5 Things to Watch in 2021
By Dave Yeager
Vol. 22 No. 1 P. 22
As with every other medical conference since March, the COVID-19 pandemic necessitated switching RSNA 2020 to a virtual format. That didn’t stop the organizers from offering a wide range of educational sessions, academic presentations, and business news, however. The hustle and bustle may have been replaced by the LED glow of computer screens, but there was still plenty to see.
Whether it was in response to the pandemic or simply a coincidence, this year’s show had more of an introspective feel than in recent years. The conference’s theme was “Human Insights/Visionary Medicine,” and many of the plenary sessions examined the need for increased medical imaging integration throughout health care. In his president’s address, “One World, One Radiology Community—A Vision for Tomorrow,” James Borgstede, MD, the 2020 RSNA president, stated that there is an opportunity in today’s health care world for radiologists to facilitate integrative medicine, and he urged radiologists to share their knowledge as well as learn from colleagues in other fields of expertise. He also emphasized that radiology must embrace collaboration if it wants to maintain influence in the years ahead.
In her opening session lecture, “The Power of Radiology to Drive Collective Action and Transform Global Health,” Kristen K. DeStigter, MD, highlighted both the urgent need for medical imaging and the challenges that radiology faces. DeStigter believes it is imperative to raise radiology’s profile so it becomes part of the nation’s health care planning. She also noted that there is a shortage of radiologists, particularly in developing nations, and it will be essential in coming years to make medical imaging more accessible to more people.
Of course, everyone talked about COVID-19, and some of the changes it has brought may last long after the virus’ threat has ebbed. AI was also a hot topic, as it has been for many years, but it is being deployed in a broader variety of ways. Part of this is due to AI’s hype having dissipated just a little, and some of it is due to necessity. As Paul Chang, MD, put it in a session about practical considerations for implementing real-world AI and machine learning, “Whether or not your perspective towards AI is one of skepticism or [anticipation], from a perspective of a practicing radiologist, I can tell you that, whatever we’re doing, we need some help because what we’re doing now is barely sustainable.”
With those thoughts in mind, the following are five things that caught my attention at RSNA 2020.
1. What Hath COVID-19 Wrought?
Might as well get this one out of the way first. As with the rest of the world, the pandemic was never far from anyone’s mind. After shutdown orders were implemented in the spring, screenings of all kinds plummeted. The numbers bounced back later in the year, but there is still a backlog of patients who would benefit from screening. This is most pronounced in breast screening, where imaging facilities were forced to overhaul their scheduling procedures and workflow processes. There was also a significant uptick in remote reading of images, to limit radiologists’ potential exposure to the virus and assist facilities that were shorthanded.
Even though the hope is that 2021 will be kinder than 2020, many people who I spoke with believe that some of these structural changes are here to stay. For example, more efficient patient throughput benefits facilities and patients alike, pandemic or no. Also, the ability to read from anywhere offers a way to utilize radiology resources more efficiently while allowing radiologists more flexibility to set their work schedules, potentially alleviating some of the burnout factors that currently exist. The movement toward remote work that COVID-19 has accelerated leads us to the next trend.
2. Off-Site Insight
Several vendors highlighted remote service options that can find, and in some cases predict, operating system problems. Some also offer virtual training programs for users. Perhaps most impressively, some offer the ability to coordinate workflow throughout an imaging center, regardless of which vendor’s scanners are used, or provide feedback to technologists to ensure that an imaging exam obtains the necessary information. This type of model promises the possibility of reducing workforce requirements and standardizing processes.
Initiatives such as this will require a more significant shift to cloud services. Several vendors noted an uptick in cloud use this past year, and it appears that the pandemic has accelerated what was already a slow but steady movement toward the cloud. Although many organizations still prefer to manage their own data, many others are finding that cloud services free up resources that can be deployed elsewhere; particularly among smaller to midsized organizations, efficient resource use is a matter of survival.
