SIIM #AskIndustry Recap: AI Platforms and Ecosystems – Where Are We Headed?

AUTHOR: Tom Hasley, MS, RT, Global Product Development FUJIFILM Medical Systems, U.S.A., Inc.

Nearly every year, Fujifilm has the opportunity to take part in SIIM’s #AskIndustry education program. Rather than the traditional in-person panel, the 2020 program took place in a webinar format, where I and four other enterprise-imaging thought leaders were selected to converse on the topic, “Artificial Intelligence (AI) Platforms and Ecosystems: Where Are We Headed?”

The discussion focused on a number of AI themes, including adoption challenges, workflow considerations, and the radiologist’s role in advancing AI technology across the entire healthcare ecosystem. Below are three critical takeaways from the conversation:

1. Radiologists continue to lead the AI era.

Though AI has started to proliferate into various -ologies, radiology continues to lead the way. To give you an example, at Fujifilm, we collaborate with a number of radiologists who have their bachelor’s degree in biomedical engineering. I’m always astounded to hear how many of these young professionals want to develop their own AI algorithms and participate in the curation of data, creation of ground truth, and validation of new algorithms to further advance the field. On a similar note, The American College of Radiology is always looking for strong AI use case submissions and is a great resource for radiologists who wish to further advance the AI field.

But radiology’s role in AI adoption goes far beyond departmental walls. Indeed, radiology is in a unique position to lead AI education for the entire healthcare ecosystem. That includes fellow radiologists, other clinical specialists, and, eventually, even patients. But for the technology to be fully trusted, the radiologist must be the primary user. It’s their education and real-world experiences that will then cascade throughout the entire clinical network.

 2. AI orchestration is essential to continued adoption.

The ultimate goal of AI is to make radiologists more accurate and efficient in their reading requirements. If an AI algorithm doesn’t seamlessly fit into their existing workflow, it runs a significantly higher risk of being rejected. The challenge is that each AI vendor is presenting its findings to the end user in a unique way. As a result, providers end up relying on a number of inconsistent workflows and eventually rejecting the algorithms that don’t complement their preferred reading style.

Similarly, another question radiologists ask our AI experts at Fujifilm is “do you immediately show AI results to downstream clinicians, even before the radiologist has the chance to vet the results and say if they agree or disagree with the findings?” Again, this is why AI orchestration is so critical to its adoption. AI must support the radiologist first and foremost. Otherwise, it will never be fully accepted.

3. Quantifying AI value continues to be a complicated endeavor.

It’s undeniable that a number of AI solutions have substantial value in today’s diagnostic arena. However, there must be substance behind these solutions beyond their foundational accuracy and efficiency benefits. For example, does the solution support today’s industry standards? Does it integrate into the existing healthcare ecosystem? Can it stand up to an unexpected regulatory audit? Being able to answer “yes” to these questions often speaks to the value of the AI solution aside from the standard workflow advantages.

Once AI’s value can be quantified and subsequently applied to a business model, a common follow-up question is who should ultimately pay for it? Should the radiologist, whom AI is helping to provide more accurate and efficient diagnoses? Should the hospital, as AI ultimately helps with its overall quality of care as an institution? Or should the insurance companies, as we’ve started to see through government reimbursement in certain instances? Only when AI’s value can be quantified and applied to a universal business model will its true return on investment start to take shape.

To hear more from me and other AI leaders in the enterprise-imaging industry, be sure to watch the full roundtable discussion here. Want to learn more about Fujifilm’s AI innovations? Reach out to us to schedule a remote demo of REiLI®, our AI platform for enterprise imaging.