Automatic Segmentation of the Left Ventricle in Cardiac MR Cine Images for Volume and Myocardial Strain Quantification
We will present an automatic algorithm for the segmentation of the myocardium in cardiac cine MR images. The major components of the system will be described, namely the detection of the heart, the detection of the left ventricular blood pool, the inverse consistent deformable registration which is used to enforce temporal consistency in the segmentation, and the shortest path recovery of the contours. The system has been tested on thousands of patient and volunteer studies over the years and it is currently used in clinical routine for cardiac function analysis. This algorithm started as a research prototype and is now part of state-of-the-art products in the Siemens Healthineers portfolio. Finally, we will show the many works-in-progress extensions of this algorithm, in left ventricle modeling for electrophysiology, myocardial strain quantification, and cardiotoxicity.