DI: How will radiology change five years from now?
February 7, 2023
What will radiology look like five years from now? Radiologists from across the United States and Canada weighed in on this topic at the annual RSNA meeting in Chicago last November. While it’s difficult to paint an exact picture of the future, the DI experts seemed to have a good sense of what will happen – in broad strokes.
The speakers had to be clear and concise in this session – they were limited to five minutes each. Here’s how they see the future unfolding:
Chair, Dept. of Radiology and Imaging Institute, Allegheny Health Network Pittsburgh, Penn.
Dr. Casagranda focused on the role of teleradiology. Is teleradiology a friend or foe, she asked.
“Twenty years ago, I would have said foe,” said Dr. Casagranda. “I was raised in academic radiology, and you always stuck it out together. You did the commute together, distributed your conferences together, you sat in the bone pit together. You did everything together. Teleradiology threatened that togetherness – it’s breaking up the family! And you don’t want to break up the family.”
However, today things have changed.
“I’m now the chair [of radiology] in a hospital system that has 10 hospitals and 28 outpatient imaging centres. We are responsible for putting out just over a million RVUs last year,” she said. (In the U.S., an RVU or relative value unit is a standard that defines the volume of work doctors perform.)
Addressing the rising demand for DI exams and the shortage of skilled personnel that plagues healthcare systems across North America, she observed: “In programs this big, the old-fashioned way of doing radiology has become incredibly difficult and some would say, impossible. So, in recent times, ignited by the (COVID-19) pandemic, there’s been a seismic shift in the way we do work.”
In short, work can be done by professionals from anywhere, anytime. The tools are available, and it makes sense to draw on remote human resources. As well, some people don’t want to be at the office all the time – work and life balance are becoming a bigger issue.
“When you take a look at this, in our own workforce, people are no longer interested in doing things the way we’ve always done them,” she asserted.
On a daily basis, she said, she is dealing with:
- Large geographic areas
- After hours – 24/7 after hours coverage.
- Multispecialty interpretation in the setting of increasing case complexity
- Reduction of turnaround times
- Expanding access to the underserved
“When I looked at the list, I realized that this is the backbone of teleradiology,” said Dr. Casagranda. “In my department, I now say that teleradiology is now copasetic.”
So, what is the real foe of academic radiology? “I’m going to be a little controversial and say that it’s been the historic inflexibility of academic radiology,” she commented.
And she had some definite ideas for things that should change.
First, Dr. Casagranda said that management should support work from home days for academic radiologists.
Second, they should engage junior staff in early leadership roles and passion projects. “It used to be that all the leadership decisions were made at the top – I recommend that you start levelling that out. You want to make it so that they never want to leave their jobs.”
Finally, she addressed what could be called ‘the old boys club’ of medicine.
Dr. Casagranda showed a slide with an historic photo of her hospital’s department chairpersons. The 35 images were all of white men!
“I’m not here to white-man bash. But what if things looked like this,” she said, bringing up a slide, with an ethnically diverse group of 35 faces. “This is 25 percent of our department. What if we were welcoming and inclusive, so that all the brightest minds went into radiology?”
In the future, she said, the secret sauce for success will be inclusivity. “I think that as we go along, we will be representing the community that we serve.”
Chair, Radiology, Vanderbilt University Medical Centre, interventional radiologist
Dr. Omary focused on the threat to human health from climate change, saying it should become a major issue for radiologists, as well as all physicians. He pointed out that in March 2020, the COVID-19 crisis struck and morbidity and death in its wake. It also changed the way medicine is practiced.
“Today another crisis is looming even larger,” said Dr. Omary. “We know that flooding has occurred, not just Harvey but Sandy in New York. California, I’ve lost track of how many wildfires there are. Pakistan had a third of its nation under water, and France has suffered from drought.
He noted that the same individuals who are likely to be affected by COVID are likely to be affected by climate change – the elderly and people with chronic conditions.
Dr. Omary asserted that healthcare is contributing to the problem. “We in healthcare are a major contributor to greenhouse gas emissions. If healthcare were a country, we’d be the 5th largest emitter in the world. The direct costs of climate change to health, according to the WHO, are over $2 billion.”
Interestingly, he observed that radiological publications on environmental sustainability and climate change are increasing – a sign of the growing interest in this issue. “There has been a sea change in radiology publications on environmental sustainability and climate change – 10 in 2020, 14 in 2022. I predict in 2023, we will break 30 and in 2024 we’ll break 100.”
Dr. Omary concluded by looking into the future. “How do we define if we’ve been successful five years from now? It’ll be if we see an EcoRadiology [stream] coming to RSNA 2027. We’d learn about green resources, we’d even be able to share our favourite plant-based restaurants. This can happen, I’m confident that it will happen.”
