Cancer trove makes move to UAMS informatics arm
They came packed up in a car.
On the 350-some miles from Washington University in St. Louis to a new home at the University of Arkansas for Medical Sciences in Little Rock. The old-fashioned way. The most feasible way, really, to move them in a month.
The Internet was too slow to transport the 40-terabyte cancer imaging archive of pictures, videos and other documents — some 4,000 times the size of the connection.
“It’s not like moving a PDF or a Word document,” said Rhonda Jorden, UAMS’ vice chancellor for information technology and chief information officer. “We have a 10-gigabyte connection, but even that is a very slow process to move that large volume of data.”
The archive is an information repository of sorts — the largest of its kind. And it left St. Louis for the first time as the team that created it moved to bolster UAMS’ Department of Biomedical Informatics, a new arm within the College of Medicine.
The field is focused on the complete set of human genetic information, called genomes. The newer part of the field, though, adds computer science in the medical realm.
“My colleagues and I at various institutions have discussed this many times: What’s the difference between computer science and biomedical informatics?” said Fred Prior, the chairman of the newly established department. “And my answer is: Biomedical informatics is in a medical school, and the people who come out of the degree programs understand biomedical research first and clinical operations. Computer scientists don’t really look
it that way, right? And that’s the real difference.”
Prior and four others — computer scientists and biomedical informatists — from Washington University created the repository in 2010 after winning a bid from the National Cancer Institute. The group pulled data like radiology images for all solid tumors stemming back 12 to 15 years from projects including the National Lung Screening Trial and the Cancer Genome Atlas.
With the lung trial, data would include all the information about every single person who participated: work history, smoking history, medical history, whether the person had cancer or developed cancer, whether the person had any other diseases. Each person was followed for 10 years.
The trick, Prior said, is managing all the data to ensure integrity of the information.
Understanding the datasets is something that has grown from the business sector, said UAMS College of Medicine Dean Dr. Pope Moseley. The “genius of WalMart,” he said, isn’t just that it understands, say, potato chips, but also what brands sell better and what flavors? The corporation knows how to meet its needs, he said.
“That’s the same kind of big data problem when you wonder what happens to a patient with a certain kind of cancer who has a particular look on an X-ray or CAT scan … who has this certain type of genetics,” he said. “Biomedical informatics is the science of developing and looking at large biomedical databases. That database could be the genes of people with certain diseases, what happens to the outcome of patients if you implement a new service; it could be looking at all the bacteria that live in your gut and how it changes when you develop diseases.”
Now, nestled away in an always air-conditioned room — a temperature gauge will read somewhere between 68 and 70 degrees — the archive stores radiological and pathological images, along with anonymous clinical trial information. And movies, too.
“It used to be a standard CT [computerized tomography] scan was 40 images. Now, it’s 400. High resolution,” Prior said. “Used to be, you’d go to the CT scanner and move the table, take an image, move the table, take an image. Now, you go on the scanner, the table moves continuously and the images are generated continuously as you go through. And basically, we’re taking movies.”
The movies show how to apply different computerized algorithms to identify the boundary of a cell and its genetic material.
“And each of the algorithms gives you different answers,” Prior said. “So you combine the answers in creative ways to come up with the best estimate.”
In part, the archive allows cancer researchers to repli cate studies, a way to help find best practices in treating certain cancers, he said. Researchers can also use the archive to develop new algorithms or software tools that scan through the radiological or pathological images and give doctors hints about what might be a cancerous tumor, he said.
It’s also a tool for medical students to familiarize themselves with health data. UAMS is growing the fledgling department, and officials are hoping to get medical students and biomedical informatics students to work in teams.
“The informatics component is getting to be ridiculously complicated,” Prior said. “And biomedical informatics is coming into being as an academic program because we recognize the need for team science and to have people on the team that understand medicine, understand the biomedical research questions, but also bring with them the computer science and data science techniques for addressing those questions.”
Introducing medical and doctoratal students to health data will allow them to make better decisions about patient care, Moseley said.
“It’s all part of precision medicine initiative,” he said.