Hospitals Looking to Outsource Medical Imaging Workload to Artificial Intelligence
With a high-volume workload staring down radiologists day after day, some hospitals are looking to lighten the load by outsourcing the work of examining medical imaging to computers running artificial intelligence and machine-learning algorithms, according to an article in Modern Healthcare.
These computer systems, according to the article, are “trained to find patterns in images, identify specific anatomical markers,” and even “spot details the human eye can’t catch.” Trials of early versions of the algorithms have shown them to be both accurate and fast. That said, the technology is still in its infancy and is not without challenges, especially when it is first launched. According to the article, artificial intelligence algorithms need to “learn,” which means that the algorithm must be trained to recognize various ailments and the nuances of the imaging that indicate them.
While the introduction of artificial intelligence systems to the workflow might make some uneasy, the goal is for the technology to share the load with physicians and free up their time from more tedious tasks, while increasing consistency and accuracy. Human radiologists would still have the final word, reading and making sense of the data produced by the computer systems. The change would be similar to the affect computer-assisted coding is currently having on the coding profession.
Zebra Medical Vision currently relies on “a vast supply of medical case data to train its algorithms so radiologists can find what they’re looking for—and what they don’t yet know they’re looking for—more accurately, more quickly and more consistently,” according to the Modern Healthcare article. “Radiologists are able to deliver better care at lower costs, and patients get the benefit of improved diagnoses,” said Elad Benjamin, co-founder and CEO of Zebra Medical Vision, to Modern Healthcare. Zebra Medical Vision has a unique algorithm dedicated to each finding. For instance, the algorithm that might be used to diagnose emphysema would not also be used to diagnose a brain issue such as stroke.
Roadblocks currently lining the road between this artificial intelligence and hospital doors include regulatory roadblocks with the US Food and Drug Administration, proper training for clinicians to effectively use the technology, and integration into existing workflows.
Sarah Sheber is assistant editor/web editor at Journal of AHIMA.