After the success of our first roundtable discussion, European trends in diagnostic radiology, Sectra has arranged yet another great session with the experts in the field. This time, the focus was on structured reporting.
What are the potential benefits of structured reporting in the radiology field? What are the keys to integration when it comes to crafting and implementing a standardized structured reporting system that can leverage these potential benefits? For the purposes of answering these questions, we brought together a panel of radiology experts and thought leaders from different countries and health systems to discuss their insights and experiences in a roundtable digital discussion. The following is a summary of the key takeaways.
Prof. Charles E. Kahn, Vice Chairman of Radiology at the University of Pennsylvania, United States
Dr. Fredrik Jäderling, Radiology Specialist at Capio S:t Görans Hospital in Stockholm, Sweden
Dr. Rhidian Bramley, Clinical Lead for Diagnostics at Greater Manchester Cancer, Consultant Radiologist, United Kingdom
Dr. Ronald van Rheenen, Nuclear Medicine Physician, Chairman of Decentralized EMR Development, The Netherlands
The roundtable was moderated by Hans Lugnegård, Global Product Manager Radiology Workflow, with co-host Fredrik Lysholm, Head of Special Development Projects
Historically, most radiology reports and health data have been organized as free-text narratives, not far from the dictation methods used by health professionals for decades. While this reporting approach may be efficient for many healthcare professionals, it leads to challenges and communication difficulties deriving from the lack of consistency and the unleashed potential of structured data required by modern-day health systems.
Today, radiology departments and health systems are slowly starting to adopt structured reporting, using templates to organize and store discrete data in a manner that enables reports to easily be communicated to other departments and health professionals. The adoption to structured reporting also includes AI integration, decision support, and quality registries, which can store and evaluate data to improve health outcomes.
Although structured reporting has not yet become mainstream in radiology, mainly because of cultural barriers and the lack of supporting technology, the far-reaching benefits of structured reporting are not difficult to imagine. Below, our four panelists discuss the potential benefits, their professional experiences, and their definitions of structured reporting in radiology.
It started when I was a resident, with many older colleagues who thought they were writing novels with a lot of words and not much information. With structured reporting, the report that you send is always structured in the same way.
To start the discussion, Hans Lugnegård asked the participants, “What is structured reporting to you, and what value provided by structured reporting is the most important?” Dr. Kahn was the first to lend his insights: “There are a couple of aspects to it. One is using reporting templates to help organize and guide radiologists to the information that needs to be included as part of an exam. In the US, it’s often the case that for certain examinations, a specific organ or body part has to be mentioned in a report, and if you fail to do so, it’s a compliance issue. Another value is the ability to provide discrete data from a report that can be aggregated and used for decision support or to train AI models.”
Radiology exists to capture, analyze, and communicate relevant data to be used in the treatment of patients. Structured reports not only help guide the radiologist to input and organize relevant data but also help different departments and health professionals extract that data efficiently.
Dr. Jäderling was next to share his views on the value of structured reporting: “It started when I was a resident, with many older colleagues who thought they were writing novels with a lot of words and not much information. With structured reporting, the report that you send is always structured in the same way. As Professor Kahn said, it’s very important to go through these steps so that you don’t miss out on any information, and you can always gather the important data quickly and easily.”
Dr. Jäderling’s unique perspective comes from years working within Sweden’s national cancer registry for prostate cancer (NPCR), which registers approximately 60,000 cases of malignant cancer per year, along with his work regarding deep neural networks for prostate tumor detection in radiology, an area that is benefiting greatly from the shift towards structured reporting.
Dr. Rhidian Bramley, Clinical Lead for Diagnostics at Greater Manchester Cancer, Consultant Radiologist
Dr. Bradley pointed out the importance of collaboration across different clinical specialties, improving efficiency as well as patient outcomes: “Working in a cancer center where we’ve been practicing standardized reporting for many, many years, not just in radiology, I see the importance of speaking a common language, using the same ontologies, and coding systems to get the right, relevant information in the report. Expressions such as ‘treatment response’ and ‘disease progression’ have specific meanings in oncology. Structured reporting goes that step beyond standardized reporting in terms of using technology to support that.”
There is this ideal situation of taking standardized reports and using computers by providing the structure to enable them to support algorithmic working that can provide real advantages.
There already exist supercomputers and digital banks dedicated to health diagnostics, but without accessible data that is easily interpreted by these systems, it is nearly impossible to
Dr. Ronald van Rheenen, Nuclear Medicine Physician, Chairman of Decentralized EMR Development
provide them with the relevant information to reach their diagnostic potential. By using structured reporting, data becomes easier to organize and analyze outside of a clinical trial paradigm, with a much larger sample size. By using discrete data and specific fields, we provide digital insight into health procedures and outcomes.
