Tony Zarola, Healthcare Segment Manager, Analog Devices examines the processing techniques used to improve the medical image quality of portable ultrasound devices.
Over the years we have seen medical devices that treat conditions such as diabetes and cancer make their way out of the hospital and into our everyday lives. This trend is set to continue and as more products make their way onto the market, processor manufacturers have a pivotal role in making products mobile without compromising quality.
One product that is increasingly expanding into the portable domain is the medical ultrasound machine. As one of the most sophisticated signal processing machines on the market, the challenge is to enable this intensive signal processing without compromising system performance. This article covers the use of portable ultrasound devices, the processing techniques used, and looks at how Analog Devices has enabled these requirements in a portable device through the development of its SHARC series processors.
Portability without compromising performance
Critical care technologies like ultrasound systems command reliability and consistent quality whether in the clinic or at a remote location. Although advancements in low power consumption are driving technologies toward portability, there are still many fundamental components in a medical ultrasound system’s design that are necessary to bring full in-hospital capabilities to disaster areas and locations where ultrasound access has previously been unavailable. It is the medical device developer’s burden of responsibility to deliver a product that provides the highest level of performance in terms of image integrity within the many environmental, business and technical constraints that are typically imposed. For portable ultrasound specifically, system performance means being able to interpret images as clearly and as accurately as a tethered system – only now it carries the constraints of fitting a specific category of weight, size, battery life and cost. These design constraints require components that have real-time computational capabilities, low power consumption and, for product design considerations, are low cost and compact. With the emergence of portable ultrasound machines, the challenge to meet low power constraints while maintaining performance levels becomes even more daunting.
CW Doppler Imaging
Ultrasound imaging technology is based on Johann Christian Doppler’s principle that moving elements will emit a detectable frequency “Doppler shift” or sound. For example, an ultrasound image of blood density is created by the detection (“sound”) of blood’s movement by a beam directed into a blood vessel. There are two main modes of Doppler ultrasound imaging, Pulse Wave (PW) Doppler and Continuous Wave (CW) Doppler. Pulsed Wave Doppler transmits pulses of ultrasound along scan lines where the relative time taken between received signals is used to calculate the Doppler frequency – the pulsed nature of the transmitter allows blood flow location information to be derived.
CW Doppler Ultrasound
This article will focus on the second method, CW Doppler ultrasound, which detects and measures the velocity of moving structures within the body. Because of its continuous wave generation, CW Doppler has high sensitivity and low bandwidth requirements, typically <100 KHz, therefore it is particularly effective for assessing large blood velocities. CW Doppler’s high velocity detection is useful for diagnosis of congenital or valvular heart disease as high blood flow velocity profile tracking is fundamental for detection of these disorders.
As implied by its name, with CW Doppler ultrasound a continuous single frequency tone is transmitted from a transmission transducer (piezoelectric crystal), while a receiving transducer simultaneously records the acoustic echo ultrasound wave signal. Since the interpretation of the beat frequency (Doppler shift) determines the blood flow velocity and direction through the cardiovascular system, high performance signal processing in the CW path is a critical element in the accuracy of the measurement. The dynamic range of the CW Doppler signal is the largest of all signals in the ultrasound system, which is due in part to the leakage from the transmit signal across the receive path (caused by the half duplex nature of the signaling) along with the reflections that are generated from the stationary body parts that are close to the surface. Examining blood flow in a vein deep in the body will therefore result in a very weak Doppler signal, hence the wide dynamic range requirement in the entire CW path signal chain. Quality ultrasound system performance is therefore directly correlated to a well integrated signal chain implementation.
Floating Point Processing’s Dynamic Range
The exponentiation inherent in floating-point computation assures a much larger dynamic range – the largest and smallest numbers that can be represented - which is especially important when processing large data sets or data sets where the range may be unpredictable. As such, floating point processors are ideally suited for computationally-intensive applications like CW Doppler ultrasound. This dynamic range processing enables portable ultrasound systems with CW Doppler to detect the very low level signals. The function of the digital signal processing unit in the CW path is to implement at minimum, Wall filtering, Envelope detection and Fast Fourier Transform (FFT).
