Regresar

Improving ventricle detection in 3-D cardiac multislice computerized tomography images

Abstract:

This paper reports a segmentation approach that enables detection of left ventricle in three-dimensional (3-D) cardiac images. The proposed approach has been tested using 4-D (3-D+ time) cardiac Multi-Slice Computerized Tomography (MSCT) images. The generalized Hough transform and a seed based clustering procedure are integrated into the segmentation method. The method also considers an image enhancement step that consists in applying the mathematical morphology operators in order to improve the left ventricle cavity information in tomography images. A validation is performed by comparing the estimated contours with respect to contours manually traced by a cardiologists. From this validation stage the average contour error considering twenty three-dimensional images (a total of 2800 bi-dimensional images) is 6.23%. © 2011 Springer-Verlag.