10 MIN READ
More than just baby pictures: The future of ultrasound in medical devices
Ultrasound only became widely available in hospitals in the 1970s and yet within 50 years it has become an essential healthcare tool. The usual perception of ultrasound is that it’s a cheap, safe and easy-to-use technology. A standard ultrasound system costs between $20,000 to $75,000 with the most expensive systems costing around $200,000. This is a bargain in comparison to MRIs or CT scanners which can cost in the millions. Furthermore, ultrasound is highly portable unlike CT and MRI systems which are expensive to install and need dedicated facilities.
Although ultrasound is more accessible, the image resolution which can be achieved has generally been considered inferior to that of modalities such as CT and MRI. However this perceived gap in imaging quality is becoming increasingly small. Over the last half century, ultrasound technology has evolved hugely: gone are the barely decipherable prenatal scans, to be replaced with real-time three dimensional videos which can show a baby wriggle and kick. Before looking at how these developments have been made and what the future may hold for ultrasound it is first important to understand the fundamentals of the technology.
Despite being best known for imaging babies, ultrasound is used to image almost every part of the body, although it isn’t good at seeing through air or bone. By harnessing the Doppler effect ultrasound can be used to measure and visualise blood flow. This is particularly useful in cardiovascular medicine, for example the flow of blood through the heart valves can be imaged to determine if the valves in the heart are correctly opening and closing and blood is flowing through the heart as expected.
When ultrasound gets agro
Before we look at how this wonderfully safe imaging technology has improved over recent years we should pause and briefly consider how ultrasound can also be used more aggressively.
A small amount of energy is inevitably lost in the process of sending ultrasound waves into the body and this absorbed energy is turned into heat. In the case of ultrasound imaging the transducer power and frequencies used mean that there is negligible heating within the body. However if the power and frequency settings are changed this extremely safe technology can become a destructive tool. By reducing the frequency and increasing the pressure waves an area of tissue where the ultrasound is focused can be heated. The heating effect is used in cancer treatment as a non-invasive way to destroy tumours in the body. This is also used in physiotherapy applications.
If the pressure is further increased a phenomenon called cavitation occurs. This effect is the result of a tiny vapour bubbles forming inside the tissues and then collapsing. If the negative pressure phase of the ultrasound cycle is sufficiently low, it will cause the liquid to instantaneously vaporise into a tiny bubble of steam. The bubble expands in the negative pressure phase of the ultrasound wave but as the positive pressure peak of the wave arrives the bubble suddenly implodes. This results in localised shockwaves and extremely high temperatures. Lithotripsy uses this powerful effect to non-invasively break apart kidney stones.
How ultrasound imaging has improved
Many decades of ultrasound research and development have greatly reduced cost, reduced size and enhanced image quality. Improvements in computing have also played a major part in improving ultrasound systems. Yet the fundamental design of ultrasound transducers hasn’t changed significantly over this time and the price of transducers remains limited by the cost of producing piezoelectric crystals.
This might be about to change with advances in microelectromechanical systems (MEMS) enabling the production of capacitive micromachined ultrasonic transducers (CMUT) to replace piezoelectric crystals. Instead of a whole crystal expanding and contracting, a
silicon membrane on a capacitor is deflected like a drum to produce an ultrasound wave. Capacitors are used in almost any circuit board and, as a result, have greatly benefitted from the advances in electronics over the years. This should mean that large numbers of CMUTs can be fabricated inexpensively, which opens up the potential of complex two-dimensional arrays. Additionally, it is possible to build the transducers on the same silicon wafer as the electronic drive circuits, thereby reducing size and cost, potentially making ultrasound low cost and small enough to find a host of new uses within healthcare.
Piezoelectric crystals are a type of material that responds to an alternating current by expanding and contracting. If the frequency is between 20 Hz and 20,000 Hz it will produce a pressure wave that can be heard by humans but above 20,000 Hz it will no longer be detectable by ear and is labelled ultrasound. To image within the body an array of piezoelectric crystals is manufactured into a transducer. This is placed in direct contact with the skin and a series of ultrasound waves are projected into the body. Different organs and tissues within the body have slightly different acoustic properties and, as a result, when the ultrasound reaches the boundaries between them, some of the ultrasound is reflected, while the rest is transmitted or absorbed. When these reflected pulses bounce back and hit the transducer it causes the piezoelectric crystals to expand and contract (albeit ever so slightly) which in turn causes the crystals to emit a very small electrical signal. The signals are amplified, analysed and the time taken for these reflections to be received across each of the crystals can then be interpreted to create an image.
