How AI-Enhanced Cardiac Analysis is Empowering Clinicians on the Frontlines

GE Healthcare

The reason why the coronavirus impacts the heart remains one of the unknowns facing clinicians in managing patients during the pandemic.  Yet, recent data from the COVID-19 frontlines show that heart issues were a frequent cause of death for 50% of COVID-19 patients.[1]

Tackling this unknown together, cardiologists are collaborating with clinicians in the ICU, ER and other emergency settings, to help triage and monitor COVID-19 patients’ hearts.  

While point of care ultrasound (POCUS) has been shown to provide many benefits for imaging COVID-19 patients’ hearts, ultrasound analysis at the point of care is still done mainly visually, in a subjective process that relies on the user’s experience.

By automating the analysis process, AI-based software enables clinicians with various levels of ultrasound experience and training to work more effectively and consistently. And as ongoing COVID-19 concerns shape new requirements for safe and effective care delivery, AI also helps reduce POCUS scanning time to minimize contact with patients who may be infected.

To maximize the benefits of AI, GE Healthcare has partnered with DiA Imaging Analysis to offer the first AI-based solution for automated Ejection Fraction analysis on handheld ultrasound, LVivo EF on Vscan Extend.

To better understand the benefits of on-device AI for handheld ultrasound, GE Healthcare and DiA Imaging Analysis spoke with Dr. Ziv Dadon, physician at the Jesselson Integrated Heart Center at one of Israel’s leading medical centers, Shaare Zedek, about his experience using LVivo EF on Vscan Extend and how he is using it to guide patient care on the COVID-19 frontlines.

How has COVID-19 impacted your practice?

“COVID-19 has made the logistics of care delivery very complex. My ability to evaluate or interact with patients is limited. At any given moment, we’re completely covered from head to toe, using coveralls and respirators, and protected with two or three sets of gloves, . Practically, there’s no such thing as using my stethoscope to do a quick physical examination which makes my chances of successfully doing a physical cardiovascular or lung assessment very low. Under these circumstances, being able to use my handheld ultrasound has become critical.”

How are you using your Vscan Extend (handheld ultrasound) for COVID-19 patients?

“With the current restrictions and the risk of infection, every unit must have a dedicated device and unlike cart-based ultrasound, handheld ultrasound is much simpler to operate and clean since there are no buttons or power cords. In addition to being portable and much easier to disinfect, I know that I can get the measurements to help me evaluate patients for signs of cardiac dysfunction or deterioration. In the emergency department, we’re using Vscan Extend for lung and cardiac ultrasound and incorporating focused cardiac ultrasound into our workups of COVID-19 patients that are symptomatic.”

What are you looking for in these patients?

“Since the beginning of the pandemic there’s an increasing evidence of myocardial injury at up to 40% of COVID-19 diagnosed symptomatic patients. The exact explanation as to this injury drivers is assumed to be multifactorial and as for now, no single reason can solely explain this phenomenon. The injury scope was surprising at first, as COVID-19 is a respiratory disease, but we now know that the disease has significant systematic effects and it’s responsible for a grave inflammatory illness. We’re looking for signs of gross left or right ventricular dysfunction and potential right ventricular dilation, which we know are negatively linked to outcomes.

Now during this time of COVID-19, when a patient presents with troponin our tendency is to assume it’s not ischemic driven (heart attack). But even if that’s generally the case, we still should evaluate each patient individually to see whether the problem is ischemic or not.  This is when I would  check the Ejection Fraction with LVivo EF on those patients and if it’s low, further workup may be required to prove that troponin is not COVID-19-related, rather that this patient should be monitored with potential urgent medical attention, whether it’s treatment or, rarely, a life-saving intervention.”      

How are you using the LVivo EF application on your Vscan Extend to quickly get information about the heart?

“LVivo EF is a helpful decision support tool, especially for non-experts when you need to immediately decide if you should start treatment or send the patient for a full echo and especially if you are not used to visually evaluating EF on a routine basis. This is not an “academic” statement but entirely relevant with practical applications for routine clinical use. Even if you get a good quality image, it can be difficult to understand what you’re looking at and even among us cardiologists, who work with echos every day, there will often be a difference of opinion about the exact results. An AI solution like LVivo EF that provides automated and objective information for evaluating Ejection Fraction can help in this case.

We know that EF can help us with early detection of heart failure patients who may be at higher risk of severe illness from COVID-19 and worse outcomes. LVivo EF can be a quick tool that tells you where you should focus.”

You recently completed and presented a study on using this application with medical students to quickly onboard them, can you tell me more about this study?

“The purpose of the study was to see the effect of using AI on handheld ultrasound and we decided to examine the use LVivo EF on Vscan Extend. The results were presented at the virtual ACC conference last March. The study involved medical students who had no prior knowledge of ultrasound and received an 8-hour training session on how to estimate EF visually by looking at the apical 4-chamber view.

The students used Vscan Extend to perform a POCUS exam on patients in the cardiology ward in order to visually assess their EF. They had to do the assessment by eyeballing the exams. Afterwards they used LVivo EF to automatically evaluate the EF and both their visual and LVivo EF results were compared to the results of echocardiographers who analysed EF by eyeballing it and using the Simpson method.

We saw that using LVivo EF to evaluate EF enabled the medical students to achieve near-perfect reliability with the expert echocardiographers. This is tremendous when you consider the benefit for non-expert ultrasound users who need immediate, reliable support to evaluate EF at the patient’s bedside, and especially now given COVID-19 protocols and precautions.”

How can this tool help other clinicians who are less familiar with point of care ultrasound during COVID-19?

“Bedside ultrasound analysis today is mostly a subjective, visual process and therefore not simple. Of course, the more training or experience you have working with these images on a regular basis, the better off you are but it’s also a question of having the confidence to trust your judgement and to set a proper treatment accordingly.

That’s where AI and solutions like LVivo EF can come in to empower clinicians with information that’s objective and accurate so they can really put the benefits of ultrasound to use. I think this goes for the experts too but is especially important for someone who is relatively new to using ultrasound, which we’re seeing a lot of now, thanks to improvements that are making ultrasound technology more accessible.”

How do you think we can apply these learnings and technologies in a post-COVID environment?

“I think COVID-19 has in many ways made clinicians more open to embracing new technologies that can help them work better and smarter. AI automation has a growing important place in the use of ultrasound, and especially at the point of care where more and more clinicians are turning to ultrasound to make faster, more informed decisions when evaluating patients. When it comes to imaging in the ICU or the ED or when you’re talking about physicians who don’t use ultrasound on a routine basis, AI makes an important difference and I think a solution like LVivo EF has a central place in ultrasound analysis of the left ventricle.”

To learn more about the use of LVivo EF on GE Healthcare’s Vscan Extend, click here.

1. Shi S, Qin M, Shen B, et al. Association of Cardiac Injury With Mortality in Hospitalized Patients With COVID-19 in Wuhan, China.JAMA Cardiol. Published online March 25, 2020. doi:10.1001/jamacardio.2020.0950