Skip to main content

Five Key Takeaways from RSNA 2022

Flywheel had a bigger presence than ever at RSNA 2022, and our team benefited from in-depth conversations at our booth, attending talks, and catching up with customers. Here are five things we’ve been talking over since our time in Chicago wrapped up: 

 

  1. "AI" (or, more accurately, machine learning, neural networks, or deep learning) is still taking radiology by storm. After an initial period of hype (alongside some confusion), the cream is starting to rise to the top: researchers and commercial enterprises are getting it right by utilizing enriched data sets for training and validating ML algorithms, cautiously rolling out ML applications, and continuously monitoring results.

  2. As ML matures, collaboration—and provenance supporting it—is becoming more and more important. Organizations need to access vast amounts of data, understand where it originated, and the curation and computation it has been subjected to.

  3. There is growing interest in uniting multimodal data for research. Combining imaging data with radiology reports, genomics data, and clinical measures is becoming more important for comprehensive analysis and, eventually, solutions serving the biomedical research community.

  4. The ongoing shortage of radiologists underscores the stakes for the industry. There are opportunities for ML technology to help radiologists improve efficiency and accuracy while enabling collaboration to ease the labor gap. However, tools need to be built with the right data and get through regulatory approval in a timely way. 

  5. The FDA is stepping up to the new reality. Radiology and ML practitioners are putting pressure on the FDA to adapt to the realities of ML: approval needs to be faster, and regulators are now allowing for applications that learn and change.

A common theme among these trends is the pressing need for advanced data management enabling end-to-end research solutions. Flywheel’s team had many conversations at RSNA about how we’re supporting advancement in radiology and continuing to develop infrastructure for cohort discovery, data sharing, federated learning, automated processing, and more. In 2023, we’re looking forward to bringing even more exciting offerings to the radiology community.