Professor, UBC Dept of Radiology, Vancouver
Dr. Forster aimed his remarks on sports medicine. In particular, he looked at the radiologist’s role in determining when an injured athlete should go back to the playing field. “In radiology we get asked a lot about injured athletes going back to their sport. The trick is to do it without causing re-injury – either in the acute sense or in the chronic sense, which could put athletes out for months or even years from their sport,” said Dr. Forster.
He observed that most often, the patients are amateurs – for them, recovering at home isn’t a major issue. However, when you get up to elite amateurs it’s more difficult – they’ve been training their whole lives. “It gets even more difficult with professional athletes. In major league baseball, injuries cost them $700 million each year.”
Dr. Forster asked, “How are we doing on this [as radiologists]? The media thinks we’re doing awesome.”
He cited a few examples, including that of Alphonso Davies, the Canadian soccer star who injured his hamstring just before the World Cup; the media reported that his upcoming MRI scans would reveal the exact extent of his injury.
“So, no pressure, no pressure there,” quipped Dr. Forster.
He then asked what the science says in terms of return to play? “It’s all over the place,” he answered. “There are some really good things we’ve learned, like using the Frederickson score system for tibia stress fractures in MRIs.” (Frederickson scores and images are a good predictor of how soon the athlete can go back to his or her sport.)
“But it’s with hamstring tears that the wheels fall off. This is a very controversial area. Some investigators have shown that when you tear the central tendon, patients take longer to return to play. High ankle sprains? We can make the diagnosis really well, but the things we see can’t seem to correlate to return to play.”
Apparently, there are stumbling blocks in diagnosing sports injuries and determining when athletes should return to play.
Dr. Forster said that improvements will occur, but they will probably require 10 years rather than just five. That’s because he’s factoring in the use of AI and machine learning in diagnosis. But collecting and analyzing the data, and building it into workflows, will take about 10 years, he predicts.
Moreover, there will be a lot of data.
“There will be data that we input that AI and ML will help us solve. But it’s not just regular images but quantitative images, T2 mapping, temperature imaging to look at muscle fibre orientation. Also, things like elastography, to look at what role fibrosis plays in return to play decisions.
“Biomechanics is super important,” he added, “and that’s got to be part of the data set. Maybe there’s also data from physiotherapy. Bone, muscle and fat mass is another important thing to include. And certainly, the ultimate to include is genomics.”
He suggested that radiologists are in an excellent position to “quarterback” the decision-making, leading the medical team. “I predict that radiologists have an incredible opportunity here,” said Dr. Forster. “We know the imaging, which is an important but not exclusive part of the decision-making. But we also have an advantage because we’re a little ahead of the game in AI and ML. As long as we do the research, we’re in a good position to become the team leader in making return to play decisions.”
Professor and Chair, NYU Langone Health.
Dr. Recht spoke about AI-enabled training as the future of radiology education. He noted that his daughter finished her radiology residency about two years ago while he finished his own approximately 30 years ago. Despite the time difference, “the structure of the residency program was essentially the same,” said Dr. Recht. “In the first three years of residency, all residents rotate through the same subspecialties, for the same length of time, regardless of their individual backgrounds and their performance on the radiology rotation.”
One of the most important facets of an educational residency program, he said, is the daily readout at the workstation. “And yet we all rely on random cases to populate the resident’s workload every day. What that means is that there is no guarantee that every resident will see the full spectrum of cases and pathology that they should see on each rotation. And, in fact, it is likely that they don’t.”
Dr. Recht observed that we don’t have a very good way of objectively evaluating a resident’s performance on the rotation. “Typically, we rely on the subjective evaluation of our attending radiologists. But we all know that our attendings only spend a few days a month with the resident, limiting their ability to truly evaluate each resident’s strengths and weaknesses.”
But thanks to AI, we can change all that. “For the first time, we can tailor each resident’s education to their own strengths and weaknesses. Let me explain how we’ve started to develop such a system at NYU Langone.”
He described how the first step was to define the types of cases and pathology that each resident should see in each year of their residency.
Second, Langone has started to build a curated set of cases that includes multiple types of pathology and types of cases for each rotation.
They’re also tracking, accurately, the types of pathology and cases that each resident sees on each rotation. “Using that information, we can supplement the cases the resident sees on their rotation with our curated cases, to make sure they see the full set of pathology,” he said.
The department has also devised an NLP algorithm that compares a resident’s dictations with that of the attending radiologist. “We can use that to identify gaps in each resident’s learning,” said Dr. Recht. “We can further supplement the resident’s worklist with cases that address their weaknesses.”
For example, if a resident is good at diagnosing musculoskeletal tumours, but not so good at differentiating the types of inflammatory arthritities, more arthritis cases can be added to their daily worklist.
“What we hope to use this information for, in the future, is to vary the length of each individual’s rotations,” said Dr. Recht. “So, if someone is really good at MSK radiology in their third year, maybe they only do a rotation of two weeks instead of four weeks. And they use those extra weeks for a rotation in something they’re not so strong in. Maybe neuroradiology.”
