It will come as no surprise to attendees of RSNA 2018 when Vijay M. Rao, MD, officially kicks off the proceedings by homing in on AI and machine learning during her president’s address on Sunday, November 25, at 8:30 a.m. After all, that topic was the talk of McCormick Place last late November and, since then, it’s only grown in importance to the profession.
“We are challenged to provide imaging services that are faster, safer and affordable, as well as information that is quantitative and precise,” Rao told RBJ in a phone interview conducted during the runup to the event. “I think these emerging [AI] applications will make radiologists more effective and allow us to make a more meaningful contribution to the world of personalized and precision medicine.”
Read on for more from our conversation with Rao, who chairs both the radiology department at Sidney Kimmel Medical College of Thomas Jefferson University and enterprise radiology and imaging at Jefferson Health in Philadelphia.
RBJ: Without giving away the store on your president’s address—“How Emerging Technology Will Empower Tomorrow’s Radiologists to Provide Better Patient Care”—how do you foresee AI technologies equipping radiologists to improve care while reducing costs?
Vijay Rao, MD: These new technologies will help make us more efficient in our daily routines, because they will improve our workflows. The greater efficiency will free up time for radiologists to better position ourselves as visible and integral members of the patient-care team. We’ll have more time to be more patient-centered, to talk directly to patients and explain their imaging findings to them. Traditionally, we have left that responsibility to our clinical colleagues.
As we start taking on that responsibility more and more, patients will recognize us as key physicians. Who understands the value of appropriate imaging better than radiologists who have trained in this fi eld for a minimum of five years before they started practicing?
On the clinical side, do you expect machine learning to take over a lot of repetitive interpretation tasks?
Yes, but more than that. Some of the repetitive work we’ve always done takes a lot of time. On the clinical side we will definitely be augmented—actually, I don’t really like the word augmented—let’s say assisted by AI. On the workflow side, we spend a lot of time on tasks like protocoling cases, image quality control, hanging protocols, etc. AI applications can assist with such tasks through machine learning algorithms.
Smart technology will be built into imaging equipment so that problems with image quality are detected while the patient is still on the [scanner] table. The protocoling of cases will be automated based on relevant clinical and lab information in the EMR and built into clinical pathways.
If you look at the life cycle of a radiology study within a radiology department, AI applications can assist with decision support for appropriateness, performance, quality control and triaging of urgent cases, among other things. It can also help with scheduling, resource optimization, etc. In fact, I almost consider the workflow bucket to be as valuable as—if not more valuable than—the image-interpretation bucket, where AI will help with pattern recognition, early detection of disease, etc.
It would stand to reason from your point that, by streamlining workflows, AI could improve imaging economics as well.
Absolutely. The power of AI applications for improving revenue cycle management is tremendous. Using the power of analytics, we’ll be able to quickly determine the carrier’s policy, the co-pay, the deductible and the balance on the bill. That will be very useful for the patient and also for the provider organizations.
Honestly, we do a lot of work that we don’t get paid for. AI can help us optimize the revenue cycle.
Speaking of money, recent surveys have shown that Millennials tend to prioritize job satisfaction over monetary rewards. What aspects of radiology could use more emphasis by residency recruiters?
A couple of thoughts. Millennials are an extremely tech-savvy generation. They embrace advances in technology, including AI, which would make radiology an attractive specialty. It is a very fertile field, one that provides opportunities for creativity, innovation, process improvement and development of products to improve patient care.
Secondly, radiology is evolving into a specialty with a wonderful work-life balance.
One can be more patient-facing or not, depending on personal preference. I think having choices can be very attractive. In the past, some medical students said no to radiology because they wanted more patient-facing time.
Looking ahead, as we mentioned earlier, there is going to be more time available for radiologists to spend with patients. As we move toward a value-based practice model and become more patient-centered, our specialty is becoming more attractive to Millennials. There is more opportunity to have a sense of personal satisfaction and a rewarding career by contributing more directly to patient care.
Are you speaking here mainly of diagnostic radiologists?
Yes. And of course our interventional radiology colleagues have been providing direct patient care and functioning as clinicians. Given rapid advances in noninvasive therapeutic interventions, a career in interventional radiology is attracting the best and brightest medical students to the field. Here at Sidney Kimmel Medical College, we have started a separate ACGME-approved residency in IR in the department of radiology. Medical students can now go directly to an IR/DR residency. That’s becoming quite popular with medical students. It’s rewarding to be able to provide direct care, including treatments, to patients. Interventional oncology is a very exciting field right now.
How is this playing out at Sidney Kimmel?
We’ve had quite a bit of interest from medical students. We certainly haven’t seen a drop in the number of medical students applying for radiology training. In fact, our numbers are going up. Two to three years ago there was a little drop. There was some concern that not enough medical students were going into radiology. But I think we are past that. We are definitely seeing an upswing in the number of applicants.
