Evaluation of cardiac biventricular segmentation from multiaxis MRI data: a multicenter study

Area of Science:

• Cardiovascular Imaging
• Medical Image Analysis
• Computational Anatomy

Background:

• Accurate volumetric assessment of cardiac magnetic resonance (CMR) images is crucial for diagnosing and managing cardiovascular diseases.
• Biventricular segmentation of CMR images presents challenges due to complex anatomy and motion artifacts.

Purpose of the Study:

• To validate a novel volumetric biventricular segmentation solution for multiaxis cardiac magnetic resonance (CMR) images.
• To compare the accuracy of automatic versus manual segmentation methods.

Main Methods:

• Segmentation of biventricular end-diastolic and -systolic phases from multiaxis cine CMR images.
• Utilized a nonrigid 3D surface model fitting approach for segmentation.
• Evaluated segmentation quality using volumetry, distance measures, and visual grading.

Main Results:

• No significant differences in volumetric measures between automatic and manual biventricular segmentation.
• Automatic segmentation showed comparable accuracy to manual segmentation for the left ventricle's endocardial border.
• Manual segmentation demonstrated superior performance for left-ventricular myocardial volume assessment.

Conclusions:

• Fully automatic segmentation provides reliable results for left ventricular volume and endocardial border assessment in CMR.
• Limited user interaction (2 minutes) is sufficient to achieve accurate segmentation for the right ventricle.