How AI-powered ultrasound is revolutionizing medical imaging, and what the future holds for medical diagnostics using ultrasound AI technology.

With the rise of Artificial Intelligence (AI) in the medical field, healthcare providers have been exploring opportunities this modern technology can bring to their practice, as well as how to mitigate concerns about its efficacy. Ultrasound powered by AI is revolutionizing medical imaging software and ultrasound procedures, helping to reduce human error and aid in image acquisition and interpretation. For instance, ultrasound AI can provide real-time support in decision-making processes, based on past and present ultrasound information. Elsewhere, AI is growing in its availability to inform and support patients who seek medical guidance. Of course, these are just a few of the endless benefits of AI in the medical field, many of which are still yet to be fully explored.

Understanding Ultrasound AI

A brief history of imaging technology

Starting with its earliest usage after World War I, ultrasound medical imaging technology underwent massive innovations during the early 20th century and it continued to enhance the understanding of specific body parts. Slowly, ultrasound practices began to intersect with diagnostic medicine during the 1950s (Newman and Rozycki, 1998), producing images of the body (Fisher and Winick, 2021) for medical interpretation. By the 1970s, ultrasound technology was normalized in creating images of the abdomen, pelvis, and other vital organs (Troxclair et al., 2011).

Today, ultrasound technology has advanced to the point where physicians can visualize more complex medical issues that were previously difficult to see. Modern developments within ultrasound technology include 3D imaging for increased accurate diagnostics, visualizing markers for inflammation, and of course, the development of ultrasound AI.

A quick introduction to AI

Artificial intelligence (AI), in general terms, studies agents that receive information from an environment, and then perform actions based on that information. AI encompasses many subfields, such as natural language processing, search, robotics, machine learning, etc. People interact with AI agents all the time: when using an application that guides them to a destination while driving, when looking at recommendations of products, music, or movies, or when interacting with virtual assistants on companies’ websites.

From all the subfields of AI, machine learning is one of the most relevant to the ultrasound field. Machine learning is a set of methods that automatically detects patterns in the data, and then uses those patterns to make predictions. This process typically involves the following steps:

1. Collecting a large amount of annotated data
2. Input the annotated data to a machine learning algorithm that will create a predictor
3. Use the predictor to make predictions on new data

For example, to identify the left ventricle of the heart in ultrasound images, data scientists will usually start by collecting thousands of images of the heart. Then, medical experts will manually delineate the contour of the left ventricle. This information will be passed to a deep-learning algorithm to produce a convolutional neural network. Finally, this convolutional neural network will be used to identify the left ventricle in new ultrasound images. This process is illustrated in the figure below.

Clinical Benefits of Ultrasound AI

The goal of adding AI to ultrasound imaging is to aid physicians during different phases of the ultrasound imaging process, from image acquisition to image interpretation and medical report generation. For example, it can automatically detect if the positioning of the probe is optimal for the desired imaging plane. When this is not the case, it can provide guidance to the user to achieve the best possible image. Similarly, it can compute in real-time, and without human input, measurements that nowadays require tedious manual labor and require considerable amounts of time.

AI has vastly improved the clinical benefits of ultrasound technology. For instance, AI can provide accurate, sensitive, and specific interpretations of focused assessments with sonographies in trauma (FASTs). Using AI to analyze FASTs, clinicians and emergency care providers can instantly identify life-threatening hemorrhages within patients, especially those who are hemodynamically unstable (Levy et al., 2023).

Besides being a valuable teaching tool, ultrasound AI can also reduce human errors when it comes to interpreting and analyzing patient data. At the point of care during urgent situations, clinicians may feel pressure and become more likely to misinterpret emergency ultrasound findings. Everybody makes mistakes, but when it comes to human lives, ensuring that errors are reduced is a priority. With the help of AI in ultrasound practices, clinicians and patient care providers can rest assured that there is another pair of eyes aiding in emergency situations. Ultrasound AI can help indicate if areas were previously overlooked and need more medical attention, or to simply ensure all areas of interest are properly explored.

What Are Some Use Cases for Ultrasound AI?

Surprisingly, you may have already interacted with AI in healthcare. Previous usages of AI in healthcare have helped advance not only medical research, but also the possibilities offered by AI-powered ultrasound. One of the most well-known, everyday applications of AI in the medical field is virtual assistants. Virtual assistants (VAs), such as chatbots, have a variety of functions: from scheduling appointments with primary care physicians to educating patients on treatment plans, these programmed applications are one of the key players today in solidifying the relationship between man, machine, and medicine. However, VAs are not the only AI healthcare applications that have positively impacted hospital settings worldwide.

