Medical images are becoming more susceptible to targeted cyberattacks as artificial intelligence becomes more advanced, according to a perspective piece in Medscape.
These warnings are based on the idea that hackers could alter medical images, leading to misdiagnosis. Hackers could potentially take over machines that use ionizing radiation and, by applying advanced artificial intelligence, could gain access to a network's imaging database to modify patients' scans, according to research presented at the 2018 Radiological Society of North America and cited by Medscape.
What's more, physicians may not be able to spot whether images have been hacked and altered by AI, according to a separate study presented at 2018 RSNA cited by Medscape.
For the study, a team of researchers looked at 680 mammograms stored in a public database and, applying a type of advanced AI, added or removed cancerous features from the images. They added cancerous features to 362 healthy mammograms and removed cancerous features from 318 unhealthy mammograms.
Three radiologists with varying levels of experience reviewed the images, but none could "reliably detect the real images from those modified by the AI," according to Medscape.
"Imagine if you infected another nation's devices; change maybe every hundredth or twentieth image, which would cause mass misdiagnosis, economically damaging the nation and undermining trust in the health care system," said the study's lead author Anton Becker, MD, a radiology resident at the University Hospital Zurich in Switzerland. He added that although this type of attack is unlikely to take place in the near future, "we need to be ready; to see it coming."
To access the complete Medscape report, click here.