Emerging advances in ultrasound

Ultrasound imaging became known for non-invasive diagnostics in healthcare and non-destructive material testing in industry. Historical as well as recent advancements in hardware, image and signal-processing algorithms, machine learning and AI, material science and precision manufacturing have considerably boosted its versatility. As a result, the quality and diversity of ultrasound solutions for applications in many different fields have grown significantly.

After the discovery of the piezoelectric effect in the late 19th century, the first ultrasound applications emerged in the early 20th century. For example, sonar systems were developed to detect submarines during World War I. From there on, pioneers in ultrasound tried to apply the technology for imaging the human body, which succeeded in the 1950s by visualizing the fetus in the womb of a pregnant woman. Technological advancements resulted in more compact, commercial ultrasound-based systems.

Nowadays, ultrasound is a very versatile technology that can be tailored to specific applications. Adjusting parameters like transmitter and/or receiver geometry and emitted frequency of the sound waves enables use of the technology across scales ranging from micrometers (microscopy) to hundreds of meters (sonar). As a result, ultrasound solutions for applications in many different fields have been developed, from the automotive and petrochemical industries to underwater exploration and medical imaging and diagnostics.

Advancements in material science, precision manufacturing and digital signal processing open up even more potential applications by moving toward increased sensitivities and higher sampling frequencies and by miniaturizing the hardware. Ultrasound remains a viable option, compared to other sensing or imaging techniques, because of its advantages. These include real-time performance, high temporal and spatial resolution, device portability, safety, accessibility, relatively low cost and absence of any ionizing radiation.

Applications

Thanks to ultrasound’s device portability, it has been an early responder to the point-of-care trend. This has been facilitated by advances in wireless-probe architectures, electronics miniaturization for beamforming and cloud connectivity. At any location, for example at the general practitioner’s or even at home, patients can be screened quickly using small ultrasound devices. Only when something suspicious has been detected, they’ll be referred to a radiology department for more advanced diagnostics. Probes can now even be connected to a smartphone or tablet with a cable or wirelessly, creating a portable device that can be used for care at home, or for imaging/diagnostics in more remote areas, particularly in developing countries. It has also promoted ultrasound application in the veterinary practice and/or in harsh environments.

In the medical field, the number of applications has grown, ranging from arterial dimension measurement to needle guidance in minimally invasive procedures. Indeed, ultrasound is still primarily known as an imaging modality for diagnostics, but it has moved to therapeutic applications. There, the ultrasound energy applied is often converted to a thermal or mechanical effect in the patient’s tissue, controlled by pulse frequency, duration and intensity. Therapeutic applications include the use of shock waves to break down large kidney stones and of focused ultrasound to locally generate heat to ablate tissue, such as tumor tissue, with sub-millimeter resolution.

Ultrasound is also widely applied across non-medical fields. For example, ultrasound-based distance sensors are used in the automotive industry for parking assistance, outperforming the light-based lidar distance sensors, especially in low-visibility conditions or on optically non-reflective surfaces. In chemical plants, they’re used for contactless level sensing in tanks. Another application is in the manufacturing industry, where high-frequency ultrasonic vibrations are used to locally heat and melt plastics for welding them together. Ultrasound also plays a key role in non-destructive testing (NDT) of materials and structures, enabling the detection of internal flaws without causing damage. Recently, more advanced applications have emerged, such as ultrasonic resonance force microscopy in subsurface semiconductor metrology.

ML and AI

Over the years, advances in probe materials and electronics have fueled the increase in ultrasound imaging quality and focusing capabilities. This has led to new applications, with higher center frequencies, larger bandwidths and higher frame rates. The use of dedicated transducers and optimal (pre)settings for specific procedures enables the fine-tuning of ultrasound equipment to dedicated applications. In addition, optimizing chip architecture design and transducer impedance ensures the real-time character and integrity of ultra-low-noise signal transmission. Thus, ultrasound monitoring has moved from qualitative to quantitative measures and, recently, from large, bulky wired systems to portable and wireless solutions as well.

Ultrasound hardware development has always gone hand in hand with image-processing algorithmic science and related automation advances. Improved ultrasound scanning techniques have been combined with more sophisticated image and signal-processing methodologies, enabled by the ever-increasing hardware processing power. Automated analysis methods have been applied to ultrasound image segmentation, classification, speckle-tracking, de-noising and quality improvements, simulation and interpretation.

