Meeting Banner
Abstract #3280

A Geometric Method Based on Mass Center Drifting Detection for Improving Basal Left Ventricle Automated Segmentation

Mengchao Pei1, Lijia Wang1,2, Jianqi Li1, Mingxia Fan1, Yi Wang2,3

1Shanghai Key laboratory of Magnetic Resonance, East China Normal University, Shanghai, China, People's Republic of; 2Department of Radiology, Weill Medical College of Cornell University, New York, NY, United States; 3Department of Physiology, Biophysics, & Systems Biology, Weill Medical College of Cornell Universi, New York, NY, United States

To improve accuracy and robustness of automated left ventricle segmentation on short-axis cardiac MRI, a novel geometric method was proposed to define basal cardiac ventricle slices and effectively correct automated segmentation at basal ventricle slices compared to previous approaches. Cardiac short cine SSFP data from 30 patients were analyzed with IRB approval and HIPAA compliance. Among the 30 patients data, the success rates of identifying and correcting erroneous effusion at basal ventricle slices were 100.0% and 89.9% respectively. Automated quantification of corrected left basal ventricle slice volumes was well agreed with manual tracing measurement (0.241.09mL, R² = 0.9646). Ejection fraction calculations with this automated segmentation method were highly correlated with manual tracing measurement (0.32.7%, R² = 0.946). This approach is promising for fully automated accurate segmentation of the left ventricle blood.