Segmentation by deformable contours of MRI sequences of the left ventricle for quantitative analysis

The article presents an application of a deformable contour model to extract the epicardial and endocardial boundaries of the left ventricle (LV) from a time sequence of cardiac MR images. The deformable model used is based on the work of Rougon and Preteux (1991). In order to exploit both the geometric and photometric information present in temporal image sequences, an additional coupling force between the contours located in adjacent images is introduced. Automatic and robust initialization procedures for the iterative optimization of the model are investigated in order to reduce interactivity. The segmentation results are then used to calculate the evolution of the LV volume during the cardiac cycle. From these measurements, it is possible to compute precisely clinical parameters such as peak filling rate or ejection fraction. This method provides therefore a first step towards an automatic quantitative characterization of the heart function