Across the U.S., what would you say has been the single most formidable challenge radiology has faced this year during your time as RSNA president?
I would say two things. Technology advances and imaging utilization along with the accompanying rising healthcare costs. However, these challenges are also our biggest opportunities. The hype and fear around artificial intelligence replacing radiologists fueled by sensational headlines is diminishing, fortunately. Many leaders in the medical field outside of radiology started telling medical students not to go into radiology because there’ll be no jobs. And that’s because of news headlines such as “Machine learning can diagnose this or that better than radiologists can,” along with venture capitalists predicting radiologists will become obsolete. This has presented an interesting challenge, but I think we’re getting over that hump.
You see a lot of people now talking more realistically about what machine learning can and can’t do to replace humans. The growth in understanding helps, doesn’t it?
Yes. The power of AI technology is real but can be overstated. I always remind people that airplanes have been using autopilot for a long time. But would you fly on a plane with no human pilot in the cockpit?
Neither would I. And would you really be willing to have your medical diagnosis made by a computer interpreting images without an expert human intervention? I do not believe that AI will replace radiologists. I expect radiologists to embrace the technology. And it’s going to influence the practice of radiology in a major, major way. So we do have to adapt as we continue to embrace this technology.
It’s interesting that you prefer assist over augment.
Assist implies that AI will share my work and reduce the amount of time I will spend on workflow tasks and interpretation. Augment implies that it makes my work more meaningful by providing additional information like precise measurements, data analytics, etc. In reality, AI will assist and augment radiologists. Ultimately AI is going to help us make imaging faster and safer—and more quantitative, precise and affordable.
What do you expect will be the profession’s single greatest opportunity heading into 2019 and beyond?
I would say global collaboration and support for advancing development and implementation of artificial intelligence applications to drive excellence in patient care, ease the shortage of radiologists and physician burnout. During my travels this year as RSNA president, I went to several continents and several international radiology conferences. Two common topics of discussion were artificial intelligence and radiologist shortages. I feel advances in AI have the potential to ease the shortage of radiologists, because AI can help with a lot of tasks that are time-consuming.
There’s a story I like to share. I have traveled to a couple of the largest hospitals in the world. In South Africa last year, I saw the third largest hospital in the world, Chris Hani Baragwanath Hospital in Johannesburg. In the radiology department, all the radiologists were busy at workstations reading CTs and MRIs. There were plain films all piled up on the floor. I asked, “What’s this?” And they said, “We don’t have enough radiologists to look at plain x-rays.”
So think about the potential application of AI here. There aren’t enough radiologists in some of the developing countries. AI could be embedded within a plain-film machine with machine learning algorithms built in to automatically scan the radiographs and highlight cases with positive findings on the work list. Radiologists can then at least review those cases.
What has been the most memorable, edifying or surprising thing you’ve observed from your vantage point as RSNA president? And how might you apply this to your work going forward?
As president, I have been fortunate to work alongside a tremendous number of tireless RSNA volunteers who passionately devote their time to advancing research and education initiatives with the goal of transforming patient care. It has been just thrilling to witness the growth of AI applications and to project its potential influence on the future practice of radiology.
I am also very proud of global collaborations that RSNA has fostered. RSNA is a global community, and, through such collaborations, we are able to share ideas with each other so that our specialty can continue to thrive. I feel very fortunate to have had an opportunity to make friendships with worldwide leaders in our field. I look forward to working with them even after my presidency.
What’s on your must-see list for RSNA’s 104th scientific assembly and annual meeting?
Everything! The meeting is gigantic and it’s all very exciting. I know one area that’s going to get a lot of traffic is the Machine Learning Showcase. Last year, there were 49 or so companies in this pavilion. This year we expect twice as many.
Also, the RSNA Deep Learning Classroom presented by NVIDIA will have certified instructors to help attendees improve their AI knowledge, and the National Cancer Institute’s Crowds Cure Cancer exhibit is returning for its second year.
3D printing exhibits and sessions were a big draw last year as well.
Yes, the 3D Printing & Advanced Visualization Showcase will feature research and innovations and continue to have a big draw. Also, the IR zone on the show fl oor where companies will present their latest product advancements in image-guided interventions will be attractive to the attendees. The Machine Learning Theater will feature presentations all through the week adding to the excitement and hype about new AI applications in medical imaging.
The conference will be a good place to get a reality check on the hype.
Correct. And we have a phenomenal cast of plenary speakers lined up. It is an exciting time in radiology. Radiologists have always embraced new technologies, going back to the emergence of CT, MR, PET and so on. And now we have another emerging technology on the horizon, artificial intelligence, which will empower us to further improve patient care.