In the future, Langone hopes to be able to use this technology not only to vary the length of the rotation, but perhaps for the entire residency – allowing professors to graduate a resident based on their individual competencies, and by hitting milestones, rather than on the current, one-size fits all approach.
MD, Chair, Radiology, Penn State University
Dr. Mosher concentrated on the HR crisis that’s impacting the healthcare sector, and industry, in general. “I predict that the 32-hour work week will become the new standard for full-time employment within the next five years,” he averred. “Those of you who might be responsible for staffing, or if you’re an administrator, you might greet this with a little bit of skepticism.”
After all, many radiologists have been working 60-hour weeks. How will management staff their departments if a 32-hour week becomes the standard?
He answered, “I would argue, you’re not going to have a choice. Because this is not an internal disruptor, this will be an external disruptor. If you look through history, the changes in work hours never came from the top down, they came from the bottom up.”
Dr. Mosher suggested that we’re on the cusp of sea-change in attitudes, where the public doesn’t want to work long hours anymore.
He mentioned “The Great Resignation”, in which workers have been retiring well before they need to. “A lot has been said about the Great Resignation, but the trend started well before the pandemic, largely due to the aging Baby Boomers who were retiring from their work,” he commented. “Their retirement funds have allowed them to retire. COVID accelerated that. It reached a peak in November 2021.”
Dr. Mosher added, “But we’re going to continue to see this attrition in the workforce.”
Those folks, of course, are older people taking earlier retirement. Pressure is coming from the other end of the age spectrum, as well.
The attitude of younger people who are just entering the workforce is different. They don’t want to be tied to a job 40 to 60 hours a week. They value their time as well as their compensation.
Already, in the US, there are a lot of companies that have switched to a 32-hour work week.
“These are the industries – often high tech – that we’ll have to compete with in the future when we want people to go into medicine,” said Dr. Mosher. “We will have to be able to compete with the flexibility being offered to people by these other industries.”
And it’s not only companies – there are many countries that are piloting a four-day work week. They include Japan, the UK, Belgium, Scotland, Iceland, Spain and Ireland.
The US House and the California state government have already introduced bills to shorten the work week. They died, but the effort has already started.
In Iceland, 86 percent of the working population is already on a shorter work week or has the right to do so in the future.
“So, on the innovation curve, we’re already past the point of early adopters,” said Dr. Mosher. “We’re close to the tipping point, as the idea is achieving acceptance by the ‘early majority’.”
He concluded by saying, “As we see further attrition of the workforce, it won’t be a matter of ‘will’ we adapt but ‘how’ will we adapt.”
MD, Chief of Muskuloskeletal Imaging, Wake Forest University School of Medicine, Winston-Salem, North Carolina
Dr. Lenchik dedicated his five minutes to radiomics, which he believes will become routine in diagnostic imaging. “I don’t know when that will happen – possibly in five years. What is routine now? Orthopedic surgeons love measuring things and radiologists hate measuring things – because (measuring is) manual, we make mistakes, we face the wrong way, and something we hate even more is segmentation. Fortunately, aside from research where we get fellows to do it for us, we don’t have to do it very often.”
However, he asserted that in the next few years, radiologists won’t have to bother with manual procedures for segmentations and measurements. Instead, it will become automated, and radiomics will be automatically applied on top of these metrics.
But in the next few years, this will change, so we will have automated segmentations and measurements. And in the next five years, radiomics will be added to the process.
“Essentially, the first step in radiomics is to do some sort of image segmentation and feature extraction,” said Dr. Lenchik. “And you get a bunch of variables, so in addition to muscle size and density, you get about 80 different variables or phenotypes, that represent texture. And it turns out that these texture features are invisible to you with the naked eye, but they’re visible to the machine.”
He noted that there is currently a lot of discussion about the use of radiomics in cancer care. “But I’m going to make an argument that it goes beyond cancer. Right now, there are people writing papers about radiomics in Alzheimer’s Disease and other non-cancerous conditions. For example, in cardiac CT and MRI, pancreatitis, renal stones, and of course, musculoskeletal imaging.”
He added, “In my view, muscle, bone and fat radiomics will improve the care of older adults. In sarcopenia (the decline of skeletal muscle tissue with age), with which I’ve been involved, there are a lot of risk factors for getting sarcopenia and a lot of things that can go wrong once you have it.
“You can die faster, your brain can decline faster, you get re-admitted to hospital, you’re miserable. So, how do we change that?”
He suggested that analyzing sarcopenia on CT, measuring muscle size and density is one way. From there, radiologists can apply radiomic analysis. “There’s freely available software – and it will be embedded in the manufacturers’ toolbox soon, as well,” he said.
He said that radiomics can be combined with other ‘omics’, like proteomics, genomics, metabolomics, to create models of health and disease that are more accurate than other models.
“This has been done by many groups and will continue to be done in the next five years,” said Dr. Lenchik. “But in my view, radiomics will become routine, by improving image-derived analytics, hopefully by improving prognostic tools and best of all, by improving patient care.