Use of low-power FPGA easily develop portable medical imaging applications

In the past, medical device manufacturers focus on development issues such as x-rays, MRI and ultrasound and other large medical devices.

But today's medical device manufacturers are moving towards portable medical electronic devices. Global population aging, and the people for their own health concerns increase, requiring medical device implementation is easy to carry and more low-cost, to facilitate a hospital outside of use elsewhere. Therefore, medical device manufacturers for some time to develop such as portable ultrasonic diagnosis equipment, blood, and other complex digital image processing technology and advanced communication technology in personal health care and medical electronics.

With 90nm and 65nm technology node when semi-conductors, allowing low power and small dimensions and high integration of the medical device to appear.

For medical device manufacturers, the main challenge is how to choose the right semiconductor devices for portable medical electronic products of power, performance, and price requirements. With a large number of internal RAM and DSP processing power of the high and low-cost DSP Spartan-3A are very suitable for the implementation of these requirements. By Xilinx FPGA devices of Spartan-3A DSP through the use of highly parallel architecture provides tremendous computing power and hardware configurability, allowing developers to construct these portable medical devices for different medical imaging equipment upgrades and communication algorithms provide customized schema design.

In medical devices, Imaging is one of the most common treatment.

Medical imaging applications typically need a lot of processing power, dynamic range and clarity, and rarely is a standards-driven, through proprietary algorithms development difference product you can get more high-definition image quality and processing capabilities. In the past, typically using DSP MCU or to deal with medical images. But since the MCU and DSP is serial devices, developers need to use ASIC or FPGA for hardware acceleration for medical imaging applications required processing power and clarity. With the FPGA parallel processing capabilities, FPGA now can easily handle images in large-scale parallel algorithms for processing needs. FPGA devices can carry out efficient hardware acceleration lies in its internal multiple RAM modules of data access capability, allowing at the same time on multiple image blocks to parallel processing. High speed LVDS PCIe, and high-speed external memory interface facilitates data acquisition front side and the user interface/display system backend and FPGA between large-scale data input and output.

Image processing algorithms enhance usually require a flexible framework for FPGA and therefore very suitable for medical electronics market.

FPGA has begun to gradually replace other technologies, are expected to become medical electronics market growth fastest semiconductor devices. Just change the bit-stream will be able to make it have a programmable, flexible and reusable capabilities so relative ASIC, FPGA has the great advantage, because it is difficult to meet ASIC medical imaging devices to frequently upgrade requirements.

In the field of medical imaging, for example, in the realization of ultrasonic application of beam generated when need to calculate a series of highly complex algorithms, algorithm is essentially the multi channel.

Highly parallel structure of FPGA provides tremendous computing processing power and hardware configurability, allowing designers to realize its vision of algorithms developed custom schema, the schema generated by the calculation of the beam. Although general DSP processors can consume multimedia applications provide a good platform to achieve parallelism and create programmable FPGA for high-end computing platform is better in portable medical imaging application to provide hardware and software co-design of significant advantages. At this point, the Spartan-3A DSP or Virtex-5 DSP can be used to implement such as enter the parallelization (deserialization), interpolation filter, beam generated delay and Apodization (apodization) and other channel beam generation features. Since most images algorithm requires the use of floating point calculations, use CoreGen IP allows FPGA with floating point computation capability.

　　

　　

Figure 1: DSP48A Slice for filter optimized integrated pre Adder for Spartan-3A DSP provides exceptional storage capacity and DSP performance.

Generally speaking, Spartan-3A DSP is by Xilinx low-cost, low power products have a maximum memory and processing capabilities of devices DSP, FPGA vis-à-vis the competition, Spartan-3A DSP 3400A products with 25% power efficiency advantage, its lowest-cost device in 250Mhz frequencies DSP performance up to 4.06GMAC/mW, and therefore more suitable for portable medical imaging application; whereas there is applicable to Virtex-DSP strictly DSP processing needs of high-end 3D medical imaging beam generator.

Author: Lin Hong SEO

Asia-Pacific region, senior product marketing manager, DSP

By Xilinx Corporation