Automated Deep Learning Based Cardiac Quantification in Hypertrophic Cardiomyopathy: A Comparative Study with Manual...

Area of Science:

• Cardiology
• Medical Imaging
• Artificial Intelligence

Background:

• Hypertrophic Cardiomyopathy (HCM) is a common inherited heart condition.
• Accurate assessment of Left Ventricular (LV) function and Mitral Regurgitation (MR) is vital in HCM.
• Cardiac Magnetic Resonance (CMR) is the gold standard, but manual analysis is time-consuming.

Purpose of the Study:

• To compare a deep learning (DL) software (SuiteHEART) against manual segmentation (syngo.Via) for cardiac parameter quantification in HCM patients.
• To evaluate the accuracy and efficiency of automated DL-based cardiac segmentation.

Main Methods:

• Prospective study of 25 adult HCM patients undergoing CMR.
• Quantification of LVEF, LVEDV, LVSV, AoF, MR, and PG using both manual and automated DL segmentation.
• Statistical analysis included correlation and Bland-Altman analysis.

Main Results:

• Strong correlations found between DL and manual measurements for all assessed parameters (r=0.81-0.91, p<0.001).
• Bland-Altman analysis showed acceptable agreement with no significant bias.
• Automated segmentation significantly reduced post-processing time (p<0.001).

Conclusions:

• Fully automated DL-based quantification accurately assesses LV function, MR, and flow in HCM patients.
• DL algorithms can streamline clinical workflows for cardiac imaging analysis.
• Further validation is necessary for complex HCM cases.