UMMC radiologist’s software helps diagnose, measure liver disease
By Jack Mazurak
New software by a University of Mississippi Medical Center radiologist that detects liver disease on CT scan images could lead to earlier detection and treatment of cirrhosis
in potentially millions of patients.
Dr. Andrew Smith, assistant professor of radiology
, got the idea during downtime on a flight for a radiology conference a couple of years ago.
“We know that liver gets nodular in cirrhosis, and this is most obvious at the surface,” said Smith, director of radiology research and an abdominal imaging specialist. “I thought if we can define the edge in an existing CT image of the liver, we can look for and measure nodularity.
“I wanted something like the tool in Photoshop that can recognize and outline the edge of an object against a different background. Once we learned to detect the liver surface, we developed tools to measure the amount of nodularity.”
The U.S. Centers for Disease Control and Prevention
reports approximately 27,000-30,000 people in the U.S. die annually from cirrhosis. While those are end-stage cases, Smith said, earlier stages of cirrhosis might affect up to 3 million Americans, or 1 percent of the population.
Its most common causes include alcoholism, fatty liver disease or hepatitis C infection.
In many cases, it goes undetected until clinical symptoms appear. At that point, patients need more drastic and expensive treatment, sometimes including transplant. Since early-stage cirrhosis is treatable, earlier detection could prevent many advanced cases.
Scientists and physicians have long known that liver nodules, especially as they approach 1-2 millimeters in size, signal cirrhosis. But there was no way to accurately and reproducibly detect and measure them on routine CT scan images, so patients currently need an invasive liver biopsy to make the diagnosis.
“A liver biopsy is the bronze standard for diagnosing cirrhosis right now,” Smith said. “We call it bronze, not gold, because livers are heterogeneous and a biopsy is a tiny sample, so they don’t always accurately show what’s going on across the organ.
“detecting and measuring liver surface nodularity on CT scan images offers a much less invasive test and allows us to screen for varices and liver cancer, common complications of cirrhosis.”
To write the software, he enlisted Ohio-based software development firm ImageIQ
, a spin-off of the Cleveland Clinic
. The team incorporated algorithms to detect the liver surface and measure nodularity.
Smith and developers at ImageIQ have been fine-tuning the program since March, putting it through multiple revisions so it can be ready for clinical trials.
Using a plastic model of the liver – called a phantom – with areas of different-size nodules, he’s tweaking the software to work best across different brands of CT scanners, including GE, Siemens, Toshiba and Philips.
Now Smith will start processing hundreds of scans to validate the program, comparing the results with biopsies taken from the same patients.
“Fortunately at UMMC, many patients have a CT scan and biopsy the same day, so we know there’s no change in the liver’s condition between the two,” he said, explaining the retrospective study that will look back at hundreds of patients’ scans and biopsies.
In skilled hands, detection and measurements of liver surface nodules on CT scan images takes only a couple of minutes.
Beyond screening general patient populations, radiologists could use the program to stage patients with known hepatitis risks. In follow-up, physicians could use it to tell how well therapies for liver disease work.
It’s a research tool, too.
“It could be used in drug clinical trials, which would move us away from doing multiple biopsies on participants,” Smith said. “The fact that UMMC supports innovation like this is very important.”
Dr. Timothy McCowan, professor and chair of radiology, said he advocates such thinking among his department’s faculty and staff.
“Radiology is linked more tightly to technology than probably any other medical specialty,” McCowan said. “It makes sense to encourage forward-thinking people to push the boundaries.
“This research comes at a great time as we increase our institutional focus on liver disease with our new liver transplant program.”
Most of Smith’s research focuses on metastatic kidney cancer, the stage where renal cancer has spread to other organs, and advanced imaging techniques for other cancers.
The software fits in a broader category of advanced CT image processing, ways to pull out more and process more information from existing scans.
He registered the name Radiostics LLC as the technology startup he hopes to spin off UMMC later this year. With it, he plans to further develop the liver surface nodularity quantification software, launch prospective clinical trials and work on other ideas for advanced CT image processing he’s got cooking for detecting stroke, osteoporosis and cystic kidney and liver disease.
The company joins a growing list of biotech spinoffs at UMMC. Within the next few years, UMMC leaders hope to start construction of a new research building that will offer incubator space for start-up companies.
“This kind of innovation drives better care for patients, helps our education mission by giving our students more training opportunities and can lead to deeper research in understanding how diseases progress,” said Dr. Richard Summers, associate vice chancellor for research.
“Locally and statewide, biotech companies like Radiostics can create jobs and impact the economy. We’re glad Dr. Smith is on our faculty and we’ll continue promoting this kind of groundbreaking work.”
Smith said he’s focused on the road ahead for his new technology.
“If this works – and I believe it will – it will be a game-changer.”