Radiology experts unpack their artificial intelligence (AI) use cases, sharing their learnings and insights around improving patient outcomes, getting systems up and running, and seeing the most value from radiology AI investments.
At this year’s virtual ASNR Annual Meeting, Elad Walach, CEO, Aidoc, and with expert panelists Dr. Roland Robert Lee, Chief of Neuroradiology at UC San Diego Medical Center; Dr. Marcel Maya, Diagnostic and Interventional Neuroradiology, Co-Chair Department of Imaging at Cedars-Sinai Medical Center; and Dr. Eli Konen, Chair of Diagnostic Imaging at Sheba Medical Center, discussed the future of radiology AI and how to get the most from AI implementation.
The webinar focused on some of the key value points of AI, how well AI operates within the radiology profession and how it streamlines workflows, and on some of the ways in which AI is being applied to different use cases and environments.
The use case for radiology AI
The algorithms from leading AI solutions such as Aidoc, offer medical institutions and practitioners a trusted and stable foundation from which to treat patients and manage workflows.
“All of our images are fed through the Aidoc algorithm for intracranial hemorrhage,” said Dr. Lee. “We use it uniformly for all of our non-contrast head CTs at UC San Diego Medical Center (UCSD). The solution is good at picking up subtle bleeds which is particularly valuable for the trainees.”
Lee points out that this is of immense value for the residents as they are already working long hours, working different rotations, and on a steep learning curve. With massive time limitations and managing multiple patients across different areas of medical practice, trainees benefit from another pair of eyes.
“Another value add is the use of the solution in organizing work lists and streamlining priorities,” added Lee. “When you get to work in the morning, there can be a number of unread studies, so I use Aidoc’s audit to go through the positives first. I also keep the system on during the day so as I’m working through the list, I can check the cases and prioritize my approaches.”
The system supports practitioners in organizing their work lists throughout the day. It allows for busy radiologists to maximize their time more efficiently, and to streamline patient care. The future of radiology AI lies not in simply sticking an algorithm into a system, but in leveraging its potential to improve use cases, to add value, to transform learning, and to integrate workflows more efficiently.
The future of radiology AI lies in workflow integration
Dr. Marcel Maya, Diagnostic and Interventional Neuroradiology, Co-Chair Department of Imaging at Cedars-Sinai Medical Center, uses the Aidoc stroke solution for large vessel occlusion and for c-spine fractures. He unpacks how important it is to ensure that any AI solution be capable of comprehensive integration into existing systems and workflows.
“We adopted the technology early, back in 2017, and intracranial hemorrhage was the first one,” he said. “We quickly moved on to cervical fractures, pulmonary embolism detection and large vessel occlusion. We had to be convinced of the value of adopting these solutions, and then we had to choose the right partner. I can’t over-emphasize how important it is that the solution works within your existing workflows of reading – even if you have the best system available, if it isn’t integrated properly it isn’t going to help.”
For Maya, this integration is not only crucial to ensuring that any AI solution actually delivers on its promised results, but to allowing for the teams to use the system in different ways, to get even more value from the technology and its capabilities.
“Different people use the technology differently, so you want a solution that works seamlessly within your workflows,” he added.
This is a view shared by Dr. Eli Konen, Chair of Diagnostic Imaging at Sheba Medical Center, one of the largest medical centers in Israel with a busy emergency room. His teams have found AI to be very helpful, for both residents and experienced radiologists.
“A key reason for the success in our department is how well it has been adopted,” he said. “It’s already become part of the normal radiologist workflow.”
The future of radiology AI
AI has the potential to elevate the influence that radiologists have on the whole care continuum by providing practitioners with robust and trusted support, allowing for the profession to evolve and adapt to changing patient and market needs.
“Radiologists are past the point of being scared of AI and recognize that they are in the best position to adopt the technology and move AI forward,” said Maya. “The challenge in the past has been that radiologists have not been able to promote our successes or advance our causes in fields that we’ve been developing. They’ve been co-opted by other specialties. This is a way of sitting at the forefront of change, in transforming clinical practice and investing into this field.”
It is patently false that AI is set to replace the radiologist. What’s actually happening is that the radiologist is leveraging AI to sit right on the cutting edge, using these tools to more effectively manage demands on radiologist time and resources. As Dr. Lee pointed out, the demand for radiologists is so high that there need to be more tools to help existing radiologists manage workloads. With AI, professionals can read more efficiently.
The future of radiology AI is not just in algorithms, but in helping radiologists transform departments into digital masterpieces that can integrate, compare, support and translate imagery across multiple touchpoints and departments. That said, the future of radiology AI is actually already here…
“It’s already very well adopted,” said Konen. “In the radiologist mind it is already part of workflow. It’s a reassuring and additional tool for experienced radiologists.”