Since the introduction of the first SIGNA™ 1.5T in 1983, our design and engineering teams have examined MR workflow and how its use changes and adapts to clinical and operational needs. The significant challenges facing healthcare today require a fresh look at MR imaging workflow. We knew we had to challenge our prior assumptions and relearn how MR imaging is performed all over the world.
MR departments are facing a perfect storm. Rising clinical demand and technologist shortages are exacerbating patient backlogs. New software and technologies are dramatically reducing scan times, leading to shorter scheduling slots and higher patient throughput. While shorter scan times are beneficial for operations and patients waiting for an appointment, it also places more pressure on technologists to work faster and move patients through their examination quicker. Some aspects of the technologist’s workflow, such as patient preparation and positioning, are manual tasks that require care and attention for each individual patient. And MR is a complex, user-dependent modality requiring knowledge of the imaging process and protocols.
We had to understand how MR is performed today in this challenging environment and relearn the technologist’s experience and workflow. What are their pain points and where do they think technology can help them improve—even in manual workflows? Our teams from the U.S. and China visited sites all over the world to capture as much information about today’s MR experience from both the technologist’s and the patient’s point of view.
Figure 1.
GE HealthCare’s Next Generation MR Platform.‡
One thing became clear: technologists are constantly moving. Whether it is physical movement during patient preparation and positioning, or the visual and mental acuity required to operate the MR system, technologists are continually performing a task to achieve a high-quality exam.
The challenges technologists face in a fast-paced environment are more acute with an increasingly older, sicker patient population. Many patients have co-morbidities, MR Conditional implants and other conditions the technologist must manage. Just as MR imaging has become more powerful and provides deeper insights into the human body, it has also become more complex. Examinations such as cardiac MR and whole-body imaging, once considered the domain of specialty groups or academic hospitals, are now becoming more routine in outpatient centers and community hospitals. These examinations require more knowledge, skill and time for the technologist to complete, further burdening an already stressed staff. While many of GE HealthCare’s artificial intelligence (AI) solutions offer the means to mitigate some of these challenges, we knew there was much more to be done to address the entire technologist workflow.
As we visited customers across the globe, our teams quickly realized the large number of small technologist movements and saw first-hand the often stressful and fast-paced environment in which they work. At the same time, we understood that people have come to expect a more interactive interface with smart devices and laptops. This expectation for an efficient and adaptive experience that began with consumer electronics has transferred to the work environment. Medical imaging is no different.
As with other design initiatives of the Next Generation MR Platform, our team visited customers’ sites to understand our users across different demographics and geographies—U.S., Asia, Europe and Latin America. We pulled opinions and observed workflows to ensure the design was appropriate for everyone and not just one specific type of user.
AI-enabled camera workflow
One of the most common complaints from MR technologists is that in an increasingly fast-paced MR imaging environment, they have less time to focus on the patient and the intricacies of the individual exam. Even the smallest errors in patient positioning and set-up may impact image quality or require additional time to re-enter the room and adjust the patient’s position, and/or repeat the imaging series.
Figure 2.
The new AI-powered camera was developed to assist technologists in landmarking anatomy and streamlining the localizer workflow. It precisely detects the patient’s body position and orientation, including 27 key anatomic markers, using an AI algorithm.
GE HealthCare has made significant progress providing AI- and deep-learning-based technologies to shorten scan times. Yet, the patient preparation workflow has remained relatively constant for many years, for example, prescribing an offset. In a shoulder exam, the offset can vary from 80 mm to 220 mm; many extremity MR exams have a similar gap. Technologists with extensive experience can likely make intuitive adjustments to protocols and strategies; but increasingly, technologists have less experience and may struggle significantly to make the appropriate adjustments to accommodate patients that fall outside the expected protocol parameters.
Another common pain point for many technologists is they often repeat the localizer to find the key anatomy before scanning. We’ve seen through our experiences, customer feedback and observational research that for many MSK exams, technologists often acquire three—and sometimes more—localizers to find the anatomy to create the main prescription. With each localizer taking approximately 30 seconds and prescription changes consuming another 15 to 20 seconds, these additional minutes add up quickly.
Recognizing the need to address this issue, our team set out to investigate if AI can help technologists detect the key anatomy faster and more accurately. We employed the same critical design thinking approach as our other engineering teams working on the Next Generation MR Platform to look at available and emerging technologies. Through this process, it became clear that an AI-powered camera‡ could assist the technologists in landmarking anatomy and streamlining the localizer workflow. In fact, our colleagues in CT developed a similar technology for their Revolution™ Apex product, where they focused on making imaging effortless by supporting technologists who must position patients within a fast-paced workflow. We made a call and tapped into their knowledge and experience for our journey.
Driving to work each day, we use automobiles equipped with large touchscreens that allow us to control nearly every feature of the vehicle. We wanted to bring these large easy-to-use touchscreen controls to the MR environment. We made the existing display monitors on the MR magnet larger and simplified the design and quantity of the physical touch buttons on previous designs. This required us to completely reimagine the workflow to leverage AI and allow safe and effective patient setup to be done faster with fewer controls, buttons and clicks.
To begin, we studied the depth-sensing camera used by our CT colleagues, but quickly determined we must exchange the camera housing for one that is MR compatible. We then developed a “smart” system that tracks the patient and 27 key anatomic markers using an AI algorithm trained on live video feeds of patients on MR tables. This gives the technologist the ability to landmark the exam with a single touch on the screen where the live camera feed shows the anatomy of interest. By giving the MR system the ability to detect the patient’s body position and orientation, it can provide the precise location of the anatomy along the length of the table, but also the patient-specific left-right offset to make shoulder, knee, elbow and even ankle exams a breeze. Providing orientation detection also helps reduce a common mistake of scanning a patient in the wrong orientation, which could happen with the option to scan head first or feet first. It also allows us to simplify the way we manage pre-set protocols. Instead of needing two localizers within one protocol, such as a Left Shoulder and Right Shoulder protocol, we can use a single localizer that uses the camera-based offset. This simplifies protocol management and ensures consistent imaging parameters on both sides of the body.
