RSNA 2018: Researchers turn to AI to protect imaging equipment from cyberattacks

Medical imaging equipment is highly susceptible to cyberattacks, putting hospitals and imaging centers at a serious risk of losing functionality of those systems and even having data stolen by an outside entity. This concerning issue is the focus of two studies being presented at RSNA 2018 in Chicago.

One study comes from a team of researchers from Ben-Gurion University of the Negev in Israel, who examined ways to improve the security of CT equipment. Hackers can manipulate a CT scanner’s behavior, they found, and this could ruin images or do significant harm to a patient.

“In the current phase of our research, we focus on developing solutions to prevent such attacks in order to protect medical devices,” Tom Mahler, a PhD candidate and teaching assistant at Ben-Gurion University of the Negev, said in a prepared statement. “Our solution monitors the outgoing commands from the device before they are executed, and will alert—and possibly halt—if it detects anomalies.”

Mahler and colleagues developed a system using machine learning and deep learning that detects when a “new, unseen command” is attempted. This system then alerts the necessary parties.

“In cybersecurity, it is best to take the 'onion' model of protection and build the protection in layers,” he said in the same statement. “Previous efforts in this area have focused on securing the hospital network. Our solution is device-oriented, and our goal is to be the last line of defense for medical imaging devices.”

The second study comes from a team at University Hospital Zurich and ETC Zurich in Switzerland, where they are working to prevent attackers from tampering with mammography results.

“As doctors, it is our moral duty to first protect our patients from harm,” said Anton S. Becker, MD, radiology resident at University Hospital Zurich and ETH Zurich in Switzerland. “For example, as radiologists we are used to protecting patients from unnecessary radiation. When neural networks or other algorithms inevitably find their way into our clinical routine, we will need to learn how to protect our patients from any unwanted side effects of those as well.”

The researchers trained an artificial intelligence (AI) network to convert images showing cancer to healthy images, to see if it was possible. Radiologists then viewed the images to see if they could tell they had been tampered with—and the radiologists could not.

Such an attack is still years away, Becker said, but it is something he and his team wanted to draw attention to with their research.