Some of the most well-known cases of ultrasound AI come from around the world. For example, back in July of this year at Sheba Medical Center, non-cardiologists were able to use an AI program to report eight critical heart functions from just two imaging angles (Ghert-Zand, 2023). Elsewhere, the feasibility of hip dysplasia screening using AI ultrasound was tested on groups of infants, proving they were able to screen infants regardless of the user’s skill level (Jaremko et al., 2023). Everywhere, ultrasound AI is helping bridge the gap between common community health disparities and the professionals capable enough to aid those in need. As ultrasound AI continues to evolve, patients and physicians can look forward to efficient care delivery, coming to them soon.

Examples of the uses of AI in ultrasound include:

• Estimation of the bladder volume
• Guidance for acquiring the correct imaging plane
• Computation of strain and ejection fraction of the left ventricle of the heart
• Detection of anomalies in lung scans
• Detection of developmental hip dysplasia
• Characterization of nodules using the TIRADS
• Gallstone detection
• Needle guidance
• Inferior vena cava (IVC) assessment
• Cardiac wall thickness (IVSd and PWd measurements)
• Obstetric and nerve assessment

Importantly, the accuracy of image capture and image interpretation could aid caregivers in doing their jobs faster and more accurately.

Will Ultrasound Techs Be Replaced by AI? The Challenges and Future of Ultrasound AI

AI's future depends upon how humans tackle the rise of ethical and medical concerns. In recent surveys conducted by Pew Research Center and PYMENTS Intelligence, around sixty percent of Americans said they would be uncomfortable with clinicians relying on AI in their healthcare. Fifty two percent of Americans also believed using AI to diagnose diseases and suggest treatments would harm the patient-provider relationships. The anxieties surrounding AI in medicine are nothing new, but the attitudes towards machine-made decisions in someone’s health persists in the present day. The need for regulatory bodies in verifying the efficacy of AI is clear. Any algorithms intended for medical usage, including AI, must attain proper FDA approval before being publicly implemented in healthcare settings.

Thankfully, patient care providers are working toward educating the public about the benefits and usage of AI in ultrasound. Many patient care providers strive to incorporate strategies that minimize potential harm from security breaches and ensure that the technology is effective for all patients, regardless of background and identity. As for the future of ultrasound AI, its potential to maximize patient outcomes through real-time results and accurate analysis is an attractive feature of growing medical technology. Ultrasound AI is projected to grow exponentially, according to Precedence Research, Korean Society of Ultrasound in Medicine, and more.

Conclusively, the practices of AI in ultrasound have yet to be fully explored. Boundless in opportunities, whether in educating new users, accelerating diagnostic procedures, or simply improving the process of providing quality care, ultrasound AI remains a field to be explored. The public can expect AI in medical settings to become increasingly common, especially given its wide arsenal of benefits for both patients and providers.

Distributing ultrasound, especially one that is AI-powered, to a broader base of users is easier said than done. However, the capacity to build operator independence, reproducibility, accuracy, and consistency across all ultrasound exams will immensely benefit the broader world of caregivers and patients. Coinciding with the rise of other technological advancements, hospitals, clinicians, physicians, and anyone involved in the medical field should expect to interact with AI at one point in their career.

What Sets Exo Apart from the Rest

Exo’s mission is to trailblaze new care pathways and modernize the medical landscape through our high-performance portable ultrasound machine, Exo Iris®, and artificial intelligence. We strive to deliver high-quality, affordable, and easy-to-use medical imaging for all. Exo’s ecosystem provides instant answers to aid critical decisions at the point of care to improve patient outcomes, and we back up our commitment to delivering reliable answers in writing. Our AI manifesto reassures patients and clinicians that we approach our AI products with clear intentions and clearly defined protocols. Through this unique approach to developing AI, we hope to deliver accurate and insightful answers.

We believe in the boundless opportunities granted by ultrasound AI and want our clinicians to treat their patients with the most accurate information on hand. Ultrasound AI aids clinicians in making the best possible choices for their patients.

Interested in learning more about the possibilities of AI in ultrasound? Visit our AI webpage to learn about our company’s mission and ethical usage of real-time AI in healthcare.