Imagine the amount of medical imaging data being collected every day in the United States through various equipment, such as ultrasound machines, emergency department radiology systems, nuclear imaging scanners and colonoscopes.
"Literally millions of images are being generated across the country" each year, said Kim Garriott, principal consultant for healthcare strategies at Logicalis, a global IT consulting and services firm. Below are seven things to know about how next-generation technology will affect medical imaging data, as explained by Garriott:
Informatics, AI and analytics for medical imaging need data volume to succeed. For example, it takes millions of images for AI to learn about disease patterns in the data, interpret the patterns correctly and then attempt to highlight potential trouble spots, Garriott said.
Algorithms analyze medical imaging data at the pixel level. In doing so, AI can learn that, in the past 500 mammographies, a type of pixel density noted in the images might always be of interest for further examination by radiologists, Garriott said. AI won't infer anything from the density, but instead will flag it for review.
Interest is growing in chest X-ray imaging analytics and informatics. Some vendors are looking into AI for Tuberculosis (TB) screening, too, Garriott said. The U.S. National Institutes of Health has set up a TB portal to collect images for further research using cutting-edge technology.
Kim Garriottprincipal consultant for healthcare strategies, Logicalis
Mobile scanners may change how medical imaging is used. To better detect strokes, the Cleveland Clinic in Ohio has purchased a mobile CT scanner in a vehicle resembling an ambulance. The scanner runs an algorithm and can assess whether a stroke occurred and the type of stroke. Images transmit to the main hospital from the vehicle. The ability to perform these measures in the field will save more stroke victim lives, Garriott said.
The cloud can serve as a useful archive for images. Lots of Garriott's consulting clients are at least evaluating cloud storage for images. "You can't keep it on premises," she said of medical imaging data. "You've got to move it" to the cloud.
Executives may not be ready for an enterprise imaging strategy. Enterprise imaging can be hard to sell to CFOs, for example, Garriott said, because at that level, spending is directly equated to improved patient outcomes. From that perspective, it is not yet easy to prove the ROI of an enterprise imaging approach. Nonetheless, Garriott said it makes sense to consolidate storage in a central point, such as a system-wide vendor neutral archive.
AI is not yet mainstream enough to threaten radiologists' jobs. "I don't see AI replacing radiologists … for many, many years," Garriott said. In the interim, AI will enable radiologists to focus on the most critical cases first, she added.