Now that machine learning (ML) and artificial intelligence (AI) are becoming mainstream in data processing, there are multiple ultrasonic imaging domains that can be approached from a very relevant data science perspective. Such ‘black-box’ data-driven approaches can integrate and synergize with more classical, ‘explainable’ algorithms, so as to automatically and efficiently process large amounts of ultrasound frames for data conditioning (de-noising) and processing, digital twins and simulations, and more.

In some cases, classical signal/image-processing techniques are preferable due to their proven performance and speed, while in cases that are computationally less well-defined and/or less deterministic, ML/AI methods may come into play. For instance, either physics-based machine learning or black-box deep learning can be used in algorithm design for image analysis and reconstruction, to ultimately support image interpretation and decision-making. Moreover, both standard algorithms and ML/AI can support ultrasound system design, for instance to optimize the tailor-made design of the beamforming unit for plane-wave imaging or of new, software-enabled modalities such as retrospective transmit beamforming.

Engineering solutions

Another driver is the progress in multi-physics modeling, for optimizing ultrasound product design as well as treatment planning, sensing and image reconstruction. Achieving the desired product functionality requires careful design decisions concerning transducer specifications such as center frequency and bandwidth, probe geometry and firing sequences. Simulating the behavior of the designed system, as well as the interaction of the sound waves with tissue, will provide valuable insights to guide these choices.

Increasingly, innovative engineering solutions are created using simulations of coupled physical phenomena, including mechanics, aero and vibro-acoustics and thermal behavior of structural materials and tissues. Think of soundwave propagation through media or the heating up of tissue through ultrasound irradiation. In this way, simulations help improve the effectiveness of ultrasound procedures. At the same time, simulations will provide insights into the safety of the designed product. Mechanical stress on tissue can be modeled, which can aid in setting safety limits for preventing unintended tissue damage.

Proof-of-principle setups can provide critical insight into key questions. Is the probe imaging resolution sufficient to see structures of interest? Is a probe powerful enough to deliver the required intensity? What’s the expected photoacoustic signal content from the tissue? Addressing such questions often necessitates precise characterization measurements to determine the pressure distribution and spectral properties of the ultrasound field. These studies are also crucial for obtaining certification of ultrasound devices. Certification, particularly concerning acoustic output, is governed by various IEC standards.

A specific challenge lies in the validation of medical ultrasound systems. Testing directly on patients is often not allowed due to safety concerns, while the variability between patients complicates standardized assessments. This is why so-called phantoms are widely used, for example, to assess image quality parameters like resolution and imaging depth. More specialized phantoms are designed for calibration and validation of, eg, Doppler imaging modes or quantitative imaging. One example is a phantom that’s flushed with blood with a known and well-controlled oxygen saturation for the validation of quantitative photoacoustic imaging.

Microbubbles

Recent developments are illustrated by projects in which Demcon is involved, such as the Capaflexus project of Delft University of Technology and Erasmus Medical Center in Rotterdam. It’s aimed at developing flexible, wearable echo probes, ie ultrasound transducer arrays, and the accompanying algorithms for monitoring the heart condition of a patient with heart failure over extended periods. The deliverable is a sticker/patch that the patient can wear on their chest for intensive yet unobtrusive cardiac monitoring.

Among the many therapeutic ultrasound applications still under investigation, one example is the combined use of ultrasound with microbubbles to reversibly permeabilize the blood-brain barrier for facilitating drug delivery and treating brain diseases. When exposed to ultrasound, intravenously injected microbubbles will start oscillating, thereby temporarily creating small pores in the blood-vessel wall. By focusing the ultrasound on specific regions of the brain, drug delivery can be improved very locally in a non-invasive manner. Demcon is developing a microbubble injector system for the formation and controlled administration of microbubbles to improve therapeutic effectiveness, reproducibility and safety of the ultrasound treatment.

Beyond the medical domain, sectors such as dredging, tunnel construction and mining require density measurement of the removed slurries. Traditionally, nuclear radiation is used for this. Alia Instruments, for which Demcon acts as an investment and engineering partner, has developed a non-nuclear alternative that’s safer and more reliable. The rubber tubes used inside the measurement chamber, however, are subject to wear. Ultrasound provided a solution: the thickness of the rubber tubes, as a measure of the degree of their wear, can be assessed by measuring the time of flight of a pulse through the measurement chamber. This non-destructive testing approach illustrates that ultrasound can also come to the rescue for industrial problems.