Skip to content

The Role of AI-powered Echocardiography Analysis To Diagnose Heart Failure with Preserved Ejection Fraction

Diagnosing heart failure (HF) can be a challenge for patients with preserved ejection fraction (HFpEF). 50 percent of HF patients with HFpEF are hospitalized, and it is a major cause of mortality. [1] In order to accurately and efficiently diagnose HFpEF patients, further measurements should be considered during echocardiograms scans, the most wide-used modality to diagnose and manage heart failure. 

In this article, we will detail what these measurements are, the challenges associated with obtaining them using manual operator reporting, and how AI-powered echocardiography can help reduce hospitalizations and improve outcomes.

Learn more about improving Heart Failure detection in echocardiography with the  help of Ultromics' AI - download and read the guide

Utilizing Global Longitudinal Strain To Diagnose HFpEF

Echocardiographers generally use measurements from left ventricular ejection fraction (LVEF) as a benchmark for diagnosing heart failure, with an LVEF under 40 percent evidencing heart failure. However, 50 percent of HF patients in the U.S. present with normal or preserved ejection fraction. [2]

This showcases the real need for more comprehensive measurement imaging which takes preserved ejection fraction into account. One way to achieve this is to measure global longitudinal strain (GLS). 

GLS measures the “stretchiness” of fibers in the heart’s myocardium, with negative values of more than -20 percent indicative of healthy left ventricular (LV) function. Changes in the composition and geometry of the LV myocardium, however, may lead to changes in LV deformation. This is especially significant where LV systolic dysfunction is present despite preserved LVEF. While these changes aren’t detectable by LVEF, they can be measured using assessment of global longitudinal strain. 

This is why the American Society of Echocardiography (ASE) and the European Society of Cardiology (ESC) suggest reporting and analyzing GLS for detection and prediction of HFpEF [3]. 

Related read: Download our ebook for further information about Strain analysis.

 

GLS Is Challenging To Measure Using Manual Echocardiography Reporting

While global longitudinal strain can help more accurately detect and diagnose HF in patients with preserved ejection fraction, it is not a straightforward addition to regular LVEF reporting. 

Measuring GLS requires increased training, specialized equipment, and additional imaging and analysis time. This may fall outside of budgetary constraints, and additionally creates the potential for a backlog of patients requiring echos. 

Also concerning is the variability of operator reporting. Manual observations are dependent on operator experience, and in some cases, the same operator may even reach a different conclusion when they re-examine the echo. 

For patients presenting with preserved ejection fraction, a more precise diagnosis is necessary. Variability in reporting, especially on global longitudinal strain, can be the difference between accurate and efficient diagnosis and delayed treatment. 

 

How AI-Powered Echocardiography Can More Efficiently Diagnose HFpEF

To address these concerns, Ultromics built an AI-powered service  which is predictive  and correlated to patient outcomes. [4] They collected a significant database of images over more than ten years, with a year’s worth of outcomes for each patient. These outcomes included cardiac biomarkers, events, and mortality. 

All of this data was combined to create EchoGo, revolutionary AI service  which allows for automation of LV GLS, EF and Volumes, without the additional time, training and variability. EchoGo connects through the cloud and sends reports to any care setting, once an echocardiogram is performed, allowing clinicians to take advantage of effective measurements with minimal effort or time.

One recent study shows that fully automated GLS measurements made by AI-powered EchoGo can be more reliable than human interpretation. [5] This is because the AI-based myocardial strain analysis has been evidenced to reduce variabilities and facilitate longitudinal follow-up of GLS in patients. 

With these factors in mind, AI reporting through EchoGo has been shown to be a significant predictor of in-hospital and follow-up mortality, with earlier detection and better prognosis in both HFpEF and HF patients. [5]

 

Conclusion 

With such a significant population of patients presenting with HFpEF, accurate and efficient diagnosis is crucial for long-term outcomes, and to prevent hospitalizations. GLS can be difficult to incorporate in routine clinical practice as it requires extra time, expense, and training, all of which could hinder the process of a fast diagnosis. Added to these challenges, cardiologists are under pressure to drive up echo volume while balance study quality, this makes GLS even harder to incorporate into practice.

This is why Ultromics' AI-powered solution. EchoGo, can be the way forward, with its ability to automate measurements, predict outcomes and send reports rapidly through the cloud at the point of care, it can make a real different to help identify heart patients with preserved ejection fraction. To learn more about how Ultromics can significantly improve heart failure patient outcomes, including a summary of the latest studies, download our guide: AI in Echocardiography: Improved Heart Failure Detection

 

Heart Failure detection - AI-supported diagnostics is proven to improve accuracy, save time, and enable repeatable, consistent readings. Download and read the guide to improving heart failure detection with the help of AI. Download

References:

  1. Assessment of Left Ventricular Function by Echocardiography: The Case for Routinely Adding Global Longitudinal Strain to Ejection Fraction
  2. Heart Failure with Preserved Ejection Fraction
  3. 2021 ESC Guidelines 
  4. Webinar: Using AI To Improve Heart Failure Prediction and Detection 
  5. Abstract 11383: Automated Measurement of Global Longitudinal Strain by Speckle-Tracking Echocardiography in Cardio-Oncology Patients Using Artificial Intelligence