You can’t spell radiology without AI

This year’s annual meeting of the Radiological Society of North America (RSNA) has been dominated by talk of artificial intelligence (AI) – how radiologists should leverage it, how it can aid in diagnosis and how it can help usher in an era of precision health.

Artificial intelligence is so important, RSNA 2019 has dedicated an entire section of McCormick Place Convention Center in Chicago to it. This area, coined the AI Showcase, features exhibiting vendors with AI-specific solutions and a theater where experts are delivering AI-focused presentations.

Heck, RSNA even created a specialized social media hashtag around the topic: #RSNAI.

AI promise and limitations

I had the opportunity to attend one of the many AI-focused educational sessions at the conference. The session, titled Learning AI from the Experts: Becoming an AI Leader in Global Radiology, brought together a panel of radiologists and radiology professors to discuss the implications of AI on the profession. Panelists pointed out how AI is pervading almost every aspect of the imaging chain – from image analysis to decision support to reporting – but they also stressed that the technology is not a threat to replace radiologists any time soon.

“While AI algorithms can recognize patterns and make predictions, they don’t have an idea of common sense,” says Eliot Siegel, MD professor of radiology at vice chair at the University of Maryland School of Medicine. “AI will be a great spell checker and grammar checker for radiologists, but it won’t replace us. Humans with an understanding of medicine will still ultimately be needed to make judgements on image sets.”

That being said, panelists were quick to highlight the benefits AI can deliver to the field of radiology. For example, the technology can prove instrumental in helping drive down the cost of healthcare by streamlining workflows, automating manual tasks and increasing productivity.

Furthermore, AI can help provide interpretation guidance in areas of the world that lack radiological expertise. Matthew Lungren, MD, assistant professor of radiology at Stanford University Medical Center, supported this claim with studies that showed how non-radiologist clinicians improved their image-reading interpretation competency after receiving AI-guided assistance.

Good AI requires quality imaging data sets

Our own Chris Magyar, senior manager of product management at Hyland Healthcare, also got into the AI act, delivering a presentation on the subject at the RSNA Innovation Theater. Magyar’s presentation, titled AI and Machine Learning in an Enterprise Imaging Environment, outlined several challenges surrounding AI and machine learning when applied to medical images.

One of the biggest challenges is providing secure access to the massive amounts of medical imaging data that health systems have accumulated to date. According to Magyar, this imaging data comes in a variety of formats and is largely stored in disparate silos. Consolidating these images into a single source and anonymizing the data for AI and machine learning purposes can be tedious and time-consuming.

Magyar demonstrated how enterprise imaging technologies such as vendor neutral archives and enterprise viewers can facilitate image integration and de-identification processes. He even shared how leading healthcare facilities, such as Northwestern Medicine and Yale New Haven Health, are using Hyland Healthcare’s vendor neutral archive solution and enterprise and diagnostic viewer to facilitate AI initiatives for early lung cancer detection and liver lesion segmentation respectively.

According to Magyar, healthcare providers achieve the best clinical outcomes when the highest quality data sets are accessible by AI and machine learning algorithms.

“Our customers rely on Hyland Healthcare enterprise imaging solutions and expertise to not just manage, but curate their medical data for clinical and research purposes,” he says. “They use our AI-powered visualization tools to enhance the diagnostic confidence of imaging professionals around the world, help AI streamline clinical workflows and automate repetitive tasks.”

To learn more about Hyland Healthcare’s role in AI as it relates to medical imaging, visit booth 4300 in the South Hall at RSNA 19.

Ken Congdon

Ken Congdon

Ken Congdon is the team lead of solution marketing at Hyland. His mission is to develop engaging content that educates healthcare providers and payers about potential solutions to their most... read more about: Ken Congdon

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.