With the continuous development and transformation of medical ultrasound technology, traditional cart-type ultrasound equipment is being replaced by handheld ultrasound equipment. These devices are so small that they fit easily into a lab coat pocket, and they are highly flexible enough to image all parts of the body, from deep organs to superficial veins, with a single probe. With advances in AI technology, not only well-trained sonographers, but also untrained professionals will be able to operate these devices in a variety of environments.
In 2018, Butterfly Network introduced the first miniature handheld ultrasound probe. In September 2023, the company Exo Imaging released a competitive product.
The key to achieving this is silicon ultrasound technology, which uses a microelectromechanical system (MEMS) that integrates 4,000 to 9,000 sensors onto a 2×3 cm silicon chip. By integrating MEMS sensor technology and sophisticated electronics on a single chip, these probes not only retain the quality of traditional imaging and 3D measurements, but also open up new applications that were previously impossible.
Conventional ultrasound probes
Conventional ultrasound probes contain sensor arrays made of piezoelectric crystals or ceramic plates such as lead zirconium titanate (PZT). When impacted by electrical impulses, these plates expand and contract, creating high-frequency ultrasound waves that bounce inside them. Ultrasound waves are transmitted from the plate, into the soft tissues and fluids of the patient's body, and create an echo signal. Capturing the echoes of these ultrasound waves is like standing next to a swimming pool and trying to hear the voices spoken underwater. As a result, transducer arrays are made of multiple layers of material that smoothly transition from a hard piezoelectric crystal in the center of the probe to the soft tissues of the body.
The frequency of energy transmitted to the body depends mainly on the thickness of the piezoelectric layer. A thinner piezoelectric layer transmits higher frequencies, resulting in smaller, higher-resolution features in ultrasound images, but with shallow imaging depths. Thicker piezoelectric layers have a lower frequency and can enter the body more deeply, but with a lower resolution.
As a result, multiple types of ultrasound probes are needed to image various parts of the body, ranging in frequency from 1 MHz to 10 MHz. To image large organs deep in the body or a baby in the womb, doctors use a 1 to 2 MHz probe with a resolution of 2 to 3 millimeters that can penetrate up to 30 centimeters deep into the body. To image blood flow in the arteries in the neck, doctors usually use a probe between 8 and 10 MHz.
How MEMS changes ultrasound
The need for multiple probes and the lack of miniaturization mean that traditional medical ultrasound systems are bulky cart-style machines. Over the past three decades, MEMS has enabled manufacturers in a wide range of industries to manufacture precise, extremely sensitive components at the microscopic scale. This advancement has made it possible to manufacture high-density sensor arrays that can produce a full frequency range from 1 to 10 MHz, enabling imaging of a wide range of depths of the human body with a single probe. MEMS technology also contributes to miniaturization, allowing all components to fit into handheld probes. Add to that the computing power of a smartphone, and there's no need for bulky carts.
The first MEMS-based silicon ultrasound prototypes appeared in the mid-90s of the 20th century, and the core component of these early sensors was a vibrating micromechanical film that generated vibrations like a drum struck to produce sound waves in the air. Subsequently, MEMS transducers with two configurations appeared. One of these is called the capacitive micromachined ultrasonic transducer (CMUT), and Pierre Khuri-Yakub of Stanford University and colleagues showed the first versions.
CMUTs are electrostatic conversion ultrasonic sensors that consist of a vibrating film and a fixed electrode as a parallel plate capacitor. One of the plates, the diaphragm (the micromechanical film mentioned earlier), is made of silicon or silicon nitride with metal electrodes. The other plate – usually a micromachined silicon wafer substrate – is thicker and harder. In emission mode, an alternating voltage is generated between the diaphragm and the substrate backplane. The resulting electrostatic force causes the film to vibrate, emitting ultrasonic waves. In receive mode, the ultrasound causes the membrane to vibrate, allowing capacitance changes to be detected. When the diaphragm comes into contact with the human body, the vibrations send ultrasound waves into the tissues. How much ultrasound is produced or detected depends on the gap between the diaphragm and the substrate, which must be one micron or less. Micromachining technology makes this precision possible.
