Graph cut-based method for segmenting the left ventricle from MRI or echocardiographic images

• Published in: Computerized medical imaging and graphics Vol. 58; pp. 1 - 12
• Main Authors: Bernier, Michael, Jodoin, Pierre-Marc, Humbert, Olivier, Lalande, Alain
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.06.2017; Elsevier Science Ltd; Elsevier
• Subjects: Algorithms; Bioengineering; Computer Science; Echocardiography; Echocardiography, Three-Dimensional; Energy; Euclidean geometry; Euclidean space; Graph cut; Heart Ventricles - diagnostic imaging; Humans; Image contrast; Image processing; Image segmentation; Internal Medicine; Left ventricular 3-D segmentation; Life Sciences; Magnetic Resonance Imaging; Medical Imaging; MRI; NMR; Nuclear magnetic resonance; Nuclear medicine; Other; Post-production processing; Signal to noise ratio; Ultrasonic imaging; Ventricle; Warp; Echocardiography; Graph cut; MRI; Left ventricular 3-D segmentation; Automated Segmentation; Constrained Level-Set; 2d-Echocardiography; Magnetic-Resonance Images; Cardiac Mri; Energy Minimization; Whole Myocardium
• ISSN: 0895-6111; 1879-0771
• DOI: 10.1016/j.compmedimag.2017.03.004

Highlights • We proposed a method for segmenting the left ventricle from MR and echocardiographic images. • The method involves an implicit as well as an explicit shape prior. • Results are obtained in less than 5 s and are guaranteed no to have any holes and to have an anatomically plausible V-shape. • The method out performs every other methods on the CETUS echocardiographic dataset. • The method is as accurate as an expert on an MRI dataset containing images from 75 patients.