Anyone who has been scanned or entered an exam room knows that MR environments are cold. Often, patients need blankets or coverings to be comfortable. But blankets—or even surface coils—can obscure a ceiling-mounted camera’s view of the patient’s body and make the landmarking process impossible. We intentionally trained our AI algorithms to retain the ability to identify the 27 key anatomic landmarks even when obscured by sheets, blankets or bulky clothing. We also appreciated that children may present particular challenges for an AI algorithm trained only on adults. For this we called our friends and family and invited them to a GE HealthCare Experience day, where they were able to visit our development facility, but also provide some much-needed AI training data to ensure our product is able to accurately function no matter the patient’s size. An advantage of working at a global organization such as GE HealthCare afforded us the opportunity to ethically and compliantly acquire training datasets from a very diverse set of volunteers. Tens of thousands of images and datasets were collected from our offices around the world to reduce bias toward any patient demographic.
We were mindful of the different workflows and orientations often preferred by technologists in regions around the world. For example, in China, technologists work in an extremely fast-paced environment. We thoughtfully developed our workflow with the support of dozens of internal and external technologists to guide us in designing an optimal workflow that does not interrupt or impede normal day-to-day workflows for technologists. Whether they use the AI features or not, it was important to help make everyone’s workflow more efficient. These features are in addition to existing capabilities that our users appreciate, such as IntelliTouch™ or manual laser landmarking.
Figure 3.
SIGNA One‡ Interface is a complete architectural change to the entire GE HealthCare software platform. Built with technologist feedback top of mind, it is designed to be intuitive and seamlessly guide the technologist through the entire MR examination.
(A) Intuitive, inline menus for common exam setup steps without pop-up windows make the interface flow. (B) Access to exam parameters is thoughtfully organized into groups and tabs to ensure the information is available but not overcrowding the screen. (C) A new way to set up voice commands makes the process easier to understand. (D) Familiar GE HealthCare features, like the coil selection screen, retain useful capabilities in familiar locations and layouts.
SIGNA™ One Interface
MR is a complex imaging modality, but the user interface does not need to be difficult to operate. A technologist must focus on many different things throughout their workflow—the easier the user interface is, the more they can focus on the patient.
Simplicity is at the core of the new SIGNA One‡ Interface. The interface interacts with the user similar to current smart devices and laptops. It is a “directed interface” that is intuitive and
designed to seamlessly flow and logically guide the technologist through each step. We’ve removed pop-up windows and provided drag-and-drop functionality in each step—protocol selection, settings, options and task list. Once the exam is saved and the prescription is confirmed, the technologist simply presses “scan” and the exam starts and works seamlessly with the new control hardware with controls like emergency stop (Figure 4).
Figure 4.
The Scan Control Interface Module (SCIM)‡ has been redesigned with simplicity, flexibility and visability in mind to offer useful functions to staff with a logical workflow.
SIGNA One Interface is a complete architectural platform change to GE HealthCare’s user interface. It is based on a domain model where each domain is characterized on the back end based on the specific workflow, and the software is aligned with a similar data model on the front end for the technologist. There is a patient domain, a protocol domain, an exam domain and a task domain, all with different levels within each.
After developing the initial design, we began the process of lean testing to further refine it and perform formative testing using prototypes. We brought together clinical, engineering, systems and product teams to ensure the design comprehensively encompassed all aspects of the workflow and addressed what we learned in our observational research. This core team had cross-functional expertise to help identify potential issues and evaluate usability.
Next, these prototypes were tested with MR technologists, both internal application specialists and external customers. Participants were asked to provide feedback on their overall impressions, including what they liked about the system and areas for improvement. Additionally, they were encouraged to push the system to its limits to help identify potential weaknesses. The SIGNA One Interface was built based on technologists’ feedback to ensure it fulfills their daily workflow needs.
Summary
Throughout the design process, we listened to the opinions of technologists—both those who work for GE HealthCare and our customers—to address daily clinical workflow pain points. The goal was clear: streamline the technologist’s workflow so they can fulfill the demands of today’s MR department while also keeping the patient at the center of everything they do. With staffing shortages, shorter examination time slots and more patient volume, technologists are being asked to do more with less. That takes a toll on the person. Anything we can do to make their job easier is a win for them and for us. We believe SIGNA One Interface does just that.
By automating the workflow—from the simplified patient positioning/setup to the streamlined localizer to the AI-based prescription—the technologist can complete each step quicker. Yet the solution must also be designed to adapt to different scenarios including outliers—the things that happen maybe 5% or less of the time—and different staffing levels in different facilities. It was a challenge to design the workflow and components to adapt to different scenarios and technologists with varying levels of expertise and knowledge of technology—yet we persevered, determined to bring forth a common vision across different customers for a simpler, more streamlined workflow that enables, not hinders, the technologist’s workflow.
The future will be driven by automation and AI, and the sooner we embed these into the MR workflow, the more providers can focus on patients and not the technology. We are proud to embrace a path that utilizes new technologies to optimize both the user experience and patient care, and free up the technologist so they can focus on the patient. Because ultimately, patient care is what’s most important.
‡ Technology in development that represents ongoing research and development efforts. Not for sale. Not CE marked. Not cleared or approved by the U.S. FDA or any other global regulator for commercial availability.
