Counter Extraction of Ventricular Wall Based on Fuzzy-C-Means Algorithm for Ultrasound Image in Cardiac Torsion Motion

Image Segmentation plays a vital role in the quantitative analysis of medical ultrasound images, but the complexity of the ultrasound images makes it difficult to segment them accurately. A contour extraction algorithm based on Fuzzy-C-Means is proposed in this paper for the ultrasound images of cardiac ventricular wall with strong noises and fuzzy edges detected in the motion of heart torsion. This algorithm firstly constructs multiple features of pixels such as gray, gradient, location values to be clustering samples. By calculating the distances between the pixels and the ventricular center of gravity, the algorithm can make best use of the location features of the pixels within the ventricle effectively. And then the optimal threshold is chosen automatically to segment the image, which is worked out from FCM clustering center. Finally, the area extraction and contour tracking to the segmented ventricle are done by the contour tracking algorithm. Experiments results demonstrate that this method has a better performance on the image segmentation of ventricular wall. It can not only keep the edges more accurate, but also satisfy the request of coherent ventricular wall in the analysis of ultrasound heart images.