Full Signal Chain Integration
As with any complex technology, tightly integrated components contribute to the overall system effectiveness and performance. For a signal processing-intensive application like portable ultrasound, the speed and efficiency of the entire signal chain contributes directly to maintaining quality, despite its portable form. Maintaining strong signal integrity from receipt through the front end analog signal processing components to the digital signal processing and back is essential for accurate analysis.
Analog Devices has been active in developing processors that meet these criteria. The latest SHARC 2147x series DSPs and analog front end (AFE) components can process the ultrasound signal throughout the signal chain. The processors are targeted for compute-intensive floating point applications and have a 32-bit floating point compute unit with 40-bit extended precision capability resulting in wide dynamic ranges and very accurate computation, and are designed to operate at high frequency while dissipating low power. These processors use a low power process technology to reduce the total power, and with other power reduction techniques, consume very little power when idling. This combination of low active power and very low idle power extends battery life an essential requirement for portable ultrasound applications.
Chip-memory is another important consideration in the development of portable ultrasound equipment. By increasing the on-chip memory, devices can be developed with minimum BOM costs and increased portability. Furthermore it can increase the total performance of the system as it minimizes accesses to external memory. In some implementations the SHARC processor’s 5mb of on-chip memory is sufficient in itself, completely eliminating the need for external memory and further reducing the bill of materials (BOM) costs.
Getting the balance right is essential and for maximum chip performance, developers should use processors that feature dedicated hardware accelerators with independent compute units and DMA memory mapping, enabling parallel processing of (FFT) functions that are frequently used to decipher and analyze the various velocity components in the returned Doppler signal. In addition, offloading the FFT computation to this parallel engine also lowers the power consumption of the FFT processing cycles.
The analog front end (AFE) components optimize the analog signal chain performance while limiting board component count and minimizing power usage. Analog Devices’ AD9276 octal receiver contains processing capabilities for not only B-mode and PW Doppler mode imaging, but with an integrated I/Q demodulator, enables CW Doppler processing in a small form factor with extremely low power consumption.
The time gain control (TGC) path provides imaging quality for high-end ultrasound systems, and includes eight channels of a variable gain amplifier (VGA) with a low noise preamplifier (LNA), an anti-aliasing filter (AAF), and a 12-bit, 10 MSPS to 80 MSPS analog-to-digital converter (ADC).
Analog Devices’ SHARC DSPs such as the SHARC 2147x series, along with the analog front ends, allow for the transformation of sophisticated, highly reliable, compute-intensive technologies like CW Doppler ultrasound into a portable package for medical designers. SHARC 2147x series DSPs support the system developer’s design goals by keeping costs low, reducing system complexity through integration, and shortening development cycles - all without sacrificing the crucial target design goal: ultrasound functionality and reliability in the field that allows diagnostic capability equal to in-hospital systems.
For more information about ADI’s full portfolio of digital signal processors, software, development tools and support, visit www.analog.com.
About the author
Tony Zarola is a Strategic Marketing Manager in the Healthcare Segment at Analog Devices, Inc. where he has been employed for 22 years. In this role, Zarola is focused on defining the business strategies for the sub-segments of Medical Instrumentation and Patient Monitoring.
Throughout his career at ADI, Tony worked in a number of technical and marketing roles at ADI in both the sales and product line organizations. He started in a design support role, progressing to field applications support and then strategic business development in the broadband business unit. Prior to his role in the Healthcare segment, Zarola was a Product Line Manager in the Processor/DSP group at ADI, where he was responsible for the Portable product strategies, planning new silicon products, such as the BF52x family, building out the software and partner eco-systems to address a broad range of mobile embedded applications
No comments:
Post a Comment