Authors
Beenish Zia, Chief Architect - Medical Imaging and Health IT, Intel Corporation
Abstract: Breast cancer is the most common type of cancer worldwide, with nearly 2.26 million cases in 2020, resulting in 685,000 breast cancer deaths globally in the same year. It is concerning to learn such statistics and as an engineer it makes me wonder how technology can help humanity stand against breast cancer. In this blog, I want to highlight a few technologies that are poised to help advance mammography and improve breast care.
Introduction: These days I rarely come across anyone who has not lost a friend or family member to some type of cancer. According to the World Health Organization (WHO), cancer is the leading cause of death worldwide, accounting for about 10 million deaths in 2020 or nearly one in six deaths. Sadly, breast cancer is the most common type of cancer worldwide, with nearly 2.26 million cases in 2020, resulting in 685,000 breast cancer deaths globally in the same year. Roughly, half of all breast cancers occur in women with no specific risk factors other than sex and age. It is concerning to learn such statistics and as an engineer it makes me wonder how technology can help humanity fight against breast cancer. In this blog, I want to highlight a few technologies that are poised to help advance mammography and improve breast care.
In general, cancer burden as well as cancer mortality can be reduced by early detection and effective intervention and care for individuals who develop cancer. For breast cancer, mammography is one of the ways to stay on top of detecting abnormalities and triaging medical treatment as needed. A small percentage of women who undergo regular mammogram screening will receive a cancer diagnosis based on the results of the exam. For many women, there’s a high degree of anxiety and unknowns when awaiting results; therefore, it’s critical to not only use mammograms for early detection of breast cancer but also to provide quick relief to healthy individuals who might have concerns about abnormalities.
Several technologies can be used to reshape mammography to meet the growing demand for speed, precision, enhanced patient experience as well as streamlining clinical staff workflow.
The Goal: Efficient Mammography Process
There are three major challenges that technologists like myself are working to achieve the goal of making the overall mammography process more efficient.
- Reducing the time that the breast is compressed: During a mammogram, X-rays are applied to the breast which gets compressed between a paddle and the detector cover. This can be uncomfortable for the patient. Proper compression helps improve the image quality by reducing the X-ray scattering, allowing clinicians to visualize cancer at earlier stages. This enables healthcare teams to provide quicker intervention and treatment which improves the prognosis. Additionally, technicians want the display of the images while shooting X-ray so the breast position, if needed can be adjusted or anything unexpected can be fixed. Furthermore, for 3D mammograms multiple shots are done with near real-time processing to display images, all of which add towards the time the breast is compressed. It is not the processing itself that adds compression time, it is the thickness of the breast (that requires longer exposures to penetrate the breast) for 2D and the speed by which the gantry (that hold both Xray tube and Detector) moves and the acquisition rate of the detector during the movement for 3D , much like a CT scanner tomosynthesis. Hence, the time for which the breast is compressed is a critical area for improving patient experience.
- Reducing time between each view: During a mammogram, several views of the breast are taken. Reducing the time between consecutive views is important to improve the overall patient and technician experience. Reconstruction (or post-processing) time needs to be as quick as possible to allow a scan of the next breast view (usually two per breast in screening).
- Speeding up overall mammogram scan time for each patient: The mammogram workflow is constrained by the need for image display in near real-time requiring compute-intensive workloads compounded with the additional stressor of a high volume of patients needing to be scanned each day. Combined with limited mammograms and technician staff, speeding up overall mammogram scan time for each patient is critical. Not accumulating any computation between scans results in moving from one patient to the next faster, improves workflow for the institution, and reduces the time in the facility for patients, thus providing a less stressful experience.
Introducing Innovative Solutions
There are innovative solutions to tackle these challenges.
Our journey is guided by a few key principles:
- Rapid Results: One of the main goals is to speed up the mammography process. We want to significantly reduce the time it takes to capture and display images during the screening. This ensures that patients spend as little time as possible in health facilities, which can often be a stressful experience.
- Optimizing Workflow: The process of taking several views of the breast could also be more efficient. Reducing the time between each view is critical, making it a smoother experience for both the technicians and the patients. By minimizing computational steps, we can move to the next patient more quickly, improving the workflow for healthcare institutions.
The Role of Hardware and Software Innovations
We are focused on collaborating with the rest of the industry to speed up the exam while at the same time improving image quality. By leveraging technologies like the hybrid architecture in the modern Intel Core processors, technologists are able to dedicate efficient cores (E-cores) for background applications and performance cores (P-cores) for the main workload. Furthermore, having an integrated GPU (iGPU) based on Intel Xe architecture, allows for writing algorithms that could split computations for different views based on available hardware resources and workload balancing between them. For example, while data is being acquired for a breast view using specific hardware, in parallel computation for the previous breast view might be happening on different hardware resources on the same platform.
A low computational task can be running on the CPU P-cores, while background applications on CPU E-cores and in parallel a compute-intensive task can run on an integrated GPU or a dedicated discrete accelerator, like the Intel® Data Center Flex Series GPU. This combination speeds up the overall computation, provides a more cost-effective solution and accelerates the overall mammography workflow.
Software, such as oneAPI , can be leveraged to write portable software that is hardware vendor neutral while having the flexibility to optimize on the hardware of choice. The industry specification of oneAPI allows programming a unified code base that can run on a heterogeneous hardware platform. It's all about striking the right balance between cost and performance, using both hardware and software to make the mammography process efficient, scalable and cost-effective.
The Future: Expanding the Ecosystem
Along with its ecosystem partners, Intel is actively investing in Artificial Intelligence (AI) for medical imaging use cases and the portability of AI models, all with the aim of increasing the efficiency of mammography technology, including AI model portability through oneAPI and the utilization of OpenVINO. By focusing on portability, we ensure that our solutions aren't locked into a single vendor, allowing for a more open and collaborative approach to innovation. This not only streamlines workflows but also contributes to a more inclusive and equitable healthcare landscape.
As we move forward, the future of mammography promises to be more efficient, patient-centric, and focused on early cancer detection. It's a journey where technological advancements and collaborative partnerships play a vital role. Join us in this revolution as we strive to make breast care technology smarter and more compassionate, ensuring quality healthcare knows no boundaries.
Reach out to Intel's Health and Life Sciences team at health.lifesciences@intel.com or learn more about what we do at https://www.intel.com/health.
______________________________________________________________________
We'd like to hear from you! Let us know in the comments or reach out to discuss – what aspects of the mammography process or technology do you believe could be improved to make the experience better and more efficient?
If you enjoyed hearing from the Health and Life Sciences team and want to hear more, give this post a like and ensure you subscribe to get the latest updates from the team.
______________________________________________________________________
About the Author
Beenish Zia is the Chief Architect for Medical Imaging and Health IT at Intel. She is an ingenious electrical engineer on a relentless journey to evolve healthcare with technology. Reach out to learn more from her here.
The author would also like to acknowledge contributors Gregoire Armand, Gustavo Reyna, Kaeli Tully, Jamal El Youssefi, Jose Izaguirre, Thomas Skwirut and YongTong Chua from Intel Corporation for all their input and review.
References:
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.