Future developments and their potential uses
As ultrasound systems gain resolution, become cheaper and smaller they are finding a growing number of uses within healthcare. We’ll finally look at some of the emerging applications where ultrasound could bring about transformative changes to healthcare. Some of these applications will be enabled by incremental improvements while others will only come about through step changes in ultrasound technology, such as capacitive micromachined ultrasonic transducers (CMUTs).
There has long been an interest in giving surgeons ‘smart tools’ which can identify different tissues and structures within the body. Several optical and spectroscopic techniques have been evaluated and demonstrated in the academic world, although none have yet made it onto the market. Miniaturised ultrasound probes could potentially be incorporated into surgical tools facilitating visualisation of structures within the surgical site during surgery. A tool which detects the presence of underlying structures would be valuable navigational aid to the surgeon.
This isn’t beyond the realm of possibility since the size of an ultrasound array is only limited by the desired wavelength of the ultrasound. And if the array is only imaging relatively shallow tissues (e.g. a few centimetres) then the transducer might only measure a few millimetres, small enough to be mounted onto surgical tools or other interventional devices. The vision is that the surgeon will be able to check what type of tissues they’re touching before cutting, this could help locate tumours or avoid other structures such as nerves or blood vessels.
Most ultrasound systems are currently quite large and require significant training to both acquire good images and to interpret them. This limits how and where ultrasound can be applied, but technology is gradually reducing costs, reducing physical size and reducing the skill required of the operator. All of which could enable ultrasound systems to become a routine diagnostic tool at the point-of-care.
There would be huge healthcare benefits in making ultrasound an accessible and easy-to-use diagnostic tool, within the hospital and pre-hospital setting. For example, paramedics could differentiate ischaemic from haemorrhagic stroke in the ambulance or accurately assess the severity of myocardial infarction before the patient even reaches the cath lab. Handheld point-of-care devices are beginning to reach the market. For example, the VScan, developed by GE Healthcare, is a compact unit with a hand-held piezoelectric transducer. It allows doctors to perform cardiac and abdominal examinations on patients within the emergency room. A study in 2015 (Andersen, 2015) showed that the VScan significantly altered the primary diagnosis of 6.5% of ER patients and provided additional diagnostic information for a further 24% of patients.
In 2018 Butterfly iQ announced the planned launch of a portable ultrasound device using CMUTs costing less than $2,000 which could be operate at multiple frequencies. At such a price, hospitals would no longer be limited by just a few devices in a hospital. Instead
physicians could use it on any patients coming in through the door to improve diagnosis and treatment decisions. The range of frequencies would mean the Butterfly iQ could be used to image many different organs and tissues.
As hardware prices continue to fall there is likely to be a greater need to deskill image of interpretation at the point-of-care. This may in fact become the greatest bottleneck. Advances in computing may reduce this bottleneck but progress is unlikely to be fast and we will not see software making diagnoses for the foreseeable future.
The biggest challenge isn’t the technology itself, in fact the domain of machine learning or artificial intelligence (AI) is moving very quickly and is currently in vogue within the tech community. Machine learning has already been used to identify regions of interest in an ultrasound image and identify abnormalities (Huang, Zhang, & Li, 2018). By far the biggest challenge for automated analysis of medical images is the regulatory pathway. Notified bodies and regulators such as the FDA require the bar very high in terms of rigour and verification of any medical software. If software is going to be relied upon to interpret an ultrasound image and offer any kind of advice or make a diagnosis, then it will require an astonishingly high level of verification.
The idea that the software could employ machine learning and get better over time sounds nice but is a minefield from a regulatory perspective. How do you verify software which is constantly learning and evolving?
Over the last few decades we have witnessed astonishing progress in the consumer technology – from smartphones to drones – we now take for granted technology which was unimaginable 30 years ago. Ultrasound as benefited from such advances and it will become an increasingly powerful healthcare tool, but we have to be realistic and accept that the evolution of ultrasound won’t be able to keep pace with the world of consumer electronics. The regulatory environment for medical devices will probably have the greatest influence on how and where ultrasound can be applied in the next few decades. Let’s hope that intellectual hunger of experts in the fields of physics, materials and ultrasound can be matched by the skill, creativity and perseverance of our regulators as they seek to balance the risks of new technologies with the benefits they can bring to mankind.