ARTICLE FOCUS:

• Beamforming technology
• Architectures and implementations
• Innovations
• Market trends

Ultrasound is one of the fastest growing medical imaging modalities due to its significantly lower cost compared with other medical imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI). Ultrasound is a safe non-ionizing operation, it’s excellent at observing motion (such as blood flow), and it’s portable. A vast underserved healthcare market in third world and developing countries is adopting ultrasound to add diagnostic imaging capability to hospitals and clinics in small towns and villages. Ultrasound is also finding many new applications outside of the traditional radiology and OB/GYN departments. Regardless of their size, cost, or clinical use, all ultrasound machines employ transducers that are constructed of arrays of elements that transmit and receive acoustic energy into and from the body. Beamforming is a common signal processing technique used to create directional or spacial selectivity of signals sent to, or received from just such an array. The beamformer is one of its most critical components in the front-end electronics sub-system of any ultrasound machine.

This article will describe the technology, architectures, and a unique configurable beamforming implementation that is driving innovation and growth in the ultrasound market.

Beamforming function in medical ultrasound
Beamforming is a common signal processing technique used in various applications such as wireless communications, including cellular and WiFi (802.11n), RADAR, SONAR, and ultrasound. These applications use an antenna that consists of an array of sensors or elements, numbering from two, to thousands of elements for transmitting and receiving radio or sound waves. Beamforming is what allows for the directional or spatial selectivity of signal transmission or reception. By coherently combining the signals from the multiple elements of the array, transmit or receive gain is achieved. Beamforming can be applied to both the transmit and the receive paths of the system to achieve this gain. In medical ultrasound, transmit beamforming can be used to create focused beams of ultrasound using a phased array. By forming a pattern of beams pointing in the same direction, receive beamforming is used to focus the echo signals received as reflections from different tissue structures in a region of interest within the human or animal body. In the receive beamformer, focusing is achieved by appropriately delaying echo signals arriving at different elements to align them in a way that creates an isophase plane. These aligned echoes are then all summed coherently, which is what provides the processing gain. This basic signal processing is called delay-and-weighting function in the time domain.

High-performance medical ultrasound beamforming poses some unique challenges to designers. Unlike RADAR or SONAR applications, receive beamforming in ultrasound is unique in that the system must capture reflections from various depths both very close and further away from the transducer. This requires dynamic focusing and an apodization capability which is a major source of differentiation for manufacturers.

Architectures and implementations
Historically, the beamformer was implemented in the analog domain. Variable analog delay lines delayed the signal from each element or channel, followed by an analog adder. Then the beamformed data was sampled by an analog-digital-converter (ADC). In contrast, in digital beamformers, more commonly used today, the signal from each channel is digitized upfront using an ADC, followed by a memory device to implement the phase delay, a multiplier to weight (apodize) the signal, and finally an adder to sum the delayed and apodized data from all the channels.

Several trends in the last decade, both in semiconductor devices and in the ultrasound market, have determined the technology used for implementing the beamformer function. Large manufacturers traditionally designed and built custom application specific integrated circuits (ASICs) that implemented the digital beamformer function. However, as ASIC process technologies advanced to 65 nm and below, the nonrecurring engineering costs for ASICs became difficult to amortize across the limited volume for a single ultrasound OEM. At the same time, field programmable gate array devices (FPGAs) have become larger and more capable of implementing sophisticated digital signal processing (DSP). These high-end FPGAs thus become the preferred device in which to implement the beamformer. The advent of DSP functions in low-end, low-cost FPGAs also allowed many smaller ultrasound companies to implement their own beamformer in these devices. Still, the performance of the beamformer algorithm is directly related to the amount of logic and memory of the chosen FPGA, with more sophisticated algorithms requiring more sophisticated and expensive FPGA.

Additionally, the high power consumption of an FPGA beamforming implementation can be the limiting factor on battery life thermal envelopes especially in small-form factor ultrasound machines. Systems in this segment, such as hand-carried ultrasound (HCU), which has a laptop form factor, can operate exclusively on batteries. While these machines initially supported only rudimentary ultrasound imaging, clinicians are now demanding, and expecting, portable ultrasound machines to have the high-performance imaging modes once found only in high-end console systems.

Ultrasound OEMs wanting to bring small-form factor machines to market are now faced with a dilemma as development of a beamformer ASIC is extremely expensive, and FPGA solutions are either performance constrained (using low-end devices) or cost, power, and size constrained (using high-end devices). To address these issues, a new implementation is needed.

Configurable beamforming
Samplify Systems, a Silicon Valley semiconductor and solutions start-up, has developed an innovative beamformer technology using phased-array processing called AutoFocus. The technology is a highly configurable beamformer that allows each OEM to implement a highly differentiated beamformer. The technology’s high degree of programmability also allows the OEM to differentiate within its internal machine lineup.

Figure 1 shows a high-level block diagram of the AutoFocus beamformer. This device has a high-performance data path supporting parallel processing of four beams simultaneously for improved imaging and frame-rates. An on-board calculation engine updates the delay coefficients and apodization weights on every sample clock for continuous dynamic focus and apodization. The inputs to this calculator come from multiple user programmable registers and tables providing multiple layers of configurability.