Artificial intelligence-based echocardiography assessment to detect pulmonary hypertension

Area of Science:

• Cardiology
• Medical Imaging
• Artificial Intelligence

Background:

• Tricuspid regurgitation jet velocity (TRJV) on echocardiography is crucial for screening suspected pulmonary hypertension (PH).
• Artificial intelligence (AI) tools offer automated echocardiogram analysis for PH detection.
• This study assesses the performance of the US2.AI tool in a PH patient cohort.

Purpose of the Study:

• To evaluate the performance and utility of the US2.AI tool for automated TRJV measurements.
• To compare AI-derived TRJV measurements with manual measurements.
• To assess the diagnostic accuracy of AI-based TRJV measurements for PH detection.

Main Methods:

• Retrospective analysis of 1031 patients investigated for suspected PH (2009-2021).
• Echocardiograms evaluated manually and by the US2.AI tool for TRJV.
• Right heart catheterisation (RHC) used to confirm PH diagnosis based on 2022 European guidelines.

Main Results:

• The US2.AI tool achieved a high interpretation yield, measuring TRJV in 87% of echocardiograms.
• Excellent agreement observed between manual and AI-derived TRJV values (ICC 0.94).
• Automated TRJV measurements demonstrated diagnostic accuracy for PH comparable to manual measurements (AUC 0.88).

Conclusions:

• Automated TRJV measurements via AI are feasible, accurate, and reliable.
• AI-based echocardiogram evaluation offers diagnostic accuracy for PH similar to manual methods.
• Findings support the integration of AI tools in echocardiography for PH diagnostic imaging.