imaging with real-world applications and implications.
“I’ve always been grateful for the colleagues that I’ve had all these years,” he said. “They have been very dedicated, very serious, brilliant people with not just the drive to accomplish these tasks, but the education that supported it. I’ve been extremely fortunate.
“For example, some of my colleagues are working on a project with the National Cancer Institute to detect cervical cancer on histology images,” said Thoma. Their algorithms identify the epithelium in the image and then use selected features to detect the degree of cancer present.
Two other medical imaging projects developed by Thoma’s team have “gone global.”
One is an effort to make it possible for more Kenyans to get tested more quickly for tuberculosis (TB). The research done in image processing and machine learning in the lab yields algorithms that detect TB and other lung diseases in chest X-rays. These algorithms are implemented in a practical system installed in trucks that travel throughout rural Kenya to more efficiently and less expensively test citizens for this infectious disease.
The other global project concerns malaria. “We also use image processing and machine learning in this project to detect cells in a microscopic image,” said Thoma. This work has resulted in an automated system to detect and count parasites in blood films for a smartphone app in collaboration with NIH’s National Institute of Allergy and Infectious Diseases (NIAID). The app is being tested by other collaborators in Thailand and Bangladesh.
Thoma’s colleagues have also developed OPEN-i, NLM’s novel system for searching and retrieving images and relevant information from the medical literature in PubMed Central. “You can even query by images. You can put in ‘X-ray of broken ankle’ as a search term and get a whole bunch of broken ankle pictures with descriptions,” he explained. “It’s a public facing system with 10,000 users a day that allows people to search for over four million images, including all the illustrations in all the open access journal articles in PubMed Central.”
Dealing with data
“In a sense, we’ve been dealing with data analytics and large stores of data, meaning big data, for quite a while,” said Thoma. “Many of our projects have large numbers of images, which after all are data, or electronic health care data. There are vast amounts of data here, and we’ve been using these to extract information from them through algorithms that we build.”
Looking to the future, he added, “We’re actively bringing in more data sources, collections of data in different medical subspecialties. Our role really is to see if text understanding and image understanding can be done in a productive way that contributes to the health and well-being of the public. That’s really what we’re aiming to do.”
The future is bright indeed.
Why he stayed
“I have to say the National Library of Medicine has been a wonderful place to challenge yourself but also to feel supported,” says Thoma. “And I have to repeat that I owe a huge amount of credit for our success to my colleagues.”
That’s part of the reason why Thoma turned what was going to be a two-year gig into a career that has lasted more than 40 years and is still going strong.