Inter-AI Agreement in Measuring Cine MRI-Derived Cardiac Function and Motion Patterns: A Pilot Study.

Area of Science:

• Cardiovascular Imaging
• Artificial Intelligence in Medicine
• Medical Image Analysis

Background:

• Manual analysis of cardiac MRI for heart motion is time-consuming.
• AI tools automate cardiac MRI analysis, but their result consistency is unknown.

Purpose of the Study:

• To investigate the agreement between AI-powered tools for measuring cardiac function and motion from cine MRI.
• To evaluate the consistency of automated cardiac MRI analysis.

Main Methods:

• Cine MRI datasets from 23 healthy volunteers were analyzed using two AI tools: Heart Deformation Analysis (HDA) and Circle CVI 42.
• Key cardiac function indices (e.g., LVEF, LVM) and motion indices (e.g., strain rate) were calculated.
• Agreement was assessed using t-tests, Pearson correlation (r), interclass correlation (ICC), and coefficient of variation (CoV).

Main Results:

• Systematic biases were observed in cardiac function and motion index measurements.
• Left ventricular ejection fraction (LVEF) and left ventricular mass (LVM) showed good agreement and were considered exchangeable between tools.
• Circumferential strain rate demonstrated good agreement between the two AI tools.

Conclusions:

• AI-powered tools for cine MRI analysis yield related but potentially differing results for cardiac function and motion indices.
• Variations in measurements obtained from different AI tools should be carefully considered when interpreting and comparing study findings.
• The study highlights the need for standardization or awareness of inter-tool variability in automated cardiac MRI analysis.