Self-driving cars and cancer radiation therapy have more in common than you might think. Both are based on mechanical technologies that have been widely used for decades. Additionally, both leverage the latest advances in software, image capture capabilities, and computing power to turn these old technologies into new and wholly transformative devices. In the radiation therapy arena, these advances enhance safety and efficacy and improve disease management in a broader population of patients than is possible with today’s technologies.
New uses for today’s technologies
In the automobile setting, a driver is really nothing more than a central processing unit (CPU) that analyzes environmental data – such as the location, direction and speed of cars and other objects. Self-driving cars replace the driver with a mechanical CPU that gathers and processes those data more quickly and accurately, and uses powerful algorithms to adjust speed, direction, and location. Features such as rearview cameras, crash alert systems, and all-wheel drive enhance the driver’s sensing and adaptation capabilities, and it’s not difficult to envision moving from these advances to fully automated cars.
Similarly, humans have been the CPUs for radiation therapy devices since they were first developed at the start of the 20th century. Today, the use of 3D grids to localize tumors, mechanical advances that enable more precise shaping of radiation beams to match the contours of a tumor, and improvements in imaging technology are increasing the sensing and adaptation power of the radiation therapist. The improved ability to differentiate tumor from normal tissue and to more precisely target radiation dosing improves efficacy while reducing damage to adjacent organs. Additionally, combining radiation therapy with other treatment methods opens the door to further innovation of radiation-based cancer therapies.
The availability and prospect of new treatment paradigms is helping to drive further technical innovation. As radiation delivery becomes more precise, dosing schedules can be adjusted to deliver higher doses over fewer sessions, an approach known as hypofractionation. While hypofractionation can improve efficacy and reduce treatment times, its safety requires more stringent control of tumor and patient motion to reduce risk of delivering radiation outside the target area. This has driven innovation of on-table immobilization devices that restrict patient movement and onboard imaging that can assess movement of the tumor and nearby tissue. Motion control systems that can sense millimeter changes in a patient’s position on the treatment table and cut off the beam when motion exceeds a pre-specified amount are also improving the precision of radiation therapy.
From algorithm to automation
The development of advanced sensing and imaging technology and computer algorithms that can respond in real time to changes in patient or tumor location or help to devise treatment plans have been critical innovations that are helping to improve safety and efficacy of radiation therapy. However, full automation of existing technologies will be an important step toward the “self-driving” radiation therapy of the future. Higher precision in radiation dosing has resulted in greater complexity and more steps in the process of delivering radiation therapy. In turn, higher complexity typically requires longer treatment time and may lead to more errors that have the potential to harm patients. Automation, fail-safes, and integrated safety checks have helped to hold down treatment time and human error even as the complexity and precision of radiation therapy has increased. These innovations have provided some degree of “auto-pilot” but current radiation therapy technology still requires a significant degree of human control and input.
Device innovation in the era of rich data
The capture, analysis and utilization of rich data have been essential to developing the self-driving car. Image-capture technology, sensors, and algorithms gather detailed data about the car and its environment and analyze it based on specific rules (traffic regulations, physics of speed and momentum, prioritization to avoid collisions, etc.). Algorithms process the outputs of these analyses and automatically adjust the mechanics of the car, resulting in changes to speed and direction. Similar advances are taking place in MR/RT.
Similar efforts to enable “self-driving” radiation therapy require significantly greater and richer real-time data than is feasible with existing technologies. Magnetic resonance radiation therapy (MR/RT), which integrates high-field MR imaging with advanced linear accelerator technology (MR-linac), allows physicians to see what they treat in real-time and adapt the treatment plan in response to changes in tumor shape, size, or location or the patient’s position. Leveraging rich data was essential to the development of MR/RT systems that use high-field MR imaging and motion detection systems to gather information about the tumor and apply sophisticated algorithms to compare planned versus actual doses. Analyses based on specific rules that increase target dosing while avoiding healthy tissue are used to calculate the treatment dose and the shape of delivery area. Just as the self-driving car adjusts the car’s mechanics in response to changes occurring around the car, MR/RT systems enable automated adjustment of radiation therapy device mechanics –changing the shape of the beam to adjust for patient or tumor motion. Similar to a guidance system “recalculating” when the car deviates from the planned route, the MR/RT system automatically updates the treatment plan to adapt the next treatment session to reflect any deviation from original plan.
The increased safety and efficacy that may be achieved with MR/RT will radically improve the precision of radiation therapy, enabling its use in additional types of cancer and broader patient populations. Imaging and treating on the same device, which is not feasible with other current radiation therapy systems, should improve clinical workflow efficiencies within radiation oncology departments. Additionally, automated planning can do in minutes what humans need weeks to do, improving staff time utilization and reducing treatment delays. Most importantly, MR/RT is expected to enhance the patient experience through more efficient treatment regimens that reduce treatment time and minimize side effects.
Rich data are also driving changes in the medical device innovation process. The development of interactive devices that can easily connect to data storage and analysis tools and the growth of the “internet of things” allow the capture and use of tremendous amounts of information. This creates multiple opportunities to move beyond developing a device and instead create tools (hardware and associated software) that help customers use technology as safely and as effectively as possible. Device manufacturers must find the balance between harnessing increasingly complex technology and decreasing the complexity of the customer experience.
Finding that balance requires that medical device manufacturers embrace cultural changes that foster innovation. This includes ensuring that mechanical and product engineers understand what machine learning algorithms are, how they are used, and what they might enable in the future. This understanding is essential for all companies – even if they don’t use this technology directly – in order for their new product offerings to be integrated into the evolving ecosystem of patient care. The reality is that rich data and automation are here and they will impact how business is conducted – regardless of a company’s industry, technology platform or product focus. Historically many companies have been reluctant to share new concepts and ideas for fear of losing a competitive edge. Today, however, it is essential to nurture international collaborations to support new product development and to employ product-centered interest groups to advance the innovation of existing products. Such collaboration is critical to accessing the best ideas and talents regardless of geography.
Caution on the road ahead
Artificial intelligence, deep learning and robotics are reducing the need for direct human involvement in a growing number of tasks and can handle vastly larger amounts of data more quickly. Any industry in which humans perform repetitive tasks has the potential to follow the self-driving car model. While the human touch will always be a critical aspect of healthcare, consider the many aspects of care delivery that could be replaced by devices that capture and process information using computer algorithms. Now imagine the exponential changes that could be enabled by having these devices communicate with others. These are truly exciting advances in technology, but they must be paired with advances in data and device security. Despite the benefits that technology provides, healthcare at its most fundamental level is truly personal. Our ability to automate more and more aspects of healthcare has the potential to radically improve care while driving down costs, but we need to make sure that these advances retain the human touch.