Detecting breast cancer early and minimizing false positives can make a critical difference in women’s healthcare. Through better detection, digital mammography can significantly improve radiology outcomes.
Analogic, a developer of direct conversion digital detector technology, makes advanced digital mammography available by developing and manufacturing flat-panel, direct conversion digital detectors used by major medical OEMs in mammography systems. By enabling digital tomosynthesis, Analogic’s detector technology provides a 3D image for improved reading accuracy and reduced false positives compared with traditional 2D film mammography.
To speed development of the next generation of digital mammography detectors, Analogic needed specific technical expertise to use high-performance image processing algorithms with its patented detection technology.
Challenge: Meet various market needs for advanced digital mammography image processing. As a premium mammography detection technology provider, Analogic needed to incorporate technically advanced image processing capabilities into a versatile new solution as quickly as possible. Because Analogic detectors are used on a range of equipment made by different manufacturers, the new system architecture design had to offer ease of integration and reuse to support multiple applications over an extended product life cycle (see Figure 1). The company also needed a solution that would allow older x-ray equipment to use its detector.
Solution: Collaborate with Orthogone to provide field programmable gate array (FPGA) technical expertise and speed development. Analogic brought Orthogone in as a partner to provide technical expertise and accelerate the development cycle. With experienced FPGA designers, embedded software designers, and hardware designers, the Orthogone team helped to develop the scalability that the Analogic team needed to achieve their product development goals.
“Our aim is to maintain our lead in the premium market for mammography detectors. In order to do that, we need high-speed communications from the detectors, and that’s definitely an area where we knew that Orthogone could help us. They brought expertise, especially in anything having to do with digital design, memory configuration, and FPGA programming,” says François Boucher, director of engineering at Analogic Canada.
Developing a state-of-the-art digital platform for medical imaging applications presented many multi-disciplinary technological challenges.
System Design and Architecture
System designers had to integrate extremely sensitive sensors with high-speed digital processing capabilities onto a highly integrated low-profile enclosure. Large-area flat panels often combine multiple sensors to form a larger high-resolution image, with high-speed links used to transfer the raw image data. In this case, the mammography sensors were divided into multiple independent sections that could each use multiple high-speed LVDS differential pairs to transfer their raw acquisition data. Data is sent to independent low-power Xilinx Artix-7 FPGAs installed around an MPSoC located in the center of the PCBA.
In this design, the Xilinx Zynq Ultra-Scale+ MPSoC integrates the Kintex UltraScale+ FPGA programmable fabric onto a single device with a processing system (PS) that includes a 64-bit dual-core Arm ® Cortex®-A53 and a dual-core Arm Cortex-R5. The architecture was designed to benefit from the FPGA fabric performances and multiple ARM cores flexibility, some of which can be used with operating systems that need to be certified for critical applications required in the healthcare industry. The design also integrates several high-speed peripheral controllers such as SATA and USB 3.0 into the Zynq MPSoC. The architecture is designed to be scalable by easily adapting to different sensor sizes.
Image Processing
Capturing high-speed raw data from the sensor is the first step in the image acquisition process. After the custom video DMA (VDMA) reorganizes the full image in external memory and all the pixels of the image are correctly reordered, the FPGA can enhance image quality by performing image processing operations, such as removing fixed pattern noise using flat field correction type processing.
Plenty of DSP block resources are provided by the Xilinx Zynq UltraScale+ MPSoC for image processing on the captured data. A ZU7CG MPSoC device that contains 1728 DSP slices is used on the platform. Image data can be routed to the DDR or directly to the back-end interface that will stream the image/video content on 10G Ethernet port(s) over a UDP session.
Hardware Design
The hardware design had to overcome multiple challenges, including minimizing noise, integrating high-speed digital with ultra-low noise circuits, physical dimensions constraints, and safety considerations. One of the factors in mammography detectors is chest wall constraints: the sensor has to be located as close as possible to the patient.
A combination of voltage regulators and low-noise buck converters were used to provide low-noise stable DC rails and reduce electromagnetic interference. The Xilinx Zynq can control and monitor all onboard supplies and disable unnecessary power rails to decrease power consumption when required.
At the center of the system is the Xilinx Zynq UltraScale+ MPSoC, which is responsible for retrieving the image sensing data coming from the Artix-7 FPGAs and creating the final image. For optimal speed and power consumption, DDR4 memory is used to store local frame buffers and to support acquisition at a fast frame rate. The Xilinx MIG (memory interface generator) automatically does the required calibration.
Software Design
Analogic’s detector software running in the Zynq UltraScale+ hardware platform is divided into different applications. These applications manage non-critical background tasks with Linux and real-time control of the image data path on the FPGA side. Multiple processors that have access to shared and private memory regions and peripherals are used by the Zynq MPSoC hardware platform. Based on the Mentor ® Embedded Multicore Framework (MEMF) library, most inter processor messaging goes through system shared memory.
Results: Deliver an industry-leading versatile solution for different medical industry markets. Working in partnership with Analogic, Orthogone used innovative design and technology implementation to develop a versatile, easy-to-use solution and deliver the project on time.
“We’ve been working with Orthogone for many years now, and we’ve had very good success. They helped us meet our development milestones, in spite of changing priorities, changing requirements, they were very flexible in adapting to our needs,” says François Boucher, director of engineering at Analogic Canada. “We have to adapt to market needs all the time, so sometimes the original idea of a product may evolve over time, and that changes some of the specifications some of the requirements so we need a lot of flexibility from our partner to change, adapt, and still meet the original milestone for the development. Orthogone has distinguished themselves helping us anchor our delivery schedule.”
With the new converter that Analogic and Orthogone developed in collaboration, older x-ray equipment can benefit from the Analogic detector for improved radiology. The flexibility, versatility, and technical capabilities that the team built into the new advanced detector allows Analogic to readily meet requirements from various manufacturers and expand into medical image processing markets beyond mammography detection.
This article was written by Alexandre Raymond, CTO of Orthogone Technologies Inc., Montreal, QC, Canada. For more information about Orthogone, visit here . For more information about Analogic, visit here .