Daily Bulletin 2017

Explore the Future of Radiology at the RSNA Technical Exhibits

Sunday, Nov. 26, 2017

Nuance Communications Develops Powerful AI Tools

Tarik Alkasab, MD


In today's evolving healthcare ecosystem, clinicians are adapting to changing payment models and the collaborative approach those changes demand. As part of this evolution, radiologists need to proactively review and redefine their roles.

New strategies and tools will be required of healthcare professionals as they migrate from their figurative silos to a more collaborative environment, and adjust to the demands of value-based care, doing more with fewer resources.

Among those tools, said Tarik Alkasab, MD, of Massachusetts General Hospital and Harvard University, will be artificial intelligence (AI).

"Radiologists really see hundreds of patients each day. Artificial intelligence tools can be an aid in automatically detecting and picking up on things and then documenting them," Dr. Alkasab said.

While AI is quickly becoming a topic of conversation in healthcare, Dr. Alkasab and Woojin Kim, MD, chief medical information officer for Nuance Communications, both said there are misconceptions about it.

"Some have tunnel vision when it comes to AI in radiology," Dr. Kim said.

He noted that when you attend radiology conference sessions associated with AI, they typically focus on image characterization, such as "detecting pulmonary nodules and intracranial hemorrhage."

In fact, Dr. Kim pointed out, "there is a much greater role that AI can play."

But, he explained, few people are aware of how artificial intelligence is already at work in every facet of their lives.

"They are surrounded by AI that does so many great things for them and they're not even aware that AI is doing it for them," Dr. Alkasab said.

With so much information and data that can be measured and with so many more partners to share it with, he said, "A radiologist is going to become a radiologist-plus.

"The job is going to be bigger, but it's going to be a job that radiologists can handle because they're doing it with the help of all these artificially intelligent assistants who are swooping in at just the right moment."

Every component of the imaging value chain — be it patient scheduling, protocol optimization, modality operations, image interpretation, actionable report creation, communication of findings to referring clinicians and patients, quality assessment, or patient safety and follow-up — will be augmented by AI.

Both doctors acknowledge that much of the promise of AI remains in the future, but the very near future. The algorithms that propel AI to assist radiologists are still being written.

"Radiologists have long been trailblazers of technology in healthcare," said Karen Holzberger, vice president and general manager of the diagnostic solutions division at Nuance Communications. "As early adopters, they have embraced innovations that improve their practice, enhance care and deliver better outcomes, starting with the introduction of picture archiving and communication systems (PACS) more than twenty years ago.

"The latest advancements in radiology embrace the power of AI to improve productivity, accuracy and remove non-reimbursable work. Yet, the full power of AI and machine learning will not be realized in healthcare until it is made widely accessible and is integrated into the radiologist's workflow in a way that's convenient and reliable. After all, transforming the delivery of patient care and combatting disease will require the most advanced technologies being readily available when it counts – at the point of care. The benefits and potential for profound and widespread impact on radiology, healthcare in general and society at large are real."

"How do we drive access and usage? How do we enable competence in the tools that are used every day?" Holzberger asked rhetorically. "It's by enabling delivery, access and usability via the systems where radiologists spend most of their time; or, in other words, at the radiologist's point of care."

It will be incumbent on companies such as Nuance to ensure that the products it supplies radiologists with, like PowerScribe 360, are enabled to meet the challenge of delivering AI tools and making them highly usable.

"Software solutions are going to have to be able to incorporate all the inputs and outputs from these neural network-inspired algorithms and other tools that are helping the radiologist," Dr. Alkasab said.

Tip of the day:

Collimation reduces patient and staff dose and improves image quality.  Attribution: Tyler Fisher and John M. Wait

The RSNA 2017 Daily Bulletin is owned and published by the Radiological Society of North America, Inc., 820 Jorie Blvd., Oak Brook, IL 60523.