Analysis of myocardial texture in resting echocardiographic images predicts recovery one year after myocardial infarction

The aim of this research is to develop a method for prediction of left ventricular recovery one year after myocardial infarction using texture parameters estimated for static ultrasound images. The study is performed for the monochrome and color (contrast based) echocardiograms that allow advanced evaluation of myocardial function. The analysis includes investigation of different texture feature selection methods and application of multilayer neural network, Support Vector Machine, and decision tree based Adaboost algorithm for recovery prediction. The obtained preliminary results are promising; estimated prediction error is lower than 30% for all used prediction tools.