Schematic diagram of how CMUT works
Another structure is called a piezoelectric micromachined ultrasonic transducer (PMUT), which works similarly to a miniaturized version of a smoke alarm buzzer. These buzzers consist of two layers: a thin metal disc fixed to their perimeter and a thin, smaller piezoelectric disc bonded to the top of the metal disk. When a voltage is applied to a piezoelectric material, its thickness as well as from one side to the other expands and contracts. Due to the larger transverse dimensions, the diameter of the piezoelectric disc changes more dramatically, and the entire structure is bent in the process. These structures in alarms are typically 4 cm in diameter and are capable of producing an alarm scream of about 3 kHz. When the diameter of the film is reduced to 100 microns and the thickness is reduced to 5 to 10 microns, the vibration frequency rises to MHz, making it usable for medical ultrasound.
Exo Imaging has developed a handheld ultrasound device using PMUT technology
In the early 80s of the 20th century, Honeywell developed the first micromachined sensors using a piezoelectric film on a silicon diaphragm. It wasn't until 1996 that Paul Muralt, a materials scientist at the Swiss Federal Institute of Technology in Lausanne (EPFL), developed the first PMUT that could operate at ultrasonic frequencies [1].
Early development of CMUT
One of the big challenges for CMUT is how to make it generate enough pressure to send sound waves deep and receive echoes. Since the gap between the diaphragm and the substrate is minimal, the movement of the diaphragm is restricted, thus limiting the amplitude of the sound waves. Combining CMUT arrays of different sizes into a single probe to increase the frequency range also reduces the sound pressure because it reduces the probe area available for each frequency.
The solution to these problems comes from the Khuri-Yakub lab at Stanford University. In experiments in the early 2000s[2], researchers found that increasing the voltage on a CMUT-like structure creates an electrostatic force to overcome the restoring force of the diaphragm, which causes the center of the diaphragm to collapse into the substrate. The collapsing film may seem catastrophic, but it was later discovered that it was a way to both improve the efficiency of the CMUT and make it easier to tune to different frequencies. The increase in efficiency is due to the fact that the gap around the contact area is very small, which increases the electric field there. And the pressure increases because the large annular area around the edge still has a good range of motion. In addition, the frequency of the device can be adjusted simply by changing the voltage. In this way, a single CMUT ultrasound probe is able to efficiently produce the entire range of ultrasound frequencies required for medical diagnosis.
Ultrasound frontier
, Like6
Inside the Butterfly CMUT ultrasound probe: the diaphragm collapses into the substrate to produce sound waves
Since then, researchers have spent more than a decade understanding and modeling the complex electromechanical behavior of CMUT arrays, as well as solving manufacturing problems. On the manufacturing side, CMUT's challenges include finding the right material and developing the processes needed to produce smooth surfaces and consistent gap thicknesses. For example, a thin layer separating film and substrate is 1 micron thick and must be able to withstand a voltage of about 100 volts. If there is a defect in this layer, a charge is injected into it, and the device can be short-circuited when the edge or diaphragm touches the substrate, causing damage to the device or at least degrading its performance.
However, MEMS foundries such as Philips in the Netherlands and TSMC in Hsinchu have developed solutions to these problems. Around 2010, these companies began producing reliable, high-performance CMUTs.
Early development of PMUT
Early PMUT designs also struggled to generate enough sound pressure for medical ultrasound. However, the sound pressure level is sufficient to be useful in some consumer applications, such as gesture detection. In this "over-the-air" use, bandwidth is not important and frequencies can be less than 1 MHz.
In 2015, PMUT for medical applications received an unexpected boost with the introduction of large 2D matrix arrays for mobile phone fingerprint sensing. In the first demonstration of this approach, researchers at the University of California, Berkeley and the University of California, Davis, connected about 2,500 PMUT arrays to CMOS electronics and placed them under a layer of silicone rubber. When a fingertip presses against a surface, the device measures the amplitude of the reflected signal at a frequency of 20 MHz to distinguish fingerprints.
This is an impressive demonstration of integrating PMUT and electronics on a silicon chip, and it shows that large 2D PMUT arrays can produce frequencies high enough for shallow feature imaging. However, the application of PMUT technology to medical ultrasound requires a piezoelectric film with higher bandwidth, greater output sound pressure, and higher efficiency.
Geneva-based ST Microelectronics came up with a way to integrate PZT thin films on silicon films. These films require additional processing steps to maintain their properties.
In order to achieve greater sound pressure, the piezoelectric layer needs to be thick enough for the film to withstand high voltages, but increasing thickness reduces the bandwidth. One solution is to use an oval-shaped PMUT film that effectively combines multiple films of different sizes into one. This is similar to changing the length of a guitar string to produce a different tone. The oval diaphragm provides strings of multiple lengths on the same structure through its narrow and wide sections. In order to effectively vibrate the wider and narrower parts of the membrane at different frequencies, an electrical signal needs to be applied to multiple electrodes in the corresponding area of the membrane. This approach enables PMUT to operate efficiently over a wider frequency range.
