AI in Echocardiography for Measuring Heart Function
Author Information
Author(s): Sveric Krunoslav M, Botan Roxana, Winkler Anna, Dindane Zouhir, Alothman Ghatafan, Cansiz Baris, Fassl Jens, Kaliske Michael, Linke Axel
Primary Institution: Herzzentrum Dresden, Faculty of Medicine and University Hospital Carl Gustav Carus, TUD Dresden University of Technology
Hypothesis
A fully automated AI-based GLS measurement will perform as well as or better than operator-dependent (Manual) and semi-automated (SemiAuto) methods in terms of accuracy and efficiency.
Conclusion
Automated GLS analysis enhances efficiency and accuracy in cardiac diagnostics, particularly for novice users.
Supporting Evidence
- The AI method showed good agreement with Manual and SemiAuto methods.
- Interobserver agreement for GLS using the AI method was 1.00.
- AI demonstrated high feasibility with 89% of Echo exams being successful.
- Automated methods reduced measurement variability compared to Manual methods.
- Beginner cardiologists performed better with AI and SemiAuto methods than with Manual methods.
Takeaway
This study shows that using AI can help doctors measure heart function more accurately and quickly, especially if they are new to the job.
Methodology
The study validated AI-based GLS analysis against Manual and SemiAuto methods using 550 Echo exams, assessing interobserver agreement and performance of a beginner cardiologist.
Potential Biases
The proprietary nature of the AI algorithms may introduce bias due to lack of transparency.
Limitations
The study's single-centre nature may limit generalizability, and the proprietary nature of the AI algorithms restricts full transparency.
Participant Demographics
Patients aged 18 to 91 years, predominantly male (59%), with a wide range of left ventricular ejection fraction.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website