When this topic was raised, Dr. Rhidian Bramley shared his professional experience: “There is this ideal situation of taking standardized reports and using computers by providing the structure to enable them to support algorithmic working that can provide real advantages.” These insights come from Dr. Bradley’s years of experience working with Christie electronic patient records, which incorporate structured reporting and big data to improve patient outcomes.
Dr. Rheenen continues by digging deeper into the different layers of structured reporting: “I’ve recently heard about data-enriched reporting, and I really like that phrase, because it says something about the results. You are reporting, but you are also certain that there is data in there which you can use, either at that point or later on.”
Structured reporting is great for research purposes, but also for quality control. And I think it’s very important to educate yourself to learn more from what you’re doing and become better.
As discrete health data from patients across various health systems is collected, compiled, and organized, computer banks and AI have the potential to then analyze this data and provide assistance to healthcare professionals. In the future, it is possible that most of the report content will be prepared by the machine, enabling the radiologist to focus on the diagnostic decision-making and not the tedious task of compiling data. In this scenario, symptoms and treatments could be reviewed and compared nearly instantly to provide the best options for health outcomes.
Meanwhile, in the nearer future, AI and structured reporting tools will be used mainly as aids in the reading process, although humans will need to be kept in the loop as AI can make mistakes that, to us, appear very silly. In other words, we want to include AI tools in the reading process so that we can simply accept or reject suggestions or get help with tedious tasks, such as volume measurements.
Professor Charles Kahn has been working in this field for several years, editing the RSNA journal Radiology: Artificial Intelligence, and was able to share his insights with the roundtable: “Part of my interest is being able to provide discrete data from a report that can be aggregated and used for decision support or, these days, to train AI models.”
Dr. Bramley also added to the conversation, providing clear examples of time-consuming reporting that has the potential to be automated in his field: “In oncology, when you start to look at comorbidities, toxicities, and ICD-10 coding, we really want that to be done in the background automatically through a bit of decision support. The last thing we want to do is make ourselves less efficient, with the huge shortages of radiologists and mismatched capacity and demand at the moment.”
Dr. Fredrik Jäderling, Radiology Specialist at Capio S:t Görans Hospital in Stockholm
Structured data can also be used to provide feedback to the primary reader and enable us to learn more. As a representative of the industry, Sectra can be a great help by presenting data in a certain way and connecting that data throughout the entire diagnostic chain.
Dr. Jäderling saw an opportunity to do exactly that: “I saw that the radiologists that did the reporting never got to follow up on what they were reporting. They wouldn’t know what happened to the patient, what kind of cancer they had, if they had cancer at all. So as soon as Sectra made it possible to paste a report directly into the system some five or six years ago, I incorporated what the urologists needed to know prior to, for example, a diagnostic biopsy, surgery, or radiation therapy to create templates. Now, we have a national database in Sweden, Inca, for all types of cancers, and the system sends a notification if a patient that you reported on two weeks ago now has a pathology report. Then you can go back and look at the images along with that report to see what the biopsy showed. Structured reporting is great for research purposes but also for quality control. And I think it’s very important to educate yourself to learn more from what you’re doing and become better.”
If you’re the oncologist or the surgeon, you want exquisitely detailed information so that you can plan therapy appropriately. If you’re the general practitioner, maybe you just want the high-level overview of what’s going on, but you don’t need to know the surgically relevant anatomy.
Before the wide-scale benefits of structured reporting can be realized, these systems must first be adopted and implemented within health systems. Implementing structured reporting very often also requires health professionals to change and evolve their workflows to adopt these new technologies.
To make this transition, solution engineers must understand the needs of these professionals and the efficiency of their systems and provide both short-term and long-term professional benefits. Our four panelists weighed in on three keys to integrating structured reporting within radiology.
What scanning technology is being utilized? What are you scanning? Why are you scanning it, and what are you looking for? Who will receive the report, and for what purposes? These questions are all important aspects that must be considered and adapted if a structured reporting template is to provide value for the radiologists.
In order to not limit the health professional’s reporting options, data templates must be comprehensive, dynamic, and flexible. “If you’re the oncologist or the surgeon, you want exquisitely detailed information so that you can plan therapy appropriately. If you’re the general practitioner, maybe you just want the high-level overview of what’s going on, but you don’t need to know the surgically relevant anatomy,” explains Dr. Kahn.
