Automatic Left Ventricular cavity detection using Fuzzy Isodata and connected-components labeling algorithms

Abstract

Extraction of the Left Ventricle (LV) contours is essential in the quantitative analysis of cardiac function. We present a new method that employs Fuzzy Isodat . a (FI) and conne . ctcd compo nents Labeling algorithms to automatically detect the LV contours. The FI algorithm is applied to each cardiac image to generate a given number of fuzzy subsets. Each heart's substructure ,<aorta . . . ) is considered as fuzzy cluster. To achieve the extraction process, a labeling algorithm is used. The method is tested by application of 32 sets of 30 images each. A comparison between the ejection fractions obtained by the proposed technique and a standard method is included.

The 3D Reconstruction of Pocillopora Colony Sheds Light on the Growth Pattern of This Reef-Building Coral

“…ISODATA, which has been widely used as a clustering algorithm (Mingchao et al, 2017), is used to divide the calices and inter-septal spaces into different polyp growth patterns in this work. As an unsupervised classification algorithm, ISODATA has the advantage of permitting an unknown number of clusters (Boudraa et al, 1992;Velasco et al, 2007) to be specified rather than requiring that value to be known a priori in the k-means algorithm method (Ahmad et al, 2013). The ISODATA algorithm uses the following process: (1) set initial parameters.…”

A Review on Medical Image Classification Algorithms

“…But in order to maximize the segmentation accuracy, we combine the SVM and ISODATA algorithm. And as an unsupervised method, ISODATA permits diverse number of clusters [23,24] to be specified rather than requiring that is known as a priori so that we can use it to correct the pixels which have been identified into incorrect parts. Then the improved SVM algorithm calculates the Euclidean distance between each pixel and the cluster centres which are determined by the hyperplane so it can correct the miss-identified pixels by putting them into the minimum Euclidean distance cluster using the adaptive method of unsupervised ISODATA.…”