Hybrid 3D heart segmentation from dynamic CT images

In this paper, a fast hybrid 3D segmentation approach is proposed to identify entire heart region from dynamic CT images. Firstly, a morphological recursive erosion operation is introduced to reduce the connectivity between the heart and its neighborhood; then an improved fast marching method is employed to greatly accelerate the initial propagation of a surface front from the user defined seed structure to a surface close to the desired heart boundary; a morphological reconstruction method then operates on this surface to achieve an initial segmentation result; and finally morphological recursive dilation is employed to recover any structure lost in the first stage of the algorithm. The approach is tested on 5 dynamic cardiac datasets, totally 50 CT heart images, to demonstrate its robustness. The measurements revealed that the algorithm achieved a mean similarity index of 0.956. The execution time for this algorithm extracting the cardiac surface from a dynamic CT image, when run on a 2.0 GHz P4 based PC running Windows XP, was 36 seconds