While Professor Kahn spoke about template customization for the intended audience, Dr. Rheenen conveyed his frustration at building templates for the myriad of needs and purposes of medical imaging: “I tried to create templates for radiology and I hit a brick wall because a lot of people wanted different things and they couldn’t agree on how things would work, so it was just a medley of people who disagreed. It was chaos.”
While structured reporting makes it possible to collect big data and analyze statistics on a massive scale, the four participants were quick to note the nuances of radiology that cannot be captured through numbers, classifications, or drop-down menus.
Even with Dr. Jäderling’s years of experience with customized templates and structured reporting, he noted the need for free-text, narrative fields: “There are some points where you are uncertain of your findings but are leaning more towards something. In those cases, classifications are not enough. But you can still do the classification and also provide some free wording or free text in order to acknowledge the difficulties or whatever you’re experiencing when assessing the images.”
Patient care comes down to more than just statistics and categorized values, and all four participants agreed that templates will nevertheless require free-text fields within standard reporting practices.
As Dr. Bradley put it, this allows for “human interpretation.” “Obviously, we will be collecting data, but we also need to provide knowledge, insight, and interpretation, and some of that requires free text, but it’s much more of a contextualized summary. You have to answer the clinical questions, but the data to get there can be structured to support machine learning and serial presentation of data while at the same time allowing us to provide a human interpretation,” explained Dr. Bradley.
Back when we first started doing this RSNA initiative, I was giving a talk at the meeting and somebody in the audience stood up and said, ‘How dare you? How dare you tell us how to dictate a head CT right?’ Fast forward six years, I’m giving more or less the same talk at RSNA and somebody in the audience gets up and says, ‘Why do you guys have three different templates for head CT? Why can’t you have just one?
Prof. Charles E. Kahn, Vice Chairman of Radiology at the University of Pennsylvania
For health professionals to modify their workflow and adopt new technologies, those solutions must provide clear and discernible value. The long-term benefits of structured reporting can only be realized if the technology is utilized and data is made available to these wider systems.
Professor Kahn was the first to share his experience of the complications that arise from professional buy-in. “Back when we first started doing this RSNA initiative, I was giving a talk at the meeting, and somebody in the audience stood up and said, ‘How dare you? How dare you tell us how to dictate a head CT right?’ Fast forward six years, I’m giving more or less the same talk at RSNA and somebody in the audience gets up and says, ‘Why do you guys have three different templates for head CT? Why can’t you have just one?’”
Professor Kahn later described the adoption of structured reporting as a “social engineering exercise,” but in Dr. Bramley’s experience, buy-in is all about short-term and long-term value added. “Structured reporting helps enforce quality and ensures that the right methods are used and the right options can be selected. But also, efficiency, because one of the challenges to practicing structured reporting is getting clinical buy-in. But if you can use the tools to really support and provide a win-win situation of improved quality and improved efficiency, then you’re a winner overall,” explains Dr. Bramley.
Before the roundtable came to an end, the participants were asked to predict the impact structured reporting will have over the next five years. While not every benefit of structured reporting has been realized and the challenges are at the forefront of development, Dr. Jäderling sees the advantages close at hand. Dr. Bramley would love to see further adoption of best practices, such as the multiparametric prostate project that Dr. Jäderling was working on in Sweden.
Dr. Bramley is also curious regarding the AI potential in this: “I think the interesting thing to me is where the AI is doing the actual image analysis and reading beyond CAD and actually generating structured reports, and then the radiologists or the second reader do a review to assure some of those.”
Finally, Dr. Jäderling contributed closing remarks, leaning upon his AI project for the development of a neural network for detection and staging of prostate tumors, and predicting this method for use across multiple medical disciplines: “I’m more of an oncology radiologist, but gynecology, breasts, thorax, and rectal cancer are all very suitable for structured reporting progress. Any field with these classifications has the potential to see the benefits of structured reporting.”
Hans Lugnegård continued: “All in all, what we will see in the coming years is that in the near future AI and structured reporting tools will be used mainly as aids in the reading process. Enterprise imaging systems will include AI-enhanced structured reporting tools in the reading process, enabling radiologists to more easily use these templates and accept/reject suggestions or get help with tedious tasks.”
We want to thank our panelists for joining us in this virtual roundtable discussion on October 6, 2021, focusing on the benefits of structured reporting in radiology. Sectra will arrange more of these virtual roundtables in the future, deep diving into different key themes. Stay tuned!