Mergers among radiology providers were already driving some of the shift, and many in the industry believe mergers and consolidation will continue to restructure the radiology landscape. Along with keeping organizations connected, remote workflows can facilitate access to subspecialty reads, which is becoming more of an issue.
3. Maximizing the Radiology Workforce
Efficient use of radiologists’ time is more important than ever. One reason that remote reading is growing is that there aren’t enough radiologists. It’s also part of the reason for the more intense focus on AI. Whether enough radiologists can be recruited to meet the expected need in the coming years is an open question, but there are already critical shortages in some parts of the world. DeStigter, in her lecture, and Bhavya Rehani, MD, in the Annual Oration on Diagnostic Radiology, highlighted this situation in detail. Deploying radiology more efficiently will require new thinking.
One strategy that may prove helpful is recruiting more women and people of color. Although there are current initiatives that focus on diversity, more can be done. In the New Horizons lecture, Shankar Vedantam discussed how unconscious bias may be holding back progress. Among many interesting points that he made, Vedantam noted that women and people of color may be reluctant to choose radiology because they don’t see many people who look like them. Also, because bias can arise from unconscious processes, he suggested that organizations put “guardrails” in place—rules that ensure diverse representation—to overcome unconscious bias. There were more than a couple of sessions dedicated to improving radiology’s diversity.
Another source of help may come from AI. AI is being used to improve workflow by distributing caseloads more fairly and efficiently, which can improve harmony within a practice and help to reduce burnout, while routing cases to the appropriate specialists. This use of AI can also ensure that urgent cases are moved to the top of the reading list.
Borgstede, DeStigter, and Rehani also advocate for more collaboration with other specialties. Maximizing radiology’s value depends on demonstrating how radiologists’ expertise improves outcomes. AI may continue identifying medical conditions with greater precision, but radiologists need to put those findings in context for other physicians and patients. Collaboration can help to identify the areas where radiologists are most helpful while incorporating knowledge from other specialties.
4. Leveraging Data
A factor that may help radiology to become more collaborative and demonstrate value is the massive trove of data it produces. As Chang put it in his lecture, radiology needs to “transition from merely storing data and information to optimally leveraging knowledge,” which is an area where AI can be useful. He notes that doing so will require larger IT investments, but he believes radiology can learn from mistakes that other industries have already made.
I have often heard it said that AI runs on data, and those data come from medical images. There was already a growing number of AI-based machine learning algorithms that can identify specific medical conditions, and COVID-19 has resulted in a surge of algorithms that look for signs of the virus. AI is also being employed with educational tools that help radiologists recognize different pathologies. Expect this trend to pick up speed.
Some in the industry are also using AI to advance radiomics—data not visible to human eyes that computers can use to identify medical conditions or even predict which patients may develop medical conditions. The field of predictive analytics is most advanced in breast imaging; some vendors highlighted predictive risk models for breast cancer among their offerings. Several sessions were devoted to radiomics and predictive analytics. Other tasks that can benefit from data analytics include decision support for exam appropriateness and 3D-assisted image annotation.
Optimizing data will require increased interoperability between radiology systems and other systems, however. The industry has been talking about eliminating “data silos” for some time, but doing so will require more of an enterprise-level view of medical imaging. This will continue to be an area of emphasis going forward, and it may be aided by more organizations opting for cloud services.
5. Patient Engagement
Part of the emphasis on interoperability is to make data in imaging reports more useful for a variety of users. One example is follow-up recommendations. Several vendors noted efforts to improve the communication and tracking of follow-up recommendations. These types of initiatives align with a broader goal of improving population health.
To truly improve population health, however, health care organizations will need to work more collaboratively with patients. There has already been some movement in this direction with patient portals, and patients are more interested than ever in their health care data, especially their medical images. Some providers are already engaging patients with text messages to schedule appointments, such as screenings, and obtain previsit medical information more efficiently. Many people I spoke with believe this trend toward consumerism is in its early stages. In the future, providers will need to include patients as partners in their health care decision making.
— Dave Yeager is the editor of Radiology Today.