The new brain aneurysm detection system involves advanced imaging scans combined with analytics technology developed by Mayo and IBM. The analytics use computer algorithms to automatically turn what would normally be viewed by radiologists as 2-D images into slices of 3-D images that are analyzed for indications of widening blood vessels, said Bill Rapp, IBM chief technology officer of healthcare and life sciences and co-director of the IBM medical imaging informatics innovation center located at Mayo's Rochester, Minn. campus.
Mayo estimates that about one in 50 Americans has an unruptured brain aneurysm, an abnormal blood vessel bulging in the brain. However, 40% of patients with aneurysms that rupture die. Aneurysms often occur in patients who have had a stroke, traumatic injury, or a family history of the problem.
The use of the computer analytics with images from non-invasive magnetic resonance angiography tests can help Mayo doctors detect aneurysms without more invasive procedures, such as injecting dye into patients' brains, a procedure that can increase neurological risks.
By using scanned images from hundreds of patients, the algorithms were "trained" to detect possible aneurysms, Rapp said. Those suspected aneurysms are highlighted by a box that appears in the brain scan images, which are then investigated further by the radiologists, said Rapp.
"The radiologist does the final call," to determine the diagnosis, he said.
Workflow software helps to do these advanced readings within minutes of a MRA test, which potentially saves lives of patients with serious aneurysms in danger of rupturing. Since launching the project last summer, the analytics has been used on 15 million images from thousands of patients, already helping to save lives, said Rapp. Mayo would not disclose estimates on how many deaths were prevented using the technology so far.
"Today, you'd have an MRA if you are showing signs of an aneurysm," such as having symptoms like headaches or a recent stroke, said Rapp. Sometimes people have potentially deadly aneurysms but don't exhibit symptoms until too late.
Moving ahead, the same kind of analytics approach using images, combined with metadata and other information from patient e-health records, could help Mayo researchers predict patients most at risk for specific kinds of aneurysms -- as well as other diseases, such as cancer -- based on family history, whether individuals smoke, and other factors, said Rapp. The technology could also be used to help detect aneurysms that occur in other places of the body, such as the liver, he said.
Such research and analytics capabilities will be available to authorized Mayo scientists via a private cloud moving ahead, said Rapp.
Also, that cloud infrastructure allows Mayo to provide patients with their radiological images and reports so that patients can add the information to e-personal health records on Web-based platforms such as Google and Microsoft HealthVault, said Rapp.
IBM said the aneurysm detection system uses algorithms developed by Mayo researchers that are executed on IBM WebSphere Process Server to model and orchestrate the automated workflow. Images are stored on IBM DB2 for Linux and Windows data service and workflow logic is run on IBM System x servers and IBM storage.
IBM and Mayo have been working together on a number of health IT initiatives, including personalized healthcare, data warehouse, and data mining projects since 2001, said Rapp.
The work with Mayo to help detect brain aneurysms isn't the first time IBM has collaborated with a leading healthcare provider for discoveries related to aneurysms. IBM has also partnered with Cleveland Clinic in a personalized medicine project to help improve outcomes for patients suffering abdominal aortic aneurysms.