From academic research to practical application
In the early 2000s, researchers began pushing CMUT technology for medical ultrasound out of the lab and into commercial development. Stanford University has founded several startups targeting this market. Leading medical ultrasound imaging companies such as General Electric, Philips, Samsung, and Hitachi have also begun developing CMUT technology and testing CMUT probes.
But it wasn't until 2011 that the commercialization of CMUT really began to take off, and a team with experience in semiconductor electronics founded the Butterfly Network company. In 2018, the launch of the Butterfly IQ probe was a transformative event. It is the first handheld ultrasound probe that can image the whole body and generate 3D image data through a 2D imaging array. The probe is about the size of a TV remote, only slightly heavier, and originally sold for $1,999, which is one-twentieth the size of a full-size cart-style machine.
Exo Imaging launched the handheld probe Exo Iris in September 2023, marking the first commercial application of medical ultrasound PMUT. Developed by a team with extensive experience in semiconductor electronics and integration, the Exo Iris is about the same size and weight as the Butterfly IQ probe. Its price of $3,500 is on par with the price of Butterfly's latest model, the IQ+, at $2,999.
With a size of 2 x 3 cm, the ultrasound microelectromechanical system (MEMS) chip in these probes is the largest silicon chip with a combination of electromechanical and electronic functions. Its size and complexity pose production challenges for the uniformity and throughput of the equipment.
These handheld devices operate at a low power, so the probe's battery is lightweight, can be used continuously for hours when the device is connected to a phone or tablet, and has a short charging time. To make the output data compatible with mobile phones and tablets, the probe's main chip performs digitization as well as some signal processing and encoding.
To provide 3D information, these handheld probes capture multiple 2D anatomical sections and then use machine learning and artificial intelligence to construct the necessary 3D data. Built-in AI algorithms also help doctors and nurses place needles exactly where they need them, such as biopsies in challenging blood vessels or other tissues.
According to a 2022 study published in NEJM Evidence [3], the artificial intelligence developed for these probes is so good that nurses without ultrasound training can also use portable probes to determine the gestational age of a fetus with accuracy comparable to that of trained sonographers. AI-based capabilities can also enable handheld probes to play a role in emergency medicine, low-income settings, and medical student training.
MEMS 超声技术刚刚起步
This is just the beginning of miniature ultrasound technology. Currently, several of the world's largest semiconductor foundries, including TSMC and STMicroelectronics, produce MEMS ultrasound chips on 300mm and 200mm wafers, respectively.
In fact, STMicroelectronics recently established a dedicated piezoelectric thin-film MEMS "Lab-in-Fab" in Singapore to accelerate the transition from proof-of-concept to high-volume production. Philips Engineering Solutions [4] provides CMUT manufacturing services for CMUT-on-CMOS integration, while Vermon [5] of France provides commercial CMUT design and manufacturing services. This means that startups and the academic community now have access to foundational technologies that enable new levels of innovation at a much lower cost than they did 10 years ago.
Industry analysts expect that as these studies are underway, ultrasound MEMS chips will be integrated into many different medical devices for imaging and sensing. For example, Butterfly Network is collaborating with Forest Neurotech [6] to develop MEMS ultrasound for brain-computer interfaces and neuromodulation. Other applications include long-term, low-power wearable devices such as heart, lung, and brain monitors, as well as muscle activity monitors for rehabilitation.
In the next five years, it is expected to see miniaturized passive medical implants based on ultrasound MEMS chips, using ultrasound to transmit power and data remotely. Ultimately, these handheld ultrasound probes or wearable arrays can be used not only for anatomical imaging, but also for reading vital signs, such as changes in internal pressure due to tumor growth or oxygenation of deep tissues after surgery. One day, fingerprint-like ultrasound sensors could be used to measure blood flow and heart rate, and wearable or implantable devices could produce ultrasound images as we sleep, eat, and live.
For the original link, please pay attention to the official account background reply MEMS to obtain
[1] https://www.nature.com/articles/s41378-023-00555-7
[2] https://ieeexplore.ieee.org/document/1235329
[3] https://evidence.nejm.org/doi/full/10.1056/EVIDoa2100058
[4] https://www.engineeringsolutions.philips.com/looking-expertise/mems-micro-devices/
[5] https://www.vermon.com/index.php
[6] https://forestneurotech.org/