MEDICAL EQUIPMENT INDUSTRY FOCUS Bringing technology up to speed
Yvonne Lin, medical marketing manager at Xilinx, explains how medical equipment designers can use high performance FPGAs to boost performance while reducing power consumption, system cost and development time
C
onsumer electronics have taken great strides over the past decade in terms of affordability, size and power. In contrast, medical electronics have yet to experience a similar type of innovation. However, an increasing world population, longer life expectancy and rising standards of living should become catalysts for a medical devices revolution. Some challenges for medical electronics
designers include features differentiation - caused by an increase in competition between manufacturers - power reduction and development time. These challenges can cause engineers to make unwanted trade-offs between features and cost, or power and size. The design of ultrasound equipment provides a prime example.
FOCUS ON ULTRASOUND IMAGING Due to its non-invasive nature, ultrasound imaging is used across many medical applications. Ultrasound machines offering the highest resolution and 4D image viewing are still cart-based and bulky. Conversly, portable or handheld versions have limitations in terms of battery life and image quality. Clearly the performance-to- size ratio still needs to be improved. One could envision a future in which personal ultrasound machines could be marketed to expectant parents. For this to happen, however, the equipment would have to reach consumer level pricing, portability and ease of use. While that may be some time away, what are the products and platforms that can help solve the design challenges of medical designers? For years field programmable gate arrays
(FPGAs) have been central to the enablement of medical ultrasound technology. FPGAs have a long product lifecycle which can match the long time-in- market requirement for medical electronics. By allowing algorithms and features to be updated without needing to change- out components, FPGAs enable systems to become essentially future-proof. With every technology node shrink, FPGAs provide increased performance to price ratio. In addition, FPGA tools support an abstracted design flow, performing automated timing closure and minimising place-and-route time.
HOW IT WORKS An ultrasound imaging system can be split into three main functions - front end, transmission and back end. Programmable technology such as FPGAs play a crucial role in each of these blocks. The aim of an ultrasound front end is to control transmission of ultrasonic pulses and capture the reflected sound data. Analog- to-digital converters (ADCs) are used to convert the reflected ultrasonic data into a digital format. Cart-based ultrasounds require high resolution (128 or even 512 channels), whereas a portable ultrasound that must meet low weight and size targets may have as low as eight channels. To support the high volume of data transmission, ADC outputs typically use low voltage differential signaling (LVDS). Because an FPGA has a large number of input/output (IO) blocks, it can aggregate data from several ADCs for pre-processing. Manufacturers seeking the highest scan resolution are pushing the limits in terms of the number of channels that can be supported, to handle high volumes of data. To alleviate this, several ADC manufacturers have adopted the JESD204B standard as the new serial protocol to support data transmission at Gbit/s rates. A JESD204B receiver configured inside the FPGA allows data to be received from the ADC at up to 12.5 Gbit/s per lane. Beam forming is a mechanism to control the sound pulse to amplify the key area of focus and de-emphasise the non-relevant area. To perform the required computations efficiently, digital beam forming is displacing traditional analog techniques. The flexible hardware structure and DSP intensive architecture of today’s FPGAs, enables complex beam forming computations to be executed quickly. Finally, pre-processing comprises a wide range of data manipulation, such as gain compensation, filtering, data and enhancement. These algorithms are implemented as custom functions inside FPGAs in order to process large sets of data.
SAVING POWER Power consumption is often a key concern in ultrasound imaging, especially portable
/AUTOMATION
Above: Xilinx’s Zynq products are aiding a new generation of medical diagnostics based on real-time measurement and analysis
Below: some challenges for medical electronics designers include features differentiation
equipment which is limited by a fixed battery pack. Although power is not as critical in a corded system, the thermal issues related to power dissipation and power cycle management must still be considered. An FPGA fabricated in 20nm technology will have 20-50% lower overall power (dynamic, static and IO power combined) compared to 28nm technology. The dynamic and static power savings is an advantage in the front and back end, as beam forming and image reconstruction are intensive processes. Also technology allows various power-down modes for different blocks for optimal system efficiency. The reduced power consumption of a 20nm FPGA, combined with intelligent power management, provides extra margins for engineers to increase system performance and enhance resolution.
REDUCING DEVELOPMENT TIME The final block within the signal chain of the ultrasound system performs image reconstruction, which translates the raw data captured by the front end into an image. There are many algorithms written in C, C++, MATLAB, or OpenCL that determine how the acquired data will be processed, converted and interpreted. This step involves complex matrix multiplication involving computationally intensive Fast Fourier Transforms (FFT). Often software engineers build models of the necessary algorithms, and hardware engineers translate these to Hardware Description Language (HDL) for implementation in the FPGA. This division of activity causes a disconnect in the design flow which can
increase development time. The solution is a tool that automatically converts C, C++, OpenCL or MATLAB constructs into HDL, automatically placing and routing the design into the FPGAs. This can result in up to 15 times shorter development time compared to a conventional manual design flow, thereby enabling medical technology designers to focus on core competencies in image construction and interpretation. As FPGA technology matures with
increased performance, reduced power and shortened development time, we can look forward to a future with next generation ultrasound imaging systems offering the advantages of improved scanning resolution and convenient portability, at an affordable price.
Xilinx
www.xilinx.com T: +1 408 559 7778
Enter 222 AUTOMATION | MARCH 2014 37
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72