Echocardiographic Image Sequence Segmentation and Analysis Using Self-Organizing Maps

Abstract

This paper presents a new approach for echocardiographic image sequence segmentation and a practical application of the results. The proposed method uses the self-organizing maps to approximate the probability density function of the image patterns. The map is post-processed, by the k-means clustering algorithm, in order to detect groups of neurons whose weights are similar. Each segmented image of the sequence is generated by correlating its pixels and clusters found in the map. The image sequence segmented was used to measure fetal heart structures. To refine the measurements we used a border detection technique based on the least-means squares error. The segmentation procedure was validated successfully by physicians.