AI Exploration & Leadership in Heart Sonography

AI Exploration & Leadership in Heart Sonography
 

ABSTRACT
A Heart sonography, or an echocardiograph, is a type of ultrasound used to look at the heart and surrounding vessels, and how blood flows through them. This test is widely common for heart health and allows doctors to diagnose diseases, monitor the heart and much more. Heart sonographers are highly trained to administer this test, but human error is always possible. AI is already implemented in machinery to increase the accuracy of the test, such as through automated measurements and improved image analysis. Furthermore, AI has been beneficial in making this test accessible to rural areas. As technology advances, we can expect it to be more involved, leading to faster, more accurate and accessible diagnosis.

Purpose

1. Researching AI Trends in Heart Sonography

2. Analyzing Case Studies

​

Research
Vscan Air SL with Caption Guidance software
The Vscan Air SL is a point-of-care ultrasound (POCUS) device, which uses AI to help sonographers get good images when administering an ultrasound. To start, POCUS devices themselves are breaking barriers with their accessibility and user-friendly features. POCUS refers to the process of bringing a probable ultrasound device to the patient rather than a patient having to come to a machine. This is beneficial in emergency situations as well remote settings where there is no proper machinery. The AI caption guidance software on this specific device gives sonographers real-time instructions as to where they should move the device to get better images. By doing this, doctors can more easily and precisely diagnose patients based on the ultrasound. This device was created by GE HealthCare, a globally well-established medical technology company. Using millions of examples of ultrasound probe motion and its impact on image quality, a machine learning model was trained. This feature allows those with less training to get high-quality images.

 

​​

Digital Networks

During COVID, many rural communities struggled to access healthcare. Sick patients would have to travel elsewhere to get diagnostic testing, but were not permitted due to the risk of spreading the virus. Health Care providers needed a new way to get these patients what they needed without seeing them physically. Dr. Tsang and her team in the UBC-VGH AI Echo Core Lab responded to this problem by creating a network which allowed rural healthcare providers to use Point-of-care ultrasound (POCUS) devices. Patients in rural communities could get the ultrasound and send it to Dr. Tsang’s lab for interpretation through the network. As time passed, they developed the network further with AI in order to get more precise images and even get automatic results.​​​​
 

 

Emerging Trends

A large emerging trend is Machine Learning. Machine learning is a type of Artificial intelligence which focuses on learning and finding trends in sets of data, rather than getting programmed to do so. This type of AI has already been extensively used in medical settings and will continue to improve efficiency and precision as technology improves. In heart health settings, the results of the echocardiogram and diagnosis are limited by the knowledge and experience of the sonographer or doctors, but machine learning algorithms can provide consistently accurate results. Another trend is the use of POCUS devices. These portable machines are becoming much more common in medical settings. Not only are they accessible and easy to use in emergency situations and low-resources settings, but also user-friendly for operators with less training. Its AI features allow all users to get high-quality images.

​

​

Case Studies

“Blinded, randomized trial of sonographer versus AI cardiac function assessment”
In 2023, about 3500 echocardiogram studies were randomly assigned to an AI group or human group of 25 sonographers for assessment. They were both tasked with observing and calculating the Left Ventricular Ejection Fraction (LVEF), or in other words the amount of blood which this specific chamber of the heart pumps out during every contraction in a percentage value. At the end of the study, it was found that the AI’s results were not inaccurate nor varied much from the sonographer group’s results. Furthermore, the AI group was considered to be more time efficient compared to the sonographers and had more consistent results since there was natural interobserver variability between the humans assessments. Finally, the AI results themselves were similar to that in a clinic setting, so Cardiologists could not see the difference between the AI and human sonographer’s results when presented to them.
 

“Enchancing Handheld Point-Of-Care Echocardiography with Artificial Intelligence: A prospective Clinical Trial”

In 2024, 659 patients, who had not recently done an echocardiogram test, were tested by untrained residents and interns with POCUS devices at Sheba Medical Centre. This device had AI features such as caption guidance to obtain high-quality images. 51% of the patients had been either effectively ruled out of diagnosis, diagnosed, had their medicine changed or needed intervention. This proved that AI-powered POCUS devices can be properly used by untraining operators. If a similar system was regularly applied in clinics, patients could be diagnosed or ruled out from diagnosis much quicker. 
 

Ethical Considerations and Challenges

AI integration in heart sonography has some ethical concerns and challenges which must be taken into consideration. Job displacement arises as AI reduces the need for highly trained sonographers. Those who studied to administer echocardiograms could be replaced by almost any untrained operator due to the AI features. Additionally, patient data privacy is a critical concern, needing proper protective software for sensitive medical information. Finally the risk of misdiagnosis, due to AI inaccuracies, is a large concern for many healthcare providers. If the results are incorrect, not only will the patient’s health be impacted, the administrator may also face legal implications. Although AI has shown much promise, there are still many barriers preventing widespread application. 

​

​

Interview

Staff Name: Adrien Boutain - Team Lead of Heart Sonography at St. Michael’s Hospital 

 

Q: How is AI currently used in your industry?

A: AI is currently new in my field. The main application I see so far is AI being able to take measurements of heart vessels and valves for us.

 

Q: How do employees interact with AI in their daily tasks?

A: So far, employees mainly only use things like ChatGPT for reports. “It's coming but has not been perfected yet. It is not used on a day to day basis yet”

 

Q: What new AI technologies or innovations are expected in the next 5-10 years?

A: In the next 5-10, I need technologies which will allow us to enhance the quality of our images and automatically take measurements. But again, this will take time. Generation over generation, the machines and technology will improve and be perfected.

 

Q:What skills are becoming important due to AI?

A: As the new technology comes, we must stay adaptable and have the ability to use AI in order to improve our imaging. We need to be informed of what it uses and how to properly operate it.

 

Q: Do you think AI will be displacing or creating jobs?

A: I think that there will be a bit of both. There will always need to be someone there to operate the AI machine, but it may be able to replace those doing certain repetitive tasks. I do not believe it is ready to replace my type of job at its current state, but it might someday. “You know the machines are getting enough that you can actually get a person to hold their own probe in places like the Northwest Territories, out in very rural areas. They are teaching people to hold the probe while we can see them on the T.V and tell them to move this way or that way. They can scan themselves.”

 

Q: Are there any ethical considerations or challenges that come with AI?

A: I think that there are always concerns but no real ethical challenges. Although, “the implementation of AI will take time and money. We need to get machines that we can’t afford. We don’t have those machines right now. It might also take a couple of generations of machines to perfect the actual technology.”

 

Comparison:

Adiren's insights aligned with my research findings. He highlighted AI's role in improving measurements and image quality, a key point in my own research. Additionally, his mention of portable probes enhancing accessibility in rural areas corroborated my findings. His observation that this technology is “coming but has not been perfected yet” was consistent with my research, which indicated ongoing development and numerous trials.

​

Conclusion

Key Takeaways:

1. AI not only improves the accuracy of echocardiogram results, but also makes the test more accessible to those in low-resource areas.

2. Although AI has shown much promise, it is not quite ready yet and will take time to be widely implemented into healthcare settings.

3. AI is revolutionary in the healthcare field and will help us create new innovations to increase efficiency and accuracy. 

​​

​

References

 

“AI-echocardiography: Current status and future direction.” Science Direct, 1 Mar. 2025, www.sciencedirect.com/science/article/abs/pii/S091450872500053X#:~:text=Currently%2C%20AI%2Dguided%20echocardiography%20devices,image%2Dacquisition%20and%20parameter%20measurement.

​

​Alsharqi, M., et al. “Artificial Intelligence and Echocardiography.” Echo Research and Practice, vol. 5, no. 4, Nov. 2018, pp. R115–25. https://doi.org/10.1530/erp-18-0056.

​​

“ENHANCING HANDHELD POINT-OF-CARE ECHOCARDIOGRAPHY WITH ARTIFICIAL INTELLIGENCE: A PROSPECTIVE CLINICAL TRIAL.” Science Direct, 2 Apr. 2024, www.sciencedirect.com/science/article/pii/S0735109724043341.

​​

He, Bryan, et al. “Blinded, Randomized Trial of Sonographer Versus AI Cardiac Function Assessment.” Nature, vol. 616, no. 7957, Apr. 2023, pp. 520–24. https://doi.org/10.1038/s41586-023-05947-3.

​​

Kim, Seungjun, et al. “Artificial Intelligence (AI) Applications for Point of Care Ultrasound (POCUS) in Low- Resource Settings: A Scoping Review.” Diagnostics, vol. 14, no. 15, Aug. 2024, p. 1669. https://doi.org/10.3390/diagnostics14151669.

​​

UBC Faculty of Medicine. “Bringing AI to the Heart of Cardiac Care - UBC Faculty of Medicine.” UBC Faculty of Medicine, 19 Sept. 2024, www.med.ubc.ca/news/bringing-ai-